Wednesday, June 5, 2019

Horizon trial: day 15 transcript

This is the transcript of Day 15 of the Horizon trial, the second trial in the Bates and others v Post Office group litigation, held at the High Court's Rolls Building on Tuesday 5 June 2019.

Jason Coyne, independent IT expert for the claimants spent the second of four scheduled days being cross-examined on his expert reports by the Post Office's QC, Anthony de Garr Robinson.

Mr Coyne's reports can be found here. The transcript follows:

                                       Wednesday, 5th June 2019
   (10.30 am)
                 MR JASON PETER COYNE (continued)
       Cross-examination by MR DE GARR ROBINSON (continued)
   MR JUSTICE FRASER:  Mr de Garr Robinson, just two things
       before we start.
           Judgment number 5, which I know you are not
       interested or involved in, went out this morning.  The
       embargo doesn't really apply because it is detailed
       reasons for decisions which were made public last week,
       but I would like a list of typographic errors by
       6 o'clock tomorrow.
           And in the interests of transparency, the learned
       usher just told me, just before I came in, that he had
       been given a message by the witness to give me, and
       I said I didn't want to hear the message and I'm not to
       take messages from witnesses in that way or indeed in
       any way, but I wanted both parties to know that that
       exchange had taken place.
   MR DE GARR ROBINSON:  My Lord, thank you for letting us
       know.
   MR JUSTICE FRASER:  All right.
   MR DE GARR ROBINSON:  Would your Lordship like to
       investigate that question?
   MR JUSTICE FRASER:  No, I don't intend to do anything at all
       but I wanted both of you to be told straightaway.
   MR DE GARR ROBINSON:  I understand.
           Mr Coyne, good morning.
   A.  Morning.
   Q.  Yesterday I'm sorry to say we gave you some homework.
   A.  Yes.
   Q.  We discussed your claim that bugs are often deferred or
       not dealt with at all on a cost benefit basis, do you
       remember making that claim?
   A.  I do.
   Q.  And I asked you whether you can think of any PEAK other
       than the particular PEAK we were looking at that shows
       that happening.
   A.  Yes.
   Q.  Have you been able to find the handful or so of PEAKs
       you referred to yesterday?
   A.  Yes, I have.
   Q.  I'm very grateful.  Do you have that on a piece of
       paper?
   A.  I do indeed.
   Q.  Perhaps the sensible thing to do would be at the break
       if you could give me the piece of paper and we can copy
       it and get it circulated to both sides.
   A.  Yes.
   Q.  Would that be acceptable?
   A.  Yes, it would.
   Q.  Thank you.
           Yesterday afternoon we were going through some
       documents you rely on in your reports with a view to
       seeing whether they justified the claims that you made
       about them.  These are some example documents.  I'm
       going to do a few more but I'm going to do them as
       quickly as I possibly can.  First of all can I ask you
       to look at paragraph 5.195 of your first report, and for
       the transcript it is at {D2/1/107}
   A.  Yes.
   Q.  You will see you say there:
           "The Post Office cash management proposals contained
       in a report dated 4 August 2017 suggests that they were
       actively considering ways to improve processes impacting
       on many of the issues raised above.  It is my opinion
       that, whilst the Post Office was looking at ways to
       improve cash management, it is also indicative that the
       system was generally far from perfect and there existed
       a real risk of bugs/errors/defects adversely impacting
       on branch accounts despite the processes in place at the
       time to prevent this."
   A.  Yes.
   Q.  Mr Coyne, if we could go very quickly to that document.
       It is at {F/1673/1}.  It would take too long for me to
       read it out loud.  Perhaps I could ask you, Mr Coyne, to
       read the first page quickly to yourself.
           (Pause)
   A.  Yes, I have read that.
   Q.  Then over the page {F/1673/2} we can see "What we
       propose to do and why", about a third of the way down
       the page:
           "We proposed to deliver the following initiatives
       through this business case."
           First of all there is a reduction of branch cash
       holdings by circa £80 million.  Perhaps I can ask you to
       read that paragraph 1(a).  (Pause)
   A.  Yes.
   Q.  Then 1(b) is "Improving branch cash declarations" where
       they say:
           "In conjunction with the above activity, a more
       strategic solution will be delivered to reduce surplus
       cash in the branch network by £80m (£60m in Sterling and
       £20m in foreign currencies) through mandatory and
       accurate cash declarations in branch."
   A.  Yes.
   Q.  That's talking about trying to get postmasters, when
       they make their cash declarations which they have to do
       everyday, to make them more accurate, to make sure they
       get them right.
   A.  Yes.
   Q.  That's explained in the following paragraphs (i), (ii),
       "Proactively manage non-conformance" and "Changing time
       of cash declaration submission", do you see that?
   A.  Yes.
   Q.  Then over the page {F/1673/3}, "Improving operational
       design, training and communications to Postmasters to
       ensure cash declaration conformance."  It is all about
       getting the SPMs to do what they already do but to get
       them to do it better.
   A.  Yes.
   Q.  Then if we go over to page {F/1673/9} of the document
       there is what is called a "Benefits Map".  It is a table
       with a series of solutions on the left-hand side,
       a series of impacts and a series of benefits in
       different columns along the page.
   A.  Yes.
   Q.  Picking it up so we can see the sort of thing it is
       dealing with, at page {F/1673/10}, the second box down:
           Discrepancy management – at the moment there is a
       lack of visibility of any inaccurate cash declaration to
       the Postmaster of one of his stock units.  If we deliver
       a technical change to the cash declaration process this
       will send any discrepancy amount to the Local Suspense
       account which will give the Postmaster immediate
       visibility and will allow faster
       corrections/investigations ..."
   A.  Yes.
   Q.  What they are suggesting is that the information that
       a postmaster gets when doing a cash declaration, some
       further information should be added which is immediately
       added to his or her suspense account which then impels
       him or her to look into the matter more closely, do you
       see that?
   A.  Yes.
   Q.  That is essentially what I get from this document.  It
       might be my fault.  What I would like to ask you is,
       going back to your statement at paragraph 5.195, why is
       it you say that this document is indicative that:
           "... there existed a real risk of bugs ... adversely
       impacting on branch accounts despite the processes in
       place at the time to prevent this."
           {D2/1/107}
   A.  Could I have the document back up, please?
   Q.  Of course.  It is {F/1673/10}.
   A.  Sorry, could we go to the first page {F/1673/1} of that
       document.  Then go on to the next, please {F/1673/2}.
       Onto the next, please {F/1673/3}.  And onto the next,
       please {F/1673/4}.  And onto the next, please
       {F/1673/5}.  And onto the next, please {F/1673/6}.  Next
       one, please {F/1673/7}.  And the next one, please
       {F/1673/8}.  Next one, please {F/1673/9} and again,
       please {F/1673/10} and the next one, please {F/1673/11}.
           And the next one please {F/1673/12}.  And the next
       one please {F/1673/13}.  Sorry next one, please
       {F/1673/14}.  And again, please {F/1673/15}.  Next one,
       please {F/163/16} and again, please {F/1673/17} and
       again, please {F/1673/18}.
           That's the end of that document, is it?
   Q.  Yes.  So can you tell me what it is you saw in this
       document which allowed you to express the opinion that
       it indicates that there existed a real risk of bugs
       adversely impacting on branch accounts?
   A.  It is incorrect to find that from that document.
   Q.  Mr Coyne, you have already accepted, and very fairly and
       properly accepted, that as an expert it is important
       when relying on documents, particularly in
       a document-heavy case where many, many documents are
       relied on in a report, you have already accepted the
       great importance of making sure any summary of the
       document, any explanation as to what the document means
       or what it indicates, it is very important to get that
       right to assist the court.
           You do seem to be looking for problems in documents
       which don't support the suggestion that those problems
       exist.  Would you accept that?
   A.  No, I don't.  It might be the case that we have
       an incorrect reference, there was an incorrect reference
       yesterday, but I don't think so in this case because the
       context of the paragraph appears to relate to this
       document.
   Q.  You are simply citing documents that don't support the
       claims you make about them, aren't you, in this report?
   A.  This particular example, there appears to be a mistake
       here, yes.
   Q.  Let's move down the page to paragraph 5.198 where you
       say -- this is in your report at {D2/1/107}:
           "It is clear that in some instances it is not always
       apparent whether recurring discrepancies were as a
       result of system bugs or the Subpostmaster's own
       actions, or other things beyond the control of the
       subpostmaster."
           Then you have two footnotes {D2/1/108}.  Then:
           "However the fact that the SSC support team were
       unable to assist or identify the root cause does
       undermine the credibility of Horizon itself."
           Correct me if I'm wrong, but I think what you are
       suggesting there is that the two documents you refer to
       in the footnote as examples of it not being apparent
       whether recurring discrepancies were a result of bugs or
       human error, you are saying those documents itself
       support the idea that the support team were unable to
       identify root causes and in a way that undermines the
       credibility of Horizon itself?
   A.  There is a number of documents, and I agreed with
       Dr Worden that often the bugs, errors or defects would
       appear as if they were mistakes made by the
       subpostmasters.
   Q.  You say "often".  We need to be careful with "often".
   A.  There are a number of occurrences.
   Q.  Let's be clear about scale, shall we?  I think you have
       agreed with Dr Worden that over the lifetime of Horizon
       there were something like 3 million branch accounts
       generated, yes?
   A.  Yes.
   Q.  And I'm taking a branch account as a monthly branch
       account.  I'm simplifying because of course between 1999
       and 2005 they were weekly accounts, weren't they, but
       let's just treat them as monthly accounts.  So there
       were 3 million branch accounts that were produced during
       the course of this period and if an error is made, one
       error is made, that means it has a branch account
       effect, that means that there is a 1 in 3 million chance
       of that error affecting a branch account, yes?
   A.  It is unlikely that one error within the system would
       only affect one single branch account.
   Q.  We will talk about that.  It depends on the nature of
       the error.
   A.  It does indeed.
   Q.  You can't say that, can you, before you know the nature
       of the error?
   A.  No, you can't say that.  That's why you need to be
       careful.
   Q.  So when you say this often happened, it gives
       an impression, doesn't it?  It gives an impression that
       suddenly large numbers -- a significant portion of
       branch accounts may be unreliable because of something
       happening.  But if you are talking about a handful of
       cases or if you are talking about 20 cases, you are
       still only talking about 20 in 3 million.  You are
       talking about 2 in 300,000.  You are talking about a 1
       in 50,000 chance, aren't you?
   A.  I wasn't talking about branch accounts, I was talking
       about bugs, errors and defects.  Bugs, errors and
       defects often appear as if it was a user error rather
       than a defect of Horizon and it was that that Dr Worden
       and I agreed on.
   Q.  What we are talking about is what undermines the
       credibility of Horizon itself.
   A.  Yes.
   Q.  What I'm suggesting to you is that if there were a dozen
       examples of something happening, given the number of
       branch accounts that are in existence, and assuming each
       of those examples only had one impact, that doesn't
       undermine the credibility of Horizon itself, does it?
   A.  Well, it needs to be considered because it is unlikely
       that it would only have an impact on one branch account.
       Sorry, let me finish, please.  If it is a defect in the
       system, the majority of the users, subpostmaster users,
       were using that system.  So it would be unlikely that it
       only impacted one.
   Q.  I'm very interested in your answer, Mr Coyne, because in
       my question I was quite careful to indicate that I was
       asking about one error that had one impact, but you
       immediately flicked to a situation where you were able
       to say, well, there are likely to have been more impacts
       in circumstances where I wasn't even specifying what the
       error was other than that it didn't have multiple
       impacts, and could I suggest to you that you did that
       because you have a view, you have a world view, you have
       a desire to maximise the impression given of any error
       that you identify, do you think that's fair?
   A.  No, I don't think that's fair.  I understood the
       question you were putting to me was about this system
       rather than a hypothetical scenario where one bug only
       impacts one account.
   Q.  Let's talk about a remote access instance.  The SSC on
       one occasion does some remote access which affects one
       branch.  Do you accept that would only have one branch
       impact?
   A.  Yes.
   Q.  So if you find one example of remote access which has
       one branch impact and you don't know which branch is
       affected, and you look at the totality of the branches
       in the network and the totality of the monthly accounts
       that have been generated in the network over 20 years,
       you would have -- if you picked a branch account at
       random, you would have a 1 in 3 million chance of
       finding a branch that was affected by that remote
       access?
   A.  If that remote access was done correctly it would only
       impact one branch, yes.
   Q.  Thank you.  Going back to this example, and if I'm wrong
       please tell me because it will save some time.  There
       are two footnotes, do you see, and they are call logs.
       Do you see that?  And they are footnotes which are given
       as instances where it is not always apparent whether
       recurring discrepancies were as a result of system bugs
       or the SPM's own action, do you see?
   A.  Mm.
   Q.  You then go on to say the fact that the SSC support team
       were unable to assist or identify the root cause
       undermines the credibility of Horizon.  Are you
       suggesting that if I look at those logs I will see
       something to justify the inference that you make here --
       that you appear to make here -- that the credibility of
       Horizon itself is undermined? {D2/1/108}
   A.  The way that the reference is introduced here is "or
       other things beyond the control of the Subpostmaster".
       {D2/1/107}
   Q.  I'm sorry?  I'm not following you and it is my fault,
       not yours.
   A.  Sorry, I have just flicked back to read what was before.
   Q.  Yes.
   A.  So these are discrepancies that were:
           "... as a result of systems bugs or the
       Subpostmaster's own actions, or ... things beyond the
       control of the Subpostmaster."
           And they should be referenced in these --
   Q.  Right.  So these logs indicate that things are happening
       which are either the result of system bugs or a result
       of SPM action or are beyond the control of the SPM, yes?
   A.  Yes.
   Q.  And do you infer from those examples -- is what you then
       say in the last sentence an inference from that?  Do you
       say that those examples show -- well, those examples
       undermine the credibility of Horizon itself?  I think
       you would say yes?
   A.  We are in the section of my report here which is
       "Opinion Summary", so it is summarising the section
       that's come before it.
   Q.  Yes, but you give these two examples here and I would
       have thought you did that for a reason.
   A.  It is likely that these are two documents that relate to
       this section, but if there was a footnote that I have
       already referred to in the section I wouldn't reference
       it again here.
   Q.  Yes.  So you are suggesting that if we look at the
       footnote we will see something that undermines the
       credibility of Horizon itself, is that right?
   A.  Yes.
   Q.  Let's have a look then.  Could we go to {F/333/1}
       please.  This is a call log.  I think it relates to
       Mrs Misra's branch and the date is -- it was opened on
       23rd February 2006.  So far as I can tell, the relevant
       passage that you would rely on is at the bottom of the
       page where it says:
           "OTI close Monday 27 February."
           Do you see that?
   MR JUSTICE FRASER:  I think we are going to need to increase
       the size.
   MR DE GARR ROBINSON:  Yes, it is very small.  Could we go to
       the bottom and increase the size, please?
   MR JUSTICE FRASER:  I think for present purposes that's
       probably magnified enough at least so I can read it.
   MR DE GARR ROBINSON:  Let me read it:
           "No transaction date and time was provided for this
       action using current date and time.  Update by Anne
       Chambers: Category 94 -- Final -- Advice and guidance
       given."
           Stopping there.  We have seen a lot of
       Anne Chambers' work, haven't we?
   A.  Yes, we have.
   Q.  My impression is that she is quite a professional
       operator.  Would that be your impression as well or
       would you disagree with that?
   A.  I don't think I could give a view on that.
   Q.  Very good.
           "I have checked very carefully and can see no
       indication that the continuing discrepancies are due to
       a system problem.  I have not been able to pin down
       discrepancies to individual days or stock units because
       the branch does not seem to be operating in a
       particularly organised manner.  In particular I have
       noted 1.  There are 6 stock units for this 3 counter
       branch, which seems a bit excessive.  2. The loss in
       euros in TP 9 appears genuine - the declared quantity
       was 4000 fewer than the system expected. It is not clear
       from the information above whether anyone found out why
       this happened (there were several rem outs, and a rem
       in, on 23rd Dec - did the pouches contain the declared
       number of euros?).  3. Stock is sometimes transferred
       out of a stock unit where it is not held.  In particular
       there were several transfers out of stock unit SMI in TP
       10. At the end of the period the stock figures were
       corrected back up to zero via Adjust Stock. This gave a
       gain of over £2000 in SMI.  Equivalent negative stock
       adjusts in AA gave a corresponding loss in AA.  4. I am
       not confident that the stock declarations are always
       correct e.g. at the end of TP 9 there was a declared
       holding of 5 £20 PO phonecards in the branch, then a few
       days later 20 were transferred from one SU to another.
       None were remmed in until a week after that.  5. The
       branch had declared 27 £20 Argos vouchers at the end of
       TP 9. Branches have now been instructed to rem out this
       product; they remmed out 17 and adjusted stock to
       account for the remaining 10 (so did they really only
       have 17 to start with?).  This has correctly caused a
       loss of £200 in SU AA.  6. Lottery instants sales are
       entered onto the system as a single transaction every 10
       days or so.  7. Stock units SMI and AA rolled over with
       non-zero cheque holding.  This may be to do with how the
       discrepancies have been accounted for but I do not
       really understand this (the total is greater than the
       sum of the branch adjustments for TP 9 and 10).  I
       recommend that this call is passed back to NBSC tier 2
       for further investigation, since there is no evidence
       that the discrepancies are being caused by a system
       problem.  If you want the above information in an email,
       let me know."
           Now, Mr Coyne, what I would like to suggest to you
       is that what that shows is that Anne Chambers did a very
       thorough job, went through the figures very carefully,
       saw there was a branch that was operating in some sort
       of chaos, forms the clear view that there's no system
       problem, but says: there are these questions, it should
       go back to NBSC to investigate.  And I would like to ask
       you why you think that that story there, told in that
       box, does undermine the credibility of Horizon itself?
   A.  I think it is that when this call is later advanced, it
       is discovered that there was a system problem.
   Q.  Perhaps you would go to the second page because if
       that's the case I have missed it but I'm happy to be
       corrected.  {F/333/2}.  What I get is what's in the
       final entry, Mr Coyne:
           "Call close by David Dawe: pm was getting
       discrepancy's ssc have investigated and adviced that the
       NBSC take a 2nd look at this as the office stock units
       appear to be in a mess."
           But please don't let me stop you reading the whole
       thing.
           (Pause)
   A.  But what we do see from that is that Anne Chambers isn't
       able to say what has actually happened with the
       discrepancy that has been seen.  She is unable to
       determine whether it is the user that has caused that or
       whether there is a potential problem with the system.
   Q.  So let me get this straight.  Let me give you
       a hypothetical case.  A branch is being run in a mess
       and things are being reported that are wrong.  So things
       are being remmed out from a particular stock unit that
       aren't in the stock unit, they are declaring cash that
       they don't have, they are declaring they have stock that
       no longer exists, there are inconsistent declarations of
       stock on different days.  It is a mess.  Clearly lots of
       erroneous figures are being entered into the system.
       There is no way that someone in the SSC is going to be
       able to correct those errors, only the subpostmaster
       will know what the true position is on the ground.
       Correct?
   A.  Yes.
   Q.  So in that situation, the SSC comes in and looks to see
       if there's a system problem and they can't find one.
       Now are you saying -- that scenario is inevitable, isn't
       it?  It doesn't matter how good the system would be; you
       could have the computer from the Star Trek Enterprise.
       The point is that in that scenario the SSC would not be
       able to say that this has happened or that has happened,
       because the data they have got is too chaotic, correct?
   A.  Yes.
   Q.  So I would like to suggest to you, Mr Coyne, that the
       sense one gets from these logs is that's what was
       happening with this branch.
   A.  Yes.
   Q.  So why do you say that this log undermines the
       credibility of Horizon itself?
   A.  No, I agree that this log in itself doesn't.
   Q.  I don't want to take any time up, but are you suggesting
       that the second log, if I go to the second log, I'm
       perfectly happy to do it, would show a different
       picture?
   A.  I would have to check it.  I believe it shouldn't but
       I'm happy to go to it.
   Q.  I do not think I have time.  Lets move on.
           What I would like to do now is to talk about
       a document on which you have built one of the major
       themes in both your reports, which is the reliability of
       Credence, and as we know that is one of the management
       information systems used by Post Office.  It is the
       Post Office system, isn't it?  And it uses it for
       various purposes, including it is one of the systems it
       uses when deciding whether to issue a transaction
       correction, yes?
   A.  Yes.
   Q.  One of the themes in your reports is that it shouldn't
       be used for the purpose of making those decisions, ARQ
       data actually should be used, is that right?
   A.  Yes.
   Q.  That is your considered view?
   A.  Yes.
   Q.  And on that basis you rely on a document which is
       called -- which has become called the Helen Rose report,
       I'm not sure it is a report, but it is a five-page
       document produced by Helen Rose who was a fraud analyst
       at the Post Office in June 2013.
           Before we go to it, can we just agree the basic
       facts.  I presume you looked at the report and the
       associated facts quite carefully so you are familiar
       with the case?
   A.  Yes.
   Q.  It related to an incident at the Lepton branch of which
       the SPM was a Mr Armstrong, is that right?
   A.  Mm.
   Q.  A bill payment transaction had failed at his branch when
       the system went down, correct?
   A.  Yes.
   Q.  And it was a cancellable transaction, correct?
   A.  Yes.
   Q.  So he completed the transaction, he took money from the
       customer, and he did that via -- the customer had
       actually got cash out from a Lloyds TSB cash withdrawal
       but that's by the by because that didn't fail.  So he
       completed the transaction and he took money from the
       customer.  Because the system went down he had to log
       back in, and when he logged back in the recovery process
       automatically reversed the cancellable transaction,
       correct?
   A.  Yes.
   Q.  And that's how the system should operate with
       cancellable transactions, correct?
   A.  Yes.
   Q.  And that left him with a surplus in his branch, didn't
       it?
   A.  Likely, yes.
   Q.  Because he had been given cash by the customer which was
       to be used to pay -- I think it was a phonecard or
       something like that.  It was a BT bill payment, sorry.
       But of course because the transaction had failed the BT
       bill payment was not made, yes?
   A.  Well, it depends at what point the counter failed and
       that's what the recovery process does, it determines how
       far the transaction got.
   Q.  Yes.  When you agreed with me a moment ago that it was
       a cancellable transaction, what that means is -- there
       are two kinds of transactions, aren't there?  There is a
       recoverable transaction and the opposite of
       a recoverable transaction is a cancellable transaction,
       correct?
   A.  It is often called nonrecoverable.
   Q.  But the technical name is cancellable?
   A.  Yes.
   Q.  Could you explain what the difference is between those
       two transactions, why a transaction is cancellable or
       recoverable?
   A.  It depends on whether it requires an interaction with
       any of the banking organisations or not.  Often with
       things like credit cards or debit card transactions
       a call will be placed to the bank to check the money and
       then if the process continues all the way to the end the
       money being requested.  If there is a failure in the
       counter at some point through that process then Horizon
       has got to understand whereabouts it failed and then
       effectively unwind that process.
   Q.  The difference is where a third party system is
       involved, isn't it?
   A.  A third party system, yes.
   Q.  So where you are doing a transaction at the counter
       there are a number of steps you take and if the
       transaction involves, I don't know, a payment being made
       from a bank, during the course of typing in the
       transaction hundreds of messages are passing back and
       forth both to the BRDB to record the nature of the
       transaction that's being keyed in and also to the
       financial institution, and the two institutions marry up
       and the financial institution says I recognise you, and
       the counter says I want you to make this payment, and
       the financial institution says I accept it, and so on.
   A.  Yes.
   Q.  Then at the end of that process the postmaster closes
       the stack.
   A.  Yes.
   Q.  He enters the transaction into the system.  I think the
       technical term is he commits the basket to the system?
   A.  Yes.
   Q.  And that's the moment at which the basket enters his
       branch accounts, is that correct?
   A.  Yes.
   Q.  Everything I have said so far is correct, is it?
   A.  It is, yes.
   Q.  Thank you.  The problem is that inevitably the moment at
       which the transaction is committed to -- the basket is
       committed to the system is different in time from the
       moment at which the payment instruction is accepted by
       the bank, yes?
   A.  Yes.
   Q.  So what happens, in a relatively rare situation you can
       have the bank accepting an instruction to make a payment
       and making the payment, then the system breaks down, and
       then that means that the transaction is not entered into
       the branch account?
   A.  That is correct, yes.
   Q.  So you have a discrepancy between what has happened in
       the real world, which is that a payment has been made by
       the bank, yes?
   A.  Yes.
   Q.  And you have the fact that that payment is not recorded
       in the branch accounts because the system has collapsed
       before the basket has been closed -- I should say
       committed, yes?
   A.  Yes.
   Q.  Do you mean collapsed?  You said the system collapsed.
   MR DE GARR ROBINSON:  It is a very loose form of --
       your Lordship understands what I mean.
           But the system goes down.  It could be a comms
       problem, it could be a systems problem, it could be
       someone has dug up the phone line outside.
   A.  A power problem.
   Q.  It could be a thousand kinds of problem, yes?
   A.  It could be lots.  I'm not sure thousands --
   Q.  I'm sorry, it is fair that you should make that
       clarification.  I am not trying to commit you to that
       number.
           So in that situation any system, let's forget about
       Horizon, again we have got a Star Trek brilliant system,
       any system is going to have to manage that problem,
       isn't it, in some way?
   A.  Yes.
   Q.  Because there are always going to be situations where
       what's happened in the real world may not actually
       accord with what's recorded in the accounts?
   A.  That is right.  Frankly what Horizon does, rather than
       make the assumption that the transaction completed
       successfully, it effectively re-looks at the elements of
       the transaction to see how far it got, to see whether it
       should roll back or roll forward.
   Q.  Yes, because in the course of the transaction being
       keyed in, before the basket is committed, all the
       elements of the transaction are actually recorded --
       let's talk Horizon Online -- they're all recorded in the
       BRDB but they are in different tables of the BRDB.  So
       they are securely kept, held somewhere, for the moment
       in time in which the transaction is committed to the
       audit store -- I shouldn't say the audit store -- to the
       database, so they are held there but they are held in
       abeyance.  Then when something goes wrong the
       transaction isn't committed to the database and the
       tables which contain the data relating to the
       transaction, and some other tables, they then throw up
       a flag saying this is a recoverable transaction.  And
       what that means is that the transaction appears -- it
       has been done in the real world but it hasn't entered
       the stacks, it hasn't entered the branch accounts, so it
       has to be looked at to see what needs to be done, is
       that correct?
   A.  Yes.
   Q.  And that's how the Horizon system was designed to work,
       correct?
   A.  Yes.  Just in pure technical terms, the raising of the
       flag, once the transaction is started that's recoverable
       a flag or a stake is put in the database to say we are
       starting a recoverable transaction.  Then at the end of
       it, once it is committed, the flag is taken down.
   Q.  Exactly.
   A.  So when the counter restarts it has a look to see
       whether there is any recoverable transaction flags
       there.  If there are, it has to deal with that before it
       boots up.
   Q.  Exactly.  And this isn't strictly relevant to Helen Rose
       but just to be clear, in that scenario, given the way
       the system is designed, indeed given the way any system
       would have to be designed, there would then have to be
       an enquiry as to what happened on the ground, wouldn't
       there?
   A.  By the humans interacting?
   Q.  Yes.  In other words, let's take this example, the
       Post Office would have to find out from the postmaster
       whether he accepted the £76, wouldn't he?  They would
       need to know whether the money was accepted or whether
       it wasn't, and only then would they know what they
       should do in relation to this transaction?
   A.  Typically a counter would know whether the transaction
       or whether the monies have been handed over because one
       of the last things that you would do at the end of the
       transaction would be -- it is called firing, you would
       fire the cash drawer and the cash drawer would come
       open.
   Q.  You are not suggesting it wouldn't have to be checked.
       You wouldn't assume that the money had passed hands, you
       would need to know whether it had.  You would need to
       ask the postmaster, wouldn't you, in order to work out
       what, if anything, needed to be done to restore the
       branch to balance?
   A.  But it is a worthwhile check to do, to find out whether
       cash has been handed over or not.
   Q.  Yes, because the system on its own doesn't know whether
       cash has passed hands, does it?  The system doesn't tell
       you.  It doesn't photograph the passing of cash from one
       to the other, there is no way in which the system would
       ever know that?
   A.  No.  It would know whether it has displayed a message on
       screen to say pay X amount and it would know whether the
       cash drawer has been opened or not, but it wouldn't know
       if physically that instruction had been followed, yes.
   Q.  So it might be, for example, the message is flashed up
       on the screen and then the system crashes, and if you
       were at the Post Office or the SSC you would not be able
       to tell, looking at the data you have, what had happened
       and you would have to make an enquiry.  And it would be
       a good practice, generally speaking, to make that
       inquiry before deciding whether any correction needs to
       be made or not, yes?
   A.  Yes.
   Q.  Thank you.  So let's go back to Lepton.  This was
       a cancellable transaction because there hadn't been
       an immediate instruction for a payment to be made by
       a financial institution.  I think you will accept with
       me that when that happens it is not in the recoverable
       category, it is in the cancellable category, yes?
   A.  I would have to check that by looking at the report.
       I can't recall precisely what the --
   Q.  Okay.  You agreed with me earlier that it was
       a cancellable transaction?
   A.  I believe so, yes.
   Q.  Which means that the standard process with -- in fact
       the universal process with cancellable transactions is
       that the transaction is then removed from the system.
       The assumption is made that the transaction should not
       be done.  Then if there is any problem that can be
       handled by manual processes.  Again you can ask the
       branch whether in actual fact, although we have
       cancelled the transaction, have you actually received
       some money?  That's how the system works, correct?
   A.  Mm.
   Q.  And that is how the system worked in this case, didn't
       it?
   A.  In this case there was a dispute between whether the
       system itself said there was the reversal or whether the
       human, the subpostmaster, chose to do the reversal or
       not.  And the indicator within Horizon was that it was
       the subpostmaster that did the reversal but it was found
       that that was incorrect and it was actually Horizon.
   Q.  So you are saying Credence said it was the postmaster
       that did it but in actual fact it was the system that
       did it.  And that's your considered view?
   A.  I believe that's what the document reflects, yes.
   Q.  Let's pick this up in your second report -- one other
       thing I should mention, actually, is that in this
       process, the way the system is supposed to operate, when
       there is a cancellable transaction like that, or indeed
       even a recoverable transaction, receipts should be
       printed by the system to allow the postmaster to know
       what's happened and what he or she should be doing, yes?
   A.  Yes.  The process should be that receipts are printed.
       There are other reports elsewhere that suggest that that
       is not always the case, but that is certainly what the
       process should be.
   Q.  Let's look.  Can we go to page 117 of your second report
       which is {D2/4.1/1}.  This is paragraph 4.78.  It is all
       under the heading "Failed Reversals" {D2/4.1/117}.
           You say at 4.78:
           "As dealt with above at paragraph 4.62, the excerpt
       from Gareth Jenkins within the Helen Rose report
       indicates that there was no evidence of the creation of
       a disconnected session receipt, unless further diagnosis
       (which I do not believe has been disclosed to me) has
       since been conducted and reviewed by Angela Van Den
       Bogerd.  I have reported on what was diagnosed
       contemporaneously by Mr Jenkins, particularly ..."
           Then you quote a piece of text that I won't read but
       I invite you to read.
           (Pause)
   A.  Yes.
   Q.  So what you are suggesting there is that the Helen Rose
       report indicates that Horizon didn't produce a
       disconnected session receipt in branch, yes?
   A.  Yes.
   Q.  If we go back to page 113, {D2/4.1/113} and look at
       paragraph 4.63, you are talking about Credence now, you
       are talking about the Helen Rose report.  You say:
           "Therefore, the contemporaneous evidence is
       consistent with the determination that Horizon initiated
       the reversal, NOT the Subpostmaster."
   A.  Yes.
   Q.  "In my first report I had explained (at paragraph 4.61)
       that the Subpostmaster had not reversed the transaction,
       this had been a reversal generated by the system as part
       of recovery."
   A.  Yes.
   Q.  "Credence data appeared to show (or was interpreted as)
       being a reversal initiated by the Subpostmaster.  This
       difference of position arose from Post Office looking at
       Credence data and Gareth Jenkins of Fujitsu looking at
       audit data and system logs."
   A.  Yes.
   Q.  "This demonstrates two positions", you say:
           "(a) Credence data, most commonly used by Post
       Office for their investigations, is either wrong or does
       not provide sufficient information to complete the full
       picture; and
           "(b) It was only after the Subpostmaster involved an
       external forensic accountant that the Audit data was
       requested."
           The external forensic accountant, are you aware of
       this, what that's a reference to, Second Sight?
   A.  In the documents that I have seen, the call logs,
       I think the subpostmaster says "I have got a forensic
       accountant involved", I do not think he mentions --
   Q.  You are not aware it was Mr Warmington from
       Second Sight?
   A.  I wasn't aware.
   Q.  Fine, I will not ask you any more about that.
           Then if we go back to what you said about this in
       your first report.  Can we go to {D2/1/67} please.  Are
       you there?  Paragraphs 5.49 to 5.50:
           "The document ('Helen Rose report') refers to an
       incident where a Transaction Correction was issued which
       the Subpostmaster duly settled financially despite the
       Subpostmaster denying conducting the reversal."
           5.50:
           "The report appears to show that the material that
       Post Office initially reviewed did not identify that it
       was the system that initiated the reversal rather than
       the Subpostmaster and therefore the Transaction
       Correction making the Subpostmaster liable was issued in
       error.  Since this is effectively a failure to
       appropriately reduce the risk of error this is also
       dealt with further ..."
           So here you are saying -- well, let's move on
       actually to page {D2/1/101}.  You have some more points
       to make about Credence at page 101.  Picking it up at
       paragraph 5.175 --
   A.  Sorry, are we in the second report now?
   Q.  The first report {D2/1/101}.  Picking it up at 5.175,
       you say:
           "The report regarding the reversal dispute conducted
       by Helen Rose states:
           "On looking at the Credence data, it clearly
       indicates that the reversal was completed by ...
       (Subpostmaster) at 10:37 ... and was reversal indicator
       1 (existing reversal) and settled to cash."
           {D2/1/102}
           "5.176:
           "It is therefore relevant to question why Post
       Office were using Credence data to initially investigate
       disputed transactions."
           Stopping there.  Your contention is that they should
       not use Credence to initially investigate, is that
       right?
   A.  It would seem that you can use Credence to conduct
       a cursory investigation but you have to go back to the
       full logs to get the full picture.  Because if there's
       a different picture being given by Credence to that of
       the logs, then ultimately both can't be correct.
   Q.  It is just your use of the word "initially".  Is there
       any significance attached to that?  That's not what you
       should look at even first, you should look at something
       else first, should you?
   A.  No, I mean, it depends what depth of investigation you
       are going to look.  If it is just a cursory
       investigation then Credence might be okay for that.
   Q.  I see, thank you.
           5.176:
           "Whilst it is evident that it was understood by Post
       Office in this instance to request assistance from
       Fujitsu for further material to investigate this dispute
       there appears to be further issues with the data
       provided by Fujitsu."
           5.177:
           "Observations of the disclosures illustrates that
       the initial report ..."
           That is the Helen Rose report, right?
   A.  Mm.
   Q.  " ... states 'a transaction at 10.42', whereas the
       Credence data file shows 10.32 with the reversal at
       10.37."
           Stopping there.  You are giving another example of
       Credence giving wrong data, yes?
   A.  There appears to be a difference between the times that
       are recorded, yes.
   Q.  "Fujitsu's data states the transactions are at 9.32 and
       9.33 and reversal timestamp is 9.37."
           You are suggesting that is a further problem with
       Credence, that it is actually an hour out as compared
       with audit data, yes?
   A.  I actually say that in the next paragraph, yes.
   Q.  Then you say:
           "5.178:
           "Whilst this hour difference between the data sets
       might be easily traceable for Fujitsu, it is not clear
       how easily it would have been to investigate issues
       where the Subpostmaster was not sure of what time things
       went on erroneously in the system ..."
   A.  Yes.
   Q.  So what you are doing here, Mr Coyne, is that you are
       making the following claims: first of all,
       a disconnected session receipt wasn't printed when it
       should have been, correct?
   A.  That is what the report says, yes.
   Q.  Secondly, that Credence data initially relied on by
       Post Office was misleading, misleading as to who
       reversed and misleading as to time, yes?
   A.  Yes.
   Q.  And, thirdly, the problems with Credence led to
       an erroneous transaction correction, inflicting a false
       loss on the subpostmaster, correct?
   A.  Whether there was an erroneous transaction correction or
       not is not clear, it depends what decision was taken
       based on the evidence, based on either the Credence or
       the ARQ log.
   Q.  We can go back to your first report, paragraph 5.50, but
       my understanding of what you said there was the
       Post Office wasn't liable and it was a false transaction
       correction.  So that is your view, isn't it, that you
       formed on the basis of reviewing the Helen Rose report
       and other documents?
   A.  If Post Office had have continued to use the Credence
       data, then the transaction correction would have been
       issued in error.
   Q.  You actually say, reading again from paragraph 5.50:
           "... and therefore the transaction correction making
       the Subpostmaster liable was issued in error." {D2/1/67}
   A.  Yes.
   Q.  You are making a claim as to what happened on the basis
       of the documents you have seen?
   A.  Yes.
   Q.  And is that your view?
   A.  Yes.
   Q.  Thank you.  Let's now go to Ms Rose's report.  It is at
       {F/1082/2}, pick it up at page 2.  There's the
       "Executive Summary" and the first paragraph has the time
       10.42 that you referred to.  You see that?
   A.  Yes.
   Q.  The report says:
           "The branch was issued with a Transaction Correction
       for £76.09, which they duly settled; however the
       postmaster denial reversing this transaction ..."
           Under "Reviewing the Data", let's read that:
           "On looking at the Credence data, it clearly
       indicates that the reversal was completed by JAR001
       (postmaster) at 10:37 04/10/2012 and was reversal
       indicator 1 (existing reversal) and settled to cash.  An
       existing reversal is where the session number/Automated
       Payment number has to be entered to reverse the item.
           "The Fujitsu logs were requested for this branch,
       but whilst waiting for these to arrive communications
       took place with Gareth Jenkins at Fujitsu for more
       details to gain an understanding what had occurred at
       this branch."
   A.  Yes.
   Q.  Now, she says the Credence data clearly indicates that
       the reversal was completed by the subpostmaster.  But it
       is fair to say, isn't it, that the Credence data did not
       actually say that the subpostmaster had initiated the
       reversal, correct?
   A.  Well, certainly whoever constructed this report said it
       clearly indicates that the reversal was completed by the
       user.
   Q.  She's inferring from the facts that she sets out there
       that the reversal itself must have been initiated by the
       subpostmaster, isn't she?
   A.  That's what she is saying.  She's saying "it clearly
       indicates".
   Q.  It is an interpretation of the data she has got.  There
       isn't a box in Credence -- she's not saying there is
       a box in Credence saying this was initiated by the
       subpostmaster, it is that the reversal has a postmaster
       reference attached to it and a reversal indicator 1, and
       she infers from that, she construes that, she interprets
       that as indicating that the reversal was specifically
       undertaken by the subpostmaster.  Would you accept that
       what we are talking about here is a mistake in
       interpretation?
   A.  There's nothing here to suggest there is a mistake in
       interpretation to me.  The words on the page say "it
       clearly indicates that the reversal was completed" by
       the subpostmaster.
   Q.  I would like to suggest to you, Mr Coyne, that what this
       suggests is she looked at the three points of
       information and she inferred from those three points of
       information that the reversal was undertaken by the
       postmaster, but the Credence system doesn't specifically
       say that.  She has made a mistake because she has put
       two and two together and made four, in fact five?
   A.  I think that really is a matter for Helen Rose.
   Q.  Would you accept it is possible?
   A.  It is possible that she was mistaken, are you asking,
       sorry?
   Q.  If we go over --
   A.  Sorry, I might have given the wrong answer.  Are you
       asking me is it possible --
   Q.  Are you suggesting that Credence did specifically state
       that the reversal was undertaken by the postmaster
       himself, rather than a reversal happened when the
       postmaster was logged on.  In fact it was the postmaster
       logging on that caused the reversal to happen?
   A.  I haven't looked at Credence myself in order to validate
       what the author of this report saw.  I have gone off
       what this paragraph says, that Credence clearly
       indicated that a reversal took place by the user.
       That's what I have based my evidence on.
   Q.  I understand, but I suggest to you that what you have
       read here is consistent with the view that what happened
       is Ms Rose misunderstood the significance of the items
       of information that were on Credence and formed the a
       mistaken conclusion?
   A.  I agree that that is possible, yes.
   Q.  Thank you.  Then if we go over to page 3 of the report
       {F/1082/3}.  For completeness I should say that on
       page 1 you will see that there are two -- this report,
       it is not really a report.  It is curious that it has
       some questions and then some answers that are provided
       by Mr Jenkins by e-mail and those answers are there in
       blue.
   MR JUSTICE FRASER:  I think you mean page 2.  We have gone
       to page 1 which is literally just the facing page.
   MR DE GARR ROBINSON:  I'm so sorry, I meant page {F/1082/2}.
       I'm sorry.
           So there are passages in blue which are quotations
       from emails she has received.  The first email is on the
       first page.  And the middle paragraph, just to be clear,
       this is the paragraph you relied on.  About halfway down
       it says:
           "The fact that there is no indication of such a
       receipt in the events table suggests the counter may
       have been rebooted and so perhaps may have crashed in
       which case the clerk may not have been told exactly what
       to do."
           I presume that was the basis upon which you said
       there were no receipts printed for this transaction,
       correct?
   A.  I think there is a more definitive statement than that
       later on in this document.  This is Gareth Jenkins
       suggesting that there wasn't a receipt.
   MR JUSTICE FRASER:  Is the blue Mr Jenkins?
   A.  Yes, I believe so, my Lord.
           Yes, it is the paragraph below where it says:
           "The reversal was due to recovery [and] was not
       an explicit reversal [made] by the clerk".
   Q.  What you are saying is this affirmatively states that no
       receipt was printed for the postmaster to tell him what
       to do.  That's what I'm asking you about, remember.
   A.  Well Mr Jenkins here, who has investigated it, has said
       from the logs that there wasn't a disconnected session
       receipt.
   Q.  And it was on the basis of that text, you said that in
       your report?
   A.  Yes.
   Q.  You remember, one of the claims you made in your report
       is that there was no session receipt printed?
   A.  Yes.
   Q.  If we go over the page, please, there was a second email
       that comes a couple of weeks later, at the top of the
       page {F/1082/3}.  It is the paragraph beginning:
           "The files 4 to 25th October ..."
           Do you see that?
   A.  Yes.
   Q.  If we can miss out the sentence that talks about those
       files.  The next sentence says:
           "Also row 70 of events 4 to 25 Oct ... shows that
       session 537803 ... has been recovered and this event has
       the same timestamp as the Reversal Session.  Also row 71
       of Events 4 to 25 Oct ... shows that a receipt was
       generated from the session 537805 (not explicitly, but
       it was the only session at that time)."
           So you will see that on the very next page,
       Mr Jenkins is saying actually the receipt was printed
       after all?
   A.  No, I think that's talking about a receipt for something
       else.  It is not a disconnected session receipt, I do
       not think.
   Q.  Are you suggesting that -- so he is talking about
       a completely different -- why would he be talking about
       a completely different session in this -- or rather why
       would she, Ms Rose, be quoting in this email
       a discussion about a completely different session?
   A.  They are actually talking about two sessions here.
       There is the session ending in 803 and the session
       ending in 805.
   Q.  Yes.  Perhaps I could read the next sentence.
   A.  Yes.
   Q.  "This receipt would have told the user that a Rollback
       had taken place (but the logs don't make that
       explicit)."
           Is that clear enough for you, Mr Coyne?  What
       Mr Jenkins is saying here is that a receipt was printed
       showing that the transaction had been rolled back,
       correct?
   A.  Right, so what Mr Jenkins is saying here is that the
       logs were missing the record that the receipt was
       printed but he believes the receipt was printed.
   Q.  Yes.  So your claim in your report that the report shows
       that the receipt was not printed, that claim is wrong,
       isn't it?  You hadn't read this document properly, had
       you?
   A.  Well, the situation here is that we have got to -- in
       order for this scenario that you are putting to me to be
       correct, we have got to assume that firstly the initial
       investigation showing that it was a user that issued the
       reversal was wrong, and then we have also got to assume
       a receipt was printed although it is not within the
       logs.
   Q.  Mr Coyne, in your report you specifically say, and
       I think you confirmed to me that it is your opinion,
       that the report -- this report shows that a receipt
       wasn't printed for the disconnected and reverse session,
       yes?
   A.  Mm.
   Q.  What I'm suggesting to you is that this report, if you
       can call it that, says nothing of the sort and that you
       haven't read it carefully enough, is that right or is
       that wrong?
   A.  Well, if Gareth Jenkins is correct then a receipt would
       have been printed.
   Q.  So would it be fair to say that in your anxiety to write
       a bad thing, to be able to write down a bad thing in
       your report about Horizon, you recorded what was said on
       the first page of the report, but you didn't look at the
       second page of the report which would have shown that
       that bad thing wasn't in fact correct?
   A.  No, but my point was about this report is to show that
       there is a difference between the view that you get of
       the data from viewing the Credence data from the ARQ
       data, and that's correct.
   Q.  Mr Coyne, if you just made that claim we would have been
       in and out of this issue within about five minutes.  The
       reason why we have spent about 20 minutes so far is
       because you made several claims, and I set them out
       orally and you agreed that you were making each of those
       claims on the basis of this Helen Rose report.  And what
       I'm suggesting to you is that the claim that we are now
       talking about is a claim that was wrong and that you
       should have known it was wrong if you had read the
       report properly?
   A.  I do agree that the report suggests that the receipt was
       printed.
   Q.  Isn't this another example of you taking a document that
       on a superficial reading could be said to say something
       critical about Horizon, and immediately writing that
       critical thing down without analysing the document
       properly to see what it actually said?
   A.  No.  This document does illustrate the point that I was
       making about the difference between Credence and ARQ
       data.
   Q.  And do you accept that a receipt -- are you suggesting
       the receipt was not printed -- are you giving up on your
       suggestion that a receipt was not printed?
   A.  The only evidence that we have here is that
       Gareth Jenkins is saying that the receipt was printed.
   Q.  So you are disclaiming --
   A.  No, no, I can accept that position.
   Q.  Very good.  For your Lordship's note, if one goes to
       {F/1095.1/1} there is an email between Mr Armstrong and
       Mr Warmington of Second Sight in which Mr Armstrong
       confirms that he did receive three receipts in relation
       to this reversed transaction at the time.  It is at page
       {F/1095.1/4} of that document.  I see it is up on the
       screen so let's have a quick look.
           To be fair to you, Mr Coyne, you wouldn't have seen
       this at the time you wrote your reports, I do not think.
   MR JUSTICE FRASER:  When was it disclosed?
   MR GREEN:  7 March, my Lord.
   MR JUSTICE FRASER:  2019?  Okay.
   MR DE GARR ROBINSON:  You will see that on 25th June
       Mr Armstrong writes an email to Mr Warmington of
       Second Sight and he says:
           "Having read your report I searched through the
       weekly records for the 4th October 2012 and found THREE
       disconnected session receipts all with the same session
       ID ..."
           If we go to the bottom of the page:
           "The time shown on these slips is 10.36 yet I had
       had the foresight to enter the time of 10.32 am on the
       customers bill alongside the amount paid of £76.09.
       This means that the customer had already left the office
       by the time these receipts were printed out by the
       system."
           So he had manually written the time of the
       transaction on the customer's bill, and one infers --
       would it be right to infer from that, Mr Coyne, that he
       hadn't actually got a receipt, he hadn't closed the
       basket and a receipt had been printed, so he realised
       something had gone wrong and he manually wrote the time
       of the transaction down on the bill he received from the
       customer, is that a fair inference?
   A.  Yes.
   Q.  So he knew something was wrong but he accepted the money
       from the customer and he allowed the customer to leave
       the premises?
   A.  Yes.
   MR JUSTICE FRASER:  I do not want to start a hare running,
       but just for my purposes that session ID of 537803 --
       this isn't a question for you, Mr Coyne, it is for
       counsel.  Can we just go back to the Gareth Jenkins blue
       extract because he mentions two sessions, 537803 and
       537805.  So are they different receipts from the 537805
       receipt that he is talking about?  It might be it
       doesn't matter.
   MR DE GARR ROBINSON:  I would strongly suggest, my Lord, it
       doesn't.
   MR JUSTICE FRASER:  But is your take on it that they are
       different receipts or is he talking about the same
       receipt?
   MR DE GARR ROBINSON:  My take on it, my Lord, is that one
       transaction was cancelled, it was this transaction, and
       the appropriate receipts were printed for it.
   MR JUSTICE FRASER:  Was that session 805 or 803?
   MR DE GARR ROBINSON:  My Lord, I'm afraid I haven't
       considered these documents sufficiently to answer that
       question.
   MR JUSTICE FRASER:  I don't want to start an unnecessary
       hare running.
           Right back to you, Mr de Garr Robinson.  {F/1082/3}
   MR DE GARR ROBINSON:  Coming to the next point, timings
       being wrong.  If we go back to 5.177 of your first
       report, that is {D2/1/102}.  So we have dealt with the
       question whether Credence affirmatively stated that the
       reversal was by the postmaster or by the system, and we
       dealt with the question whether a session receipt --
       session receipts, I should say, were printed or not.
           We now come to this further criticism which is that:
           "Observations of the disclosure illustrates that the
       initial report states 'a transaction at 10.42', whereas
       the credence data file shows 10.32 with the reversal at
       10.37."
           I would like to suggest to you, Mr Coyne, and we
       might be able to save some time, that the transaction
       was clearly at 10.37, indeed we have Mr Armstrong
       himself saying so in the email we have just read.
   A.  Yes.
   Q.  Clearly what happened is there is a typo in Ms Rose's
       report.  If we go to {F/1082/2}, please, at page 2.
           At the top of the page she says:
           "A transaction took place at Lepton ... on the
       04/10/2012 at 10.42 for a British Telecom bill payment
       ..."
           Then she then says in the next sentence:
           "At 10.37 on the same day the British Telecom bill
       payment was reversed out to cash settlement."
           Now, I would just like to give you an opportunity to
       correct what you are saying in 5.177.  Isn't it fairly
       clear that the reference to 10.42 here was her error,
       because you can't have a transaction that's reversed
       five minutes before the transaction is done.  In fact
       she should have written 10.32, yes?
   A.  Well, either she has got it wrong or the system has
       recorded it wrongly, I don't know which.
   Q.  Are you really seriously suggesting that Credence was
       indicating that the transaction was done at 10.42 and
       that that's a reason for suggesting, for thinking, that
       Credence is unreliable?  Is that really your contention?
   A.  Times on computers can be out.  They do drift.  It is
       possible that it's got the time wrong.  I agree with
       your position that it could well be a mis-key on behalf
       of Helen Rose.
   Q.  It wouldn't be a sound basis for suggesting that
       Credence wrongly records the time done of transactions,
       would it?  This wouldn't be a sound basis for making
       that claim about Credence?  What it is a sound basis for
       saying is that when people write documents sometimes
       they press the wrong keys, would you agree?
   A.  Yes.  It is one of those two scenarios, yes.
   Q.  As for the second point made at 5.177, that the ARQ data
       always works in accordance with Greenwich Mean Time,
       whereas everybody else at the time was working on
       British Summer Time, that's not a serious problem, is
       it?  It's not something that is going to cause great
       difficulties to anybody, is it?
   A.  As soon as you know that you are an hour adrift then it
       becomes very easy to deal with, but if you don't know
       that it is problematic.
   Q.  So are you imagining a world in which Mr Armstrong is
       provided with ARQ data but nobody tells him that ARQ
       data is based upon Greenwich Mean Time, is that your
       assumption?  And that's a problem, because nobody tells
       him that ARQ data is based on Greenwich Mean Time?
   A.  No, my answer is if you are told then it becomes very
       clear very quickly, but if you are not told it is
       confusing.
   Q.  But in 5.178, Mr Coyne, you seem to be assuming
       {D2/1/102}, remarkably, that no one would have told him.
       You say:
           "... it is not clear how easily it would have been
       to investigate issues where the Subpostmaster was not
       sure of what time things went on erroneously in the
       system ..."
           Why are you assuming that, having reached a point
       where the subpostmaster actually has the ARQ data, no
       one is going to help him understand that there is
       an hour discrepancy between the ARQ data and British
       Summer Time?
   A.  The point that I'm making is that unless somebody tells
       him it wouldn't be clear.  I do not think a user would
       typically know that the computer would be an hour out.
       I think the assumption would be that if it is an audit
       system of some description, that the clock difference
       would actually be dealt with correctly.
   Q.  What I would like to suggest to you, Mr Coyne, is that
       in this section what you are doing is you are trying to
       squeeze as much criticism as you can out of the
       Helen Rose report that you can level at Post Office.
       This isn't a fair-minded explanation of what happened,
       it is an exercise in trying to extract bad points as and
       where you can find them.  What would your response be to
       that?
   A.  That's not true.  And with regard to the time, when
       I point out that the time was wrong, the next paragraph
       explains how it is likely that it was wrong.
   Q.  You say "it is not clear how easy it would have been to
       investigate", that's a suggestion that in fact the
       subpostmaster would ...
           I'll read the whole of it:
           "... it is not clear how easy it would have been to
       investigate issues where the Subpostmaster was not sure
       of what time things went on erroneously ..."
           What are you saying that is a suggestion of?
   A.  Well, the subpostmaster might not necessarily know what
       the actual time was that the error took place.  So if
       they have got to then work out what time it actually
       took place precisely, and then look at two different
       times because the clock might be right on the audit log
       or it might be an hour forward or an hour backward on
       the audit log it just makes the process more difficult.
       But I do accept that if somebody explains to the
       reviewer that it is an hour behind, then that makes the
       process easier.
   MR DE GARR ROBINSON:  My Lord, I wonder whether this is
       a convenient moment.
   MR JUSTICE FRASER:  By all means.  We will have a 10-minute
       break.
   MR DE GARR ROBINSON:  Could we make it five minutes,
       my Lord?
   MR JUSTICE FRASER:  One of the transcribers has a back issue
       and that's why we are having 10 minutes.
   MR DE GARR ROBINSON:  Very good.
   MR JUSTICE FRASER:  But we can go on a little bit past 4.30
       if you are worried about losing time.  The trouble with
       five minutes is it is not really sufficient for current
       purposes.  So a 10-minute break and come back in at
       11.55 am.
   (11.45 am)
                         (A short break)
   (11.55 am)
   MR DE GARR ROBINSON:  My Lord.  Mr Coyne, we were talking
       about your criticisms of the use of Credence.  This
       isn't -- the discussion we have just had, the points we
       have just been discussing -- actually before finishing
       on this system, would you accept that what happened in
       the Lepton case was a customer gave cash for a BT bill
       payment to be made, in fact the BT bill payment was not
       made but the branch accepted cash for that payment, it
       therefore had a surplus of cash and so a TC had to be
       issued to correct for that surplus.  Do you accept that
       that's what happened?
   A.  Yes, I believe that that's what happened.
   Q.  So do you accept that the TC was not erroneously issued,
       in fact it was correctly issued?
   A.  Yes.
   Q.  Thank you.  Then let's move on to another criticism you
       have of Credence at {D2/1/101}.  This is your first
       point about Credence.  It is paragraph 5.174.  You say:
           "The End to End Reconciliation Reporting document
       from 27 February 2012 states:
           "There is no formal reconciliation produced between
       the POLSAP System and the Credence transaction stream.
       The Credence stream should therefore not be used to
       verify financial integrity and Post Office should ensure
       the POLSAP System Transaction information is used for
       this purpose."
   A.  Yes.
   Q.  That's one of the bases upon which you suggest that
       Credence shouldn't be used in order to make decisions on
       transaction corrections, right?
   A.  Yes.
   Q.  Okay.  Let's look at the document itself.  It is at
       {F/896/1}.
           I'm afraid I can't see the date on the version on
       the screen.
   A.  27 February 2012.
   Q.  Thank you very much.  It is called "End to End
       Reconciliation Reporting", so it is a document about the
       reconciliation process that we discussed yesterday.
   A.  Yes.
   Q.  Which is the process by which data goes into POLSAP and
       the data in POLSAP is then compared with client data,
       data from banks other institutions and that sort?
   A.  Yes.
   Q.  Then exceptions are identified and looked into?
   A.  Yes.
   Q.  Right.  If we could go forward to page {F/896/65}.
       Section 5 at the top of the page says "TPS
       Reconciliation Reports Specified".
           It starts by saying:
           "The Transaction Processing System (TPS) Report Set
       has been designed to enable reconciliation of the
       transactions carried out in Post Office branches using
       the Electronic Point of Sale Service (EPOSS) which are
       sent to POLSAP and POLMIS."
   A.  Yes.
   Q.  Just to be clear, the TPS system is the system which
       takes -- it is almost the highway which takes data from
       the Horizon branch database and transfers it into
       Post Office's own systems?
   A.  Mm.
   Q.  They are generally referred to as back office systems?
   A.  Yes.
   Q.  They are actually separate systems that belong to
       Post Office, yes?
   A.  Yes.
   Q.  In the course of that process there are -- that's when
       the reconciliation --
   A.  Yes.
   Q.  Once they arrive at POLSAP the reconciliation process is
       undertaken, yes?
   A.  Yes.  I mean reconciliation, this is talking about end
       to end reconciliation, so it is all the way through the
       whole -- the entirety of the systems.
   Q.  But when the comparison occurs -- the figures hit POLSAP
       and then the comparison with client data occurs, does
       it?
   A.  Yes.
   Q.  I see.  So:
           "The ... (TPS) Report Set has been designed to
       enable reconciliation of the --"
           Sorry, I have just read that sentence.  Let's move
       on:
           "The TPS exceptions report set identified herewith
       reports errors that have occurred within counter
       transactions or during the harvesting process.
           "NB: for the avoidance of doubt, there is no formal
       reconciliation produced between the POLSAP and POLMIS
       transaction stream.  The POLMIS stream should therefore
       not be used to verify financial integrity and Post
       Office Ltd should ensure the TPS Report Set and POLSAP
       transaction stream are used for this purpose."
           This is the document you referred to and it refers
       to POLMIS here.  Actually there is a later version of
       this document that refers to Credence.
   A.  Yes.
   Q.  It may be that the document reference you have given is
       erroneous?
   A.  I think that that's right.  It must be a later version
       of the same document.
   Q.  I presume you didn't check all the document references,
       it would have been very difficult for you to do that.
   A.  I have put what's called a MD5 reference at the bottom
       of there, so I'm not sure.  I don't think there's any
       easy way of checking that.
   Q.  I have found a later version of the document that does
       refer to Credence so I'm not going to challenge you on
       that, Mr Coyne.
   MR JUSTICE FRASER:  Would you give me, for my note, that
       reference at some point.  You don't have to do it now.
   MR DE GARR ROBINSON:  Yes, my Lord.  It is {F/1686/1}.
   MR JUSTICE FRASER:  Thank you very much.
   A.  Sorry, my footnote does say 27 February and the date at
       the bottom of this one is 22 June.
   MR JUSTICE FRASER:  No, this is 27 February but I think
       an earlier version was 22 June of 2011.
   A.  Forgive me, my Lord, I'm just looking at the bottom of
       what's on the screen at the moment, towards the bottom
       right.
   MR JUSTICE FRASER:  Yes, that's because you can't see the
       way the colours have been struck through.
           That is right, isn't it, Mr de Garr Robinson?
   MR DE GARR ROBINSON:  My Lord, I believe so.
   MR JUSTICE FRASER:  For some reason I have got in my
       {F/896/1} on my trial bundle, I have got a coloured
       track change version of the same document which does
       show 22 June crossed out.
   MR DE GARR ROBINSON:  Mine does --
   MR JUSTICE FRASER:  Yours does as well?
   MR DE GARR ROBINSON:  I have a hard copy and mine does,
       which I printed off the trial bundles, my Lord.
           Should I say "from" the trial bundles?  I never
       know.
   MR JUSTICE FRASER:  That's fine.
   MR DE GARR ROBINSON:  If we go back to page {F/896/8} of
       this document, I hope it is of this document and not the
       1696 version.
           It explains, if we look at section 2, "Scope":
           "This document defines the format and content of all
       reconciliation reports for HNG-X ..."
           That is Horizon Online?
   A.  Mm.
   Q.  "... which satisfies the DRS, APS and TPS reconciliation
       requirement."
           These are all different forms of reconciliation.
       Can you take us through them.  What's DRS?
   A.  It is something reconciliation service but I can't think
       what.
   Q.  APS?
   A.  Automated payment system.
   Q.  Then we have TPS.
   A.  Transaction --
   Q.  These are all separate systems beyond, as it were, the
       branch data?
   A.  Yes.
   Q.  It is beyond branch accounts.  This is information taken
       from the BRDB and pushed through those streams to allow
       different forms of reconciliation to be undertaken?
   A.  Yes.
   Q.  It does not attempt to define within the operating
       systems how the transactions are processed.
           "This document does not attempt to define the
       business processes undertaken within Fujitsu Services
       and Post Office Ltd with respect to the resolution of
       any exceptions which may arise, nor does it scope the
       requirement for any systems that may be required to
       assist in this process.  This information can be found
       in the associated documents."
           So it is just talking about the format and content
       of all reconciliation reports and it doesn't talk about
       the business processes undertaken with respect to the
       resolution of any exceptions.
           Now, if we go back to -- so just to be clear, it has
       nothing to do with the Post Office business processes
       leading to decisions on transaction corrections, does
       it?
   A.  It does -- well, the decision to make a transaction
       correction is a comparison between, in rudimentary
       terms, front end and back end systems.  Because what is
       going on here through transaction processing has
       a potential to change transactions in the back end.
   Q.  Mr Coyne, I'm getting slightly concerned about time.
       I asked quite a simple question which was this
       document -- paragraph 2, section 2 that we have just
       read, makes it clear it is not about the business
       processes which lead to the decisions made to issue
       transaction corrections.  That is a yes or no answer, if
       I may suggest.  Could you give me one?
   A.  Yes, this document does not go to --
   Q.  Thank you.  It is not what this document is concerned
       with at all, is it?  It is concerned with the
       reconciliation process.  And that process may end up
       leading to investigations that result in decisions being
       made but it is not about that side of the divide at all,
       is it?
   A.  No, that is right.
   Q.  Thank you.
   A.  I have the DRS, it is data reconciliation service.
   Q.  Thank you.
           Then if we go back to page 65 {F/896/65}, this is
       going back to section 5.  This is describing the process
       by which exceptions are identified, yes?
   A.  Yes.
   Q.  Then the various reports that are produced are then
       discussed.  So at 5.1 there is the TPSC250 report, "Host
       Detected Transaction Control Errors":
           "This report is produced daily and shows detail for
       any Post Office branch where the control totals for the
       transactions output by the Host to POLFS and POLMIS do
       not match the daily transaction totals calculated by the
       counters."
           So that is quite a good check.  It shows if there is
       a discrepancy between what the counters have done and
       the information going into Post Office's systems.
       That's quite a useful check, isn't it?
   A.  Yes, it is a report that's available to the Post Office.
       It isn't given to the branch I don't believe.
   Q.  No, it's not.  I think actually it is given in the first
       instance to Fujitsu.  It is only given to Post Office if
       Post Office request it, that is right, isn't it?
       Because Post Office isn't involved in this, it is
       a Fujitsu process, correct?
   A.  There is a suite of reports that is handed over between
       Fujitsu and Post Office automatically each morning.  I'm
       not sure whether this is one of the reports that's sent
       to them.
   Q.  If one goes over the page, 5.2, TPSC254.  This is
       another form of report that's generated each day, yes?
       {F/896/66}
   A.  Yes.
   Q.  This is called "Harvester Exceptions":
           "This report is produced daily and shows a list of
       exceptions detected by the BRDB copy process when
       failing to process one or more messages."
           These are all countermeasures, aren't they?  They
       are the sort of countermeasures Dr Worden is talking
       about, spotting discrepancies between data that ought to
       be the same.  It is a good example of redundant data
       storage and MID and all those other acronyms that
       Dr Worden uses, correct?
   A.  These are reports that are available so as long as
       somebody looks at the reports they should be able to
       pick it up.
   Q.  Mr Coyne, you are not suggesting that people don't look
       at -- people look at these reports every day, don't
       they?  These are the reports that produce all those
       exceptions about which you make so much hay in your
       report?
   A.  They certainly should do, yes.
   Q.  Are you suggesting -- let me get -- are you suggesting
       that people don't look at these reports?  Have you seen
       evidence to suggest that people don't look at these
       reports?
   A.  No, what I'm saying is somebody needs to read the
       reports.
   Q.  Very good.  Then if one goes over the page to 5.3
       {F/896/67}, TPSC257, that's "POLFS Incomplete Summaries
       Report".  That's another daily report, isn't it?
   A.  Yes.
   Q.  "This report identifies all Post Office branches on a
       daily basis in which the net total of transactions
       (debits/credits) does NOT net to a value of zero."
           So this, for example, picks up receipts and payments
       mismatches, doesn't it?
   A.  Yes.
   Q.  So if there is a receipts and payments mismatch at any
       branch on any day of the week it will be automatically
       reported to Fujitsu who will be aware of it and can
       investigate, isn't that right?
   A.  Yes, I believe that this is the report that's printed to
       indicate that, yes.
   Q.  And would I be right in thinking that you have seen
       hundreds of PEAKs which show that Fujitsu do absolutely
       investigate these exceptions when they arise?
   A.  There are certainly PEAKs that talk about the
       investigation from these incomplete summary reports,
       yes.
   Q.  I'm interested, would you accept that there are lots of
       PEAKs that do that?
   A.  I don't know exactly what the number would be but there
       are a number, yes.  There are many.
   Q.  Here's what interests me about that answer, Mr Coyne.
       You are perfectly happy when you see an example of
       a handful of things happening to say things often happen
       when they favour a case -- that they help build a case
       that Horizon is bad.  But when I ask you a simple
       question, "You have seen lots of PEAKs in which these
       exceptions are investigated?" you are unwilling even to
       concede that it happens a lot of times.  I'm quite
       interested in why you should have a different attitude
       depending on whether or not something is a criticism of
       Horizon or in praise of Horizon?
   A.  I'm attempting to be as precise as possible with my
       answers to you.
   Q.  When it comes to saying something positive about Horizon
       you are very precise indeed.  Can I suggest to you, Mr
       Coyne, that you are rather less precise when it comes to
       criticising it.
   A.  That's certainly not my intention.
   Q.  Let's go back to page {F/896/65}.  This is the third
       paragraph under section 5:
           "NB: For the avoidance of doubt there is no formal
       reconciliation produced between the POLSAP and POLMIS
       transaction stream."
           We can call that Credence.
           "The POLMIS stream should therefore not be used to
       verify financial integrity and Post Office Ltd should
       ensure the TPS Report Set and POLSAP transaction stream
       are used for this purpose."
           You appear to suggest in the paragraph of your
       report that we have just read, paragraph 5.174, that
       this is an indication that Post Office should not be
       using Credence for the purposes of making decisions
       about transaction corrections.  That is your claim,
       isn't it?
   A.  Yes.
   Q.  Could I just suggest to you, Mr Coyne, that when this
       report is talking about financial integrity that's
       a reference to the integrity of the financial data, it
       is a reference to ensuring that the data for the given
       day is complete so that it can be used for
       reconciliation.  It is not a statement about what should
       be done when making decisions on transaction
       corrections?
   A.  But some of the information that is reported here and
       finds its way back into POLSAP will be required in order
       to make a decision on whether to issue a transaction
       correction or not.  And if they are not reconciled
       together, you could have the scenario where Credence
       data differs from POLSAP data.
   Q.  So as I understand it, you are using this as part of
       an argument -- and it becomes a theme of your second
       report -- that Credence data shouldn't be used for the
       purposes of deciding on TCs, actually it should be ARQ
       data?
   A.  My point is that Credence data alone shouldn't be used.
       The ARQ data will give the full picture of what went on
       at the counter.
   Q.  If in this report they are talking about that process,
       and I have already suggested to you, Mr Coyne, that that
       question, the data that is used for the purposes of
       transaction correction decisions, has got nothing to do
       with this report.  The writer isn't concerned with that.
       I have already suggested that to you and I think you
       have accepted it.  But are you suggesting that the
       writer has decided to say something that's outside the
       scope of this report because he is concerned that
       inappropriate data is being used for the purposes of
       making transaction correction decisions?  Is that how
       you construe this paragraph?
   A.  Well, I mean I don't exactly know what was in the mind
       of the author when they put this together, but they saw
       fit to put a specific note to say that there was a doubt
       over what should and shouldn't be used to verify
       financial integrity, and what should be used to verify
       financial integrity is the TPS reports in POLSAP.
   Q.  Mr Coyne, from the get-go Post Office has used its
       management information systems in order to decide on
       whether or not to issue transaction corrections, is that
       right?
   A.  Yes, I would think so.
   Q.  And it has used Credence and any predecessor -- I'm
       presuming here that POLMIS might be a predecessor of
       Credence -- would that be right?
   A.  It is Post Office Management Information System.
       Whether that later became Credence or not, I would have
       to check.
   Q.  I'm afraid I don't know.  That was a genuine question.
           So we have a business, the Post Office, which has
       had a practice since the beginning of making decisions
       in relation to transaction corrections based upon its
       management information systems?
   A.  Its range of management information systems.
   Q.  And you are suggesting, are you, here in this paragraph,
       that the writer of this report in 2012, February 2012
       and thereafter, is suggesting that what Post Office has
       been doing for the previous 12 or 13 years is completely
       wrong?  Do you honestly think that that's what the
       writer of this sentence was intending to convey?
   A.  No, I don't think they are saying that what you have
       been doing for the last 12 years is completely wrong.
       They are providing a warning that you should use one set
       of systems rather than another set of systems because
       the two do not reconcile.
   Q.  And what I would like to suggest to you, Mr Coyne, is
       that when this report talks about financial integrity,
       it is talking about the integrity of the data that's
       compared as between the client and Post Office.  It is
       not talking about the process of making decisions about
       transaction corrections.  Do you not accept that?
   A.  But the integrity of the data between the Post Office
       and its clients could well have an impact on branch
       accounts, because if there's an issue between
       Post Office and its clients, the client will report
       a different view of the transaction.
   Q.  Let me move on.  Let's move on to the conclusion that
       you then draw from the passages that we have seen, the
       Helen Rose report, and the end to end reconciliation
       reporting.
           The conclusion you draw is that when faced with
       a problem, an apparent discrepancy, an apparent
       exception in accounting figures, when therefore called
       upon to make a decision about whether to decide on
       a transaction correction or not, Fujitsu and Post Office
       should always use the raw ARQ data that's held in the
       audit store, that is your claim, isn't it?
   A.  In order to get the definitive position on it they
       should, yes, because that is a record of what actually
       happened.
   Q.  So let's take this in stages.  It is a good thing for
       a complex system like Horizon to have a secure place to
       store a copy of all the transaction data that comes in
       from branches, isn't it?
   A.  Yes.
   Q.  Because one can then go back to look at that pristine
       copy months or years later if there is a concern about
       the accuracy of the data in the management systems,
       correct?
   A.  Yes.
   Q.  And the whole point of an audit store is that the data
       in it is effectively locked away in a secure place and
       only extracted when it is necessary for checking against
       the other sources, correct?
   A.  Yes.
   Q.  Now, were you in court when Mr Dunks gave evidence about
       the process of obtaining data from the audit store, do
       you remember?  Were you here when he gave that evidence?
   A.  I'm not sure that I was.
   Q.  You will recall his witness statement where he describes
       it, yes?
   A.  Yes.
   Q.  It is a slow and careful process, isn't it, extracting
       the data in a reliable way?
   A.  I don't know if "careful" is the right word but it
       certainly would be slow, I can imagine.
   Q.  And it is done from a very small number of very
       carefully managed secure sites, correct?
   A.  Likely, yes.
   Q.  And it is labour intensive and quite expensive, correct?
   A.  I can't imagine why it would be labour intensive.
       I imagine you'd put a search into the computer system
       and press go, I would imagine, and it would --
   Q.  Mr Dunks' witness statement describes the care with
       which these processes are undertaken, the care with
       which access to the relevant systems is carefully
       controlled.
   A.  Yes.
   Q.  It is only particular people that are allowed to do that
       particular job --
   A.  Indeed.
   Q.  -- and they have to have particular authorisation and
       particular qualifications?
   A.  Yes.
   Q.  And I think you have already accepted it can be a slow
       process?
   A.  Yes.
   Q.  Particularly if a large amount of data is being
       extracted you would accept, would you, that it could
       take weeks and weeks for really huge quantities of data
       to be extracted?
   A.  I would be surprised if that was the case.  But I mean
       typically you would be extracting a day's worth of
       transactions, perhaps even less than that, to understand
       what went on around the particular hour or --
   Q.  And it is quite an expensive process, isn't it?
   A.  I believe that there are -- there is a certain number of
       requests that can be made within Fujitsu's service level
       agreement and then after that there is a charge that's
       made, yes.
   Q.  And the charge that's made over the allowance of 720
       a year, it is over £200, are you aware of that?
   A.  I think I did see that figure, yes.
   Q.  Right.  And what you get when the data is extracted is
       not data organised into the form of elaborate reports,
       it is raw data which actually needs packaging even to
       put it in a spreadsheet.  It is very difficult to manage
       this kind of data, isn't it?
   A.  I believe the process is that there is a raw version but
       there is also a version that is packaged so it can be
       read in Excel.
   Q.  You mean in a spreadsheet?
   A.  In a spreadsheet.
   Q.  Do you mean the spreadsheets that the witnesses -- the
       claimant witnesses were taken to during the course of
       their evidence at the beginning of the trial?  Because
       my experience of those spreadsheets is that they are
       very difficult to manage your way through, but would you
       suggest not?
   A.  Absolutely.  You would have to be reasonably experienced
       in interpreting the data that's given to you, it would
       be quite a complex spreadsheet.  But it can be opened up
       in a conventional spreadsheet.
   Q.  The controls and checks described by Mr Dunks are what
       you would expect if the idea is to have a gold standard
       store of data that cannot be altered or corrupted or
       lost in the meantime, yes?
   A.  Yes.
   Q.  Perhaps I could go to your second report now at
       {D2/4.1/7}.  This is your executive summary of your
       second report.
   A.  Yes.
   Q.  In paragraph 1.2 you say:
           "I consider that Horizon is less robust than as
       originally expressed in my first report.  My primary
       reasons for this are as follows ..."
   A.  Mm.
   Q.  And you talk about remote access.
           If you go down to paragraph (c).
   A.  Yes.
   Q.  "Post Office do not consult the full audit data before
       ruling on a discrepancy, instead using third party
       client reconciliation data or subsections of the audit
       data from within Credence or HORice."
   A.  Mm.
   Q.  So this is something -- your discovery that this was
       happening is something that caused you to have a change
       of heart on robustness, is that right?
   A.  Yes.  It was my original belief that the audit data was
       consulted.
   Q.  So when you drafted your first report you thought that
       every time Post Office was faced with some kind of
       discrepancy that might lead to a TC, you thought in
       every case regard was had to the full audit data that
       was held in the audit store, did you?
   A.  Well, certainly that was my original opinion, yes.
       I thought that was the purpose of the audit store, to
       actually go back and see what happened at branches.
   Q.  But you knew, Mr Coyne, that the audit store was copied
       from the BRDB and sealed and maintained for seven years,
       didn't you?
   A.  But it only needs to be sealed from a write perspective.
       You can seal something and still have read access to it.
       There is generally no problem with that.  That doesn't
       tamper with any seals --
   Q.  Didn't you know, in fact wasn't it obvious, bearing in
       mind all the controls that we just discussed with
       Mr Dunk's report, witness statement and so on, that
       Post Office would generally rely on its own management
       information systems when making decisions on transaction
       corrections?  Wasn't that obvious to you?
   A.  No, it wasn't obvious to me.  I perceived that the
       management information systems would be part of it, but
       that to get the true picture of what had happened at the
       branch the audit data would be consulted.
   Q.  Well, if we could go back to your first report, it is
       {D2/1/119}.
   MR GREEN:  My Lord, in fairness to the witness, it does
       pre-date the witness statement being referred to.  The
       witness statement is November.
   MR DE GARR ROBINSON:  I'm grateful to my learned friend.
       Thank you.
           If we look at paragraph 6.46, you will see it is
       under the heading "Reconciliation Summary".  This is
       your first report, yes?
   A.  Mm.
   Q.  You say:
           "In consideration that Branch account positions were
       interpreted and reviewed from data flows through to Post
       Office back end systems (which would determine whether
       Transaction Corrections were to be applied), the
       following is considered relevant."
           {D2/1/120}
           Over the page you say:
           "POLSAP – Following investigation by Fujitsu, Logica
       and Ingenico, the root cause of a long outstanding
       problem with missing data within POLSAP was identified
       as out of range dates which failed the Credence
       validation (in excess of 90 days).  Ingenico has
       corrected the data and P&BA has advised that the
       mismatches have been cleared ..."
           Here you appear to be saying, indeed you appear to
       be raising it as a criticism of Post Office, that when
       making management -- decisions on transaction
       corrections, Post Office were using management data that
       could be wrong.
           Now I would like you, if you would, to explain why
       having made that criticism there you claimed just three
       and a half months later in your next report that in fact
       you made the opposite assumption, namely, that
       Post Office always looked at all the core audit data?
   A.  Sorry, I don't understand the question.
   Q.  This is your first report, paragraph 6.46 is your first
       report.
   A.  Yes.
   Q.  And we just read your second report where you said: when
       I produced my first report I believed that when
       decisions were made about transaction corrections the
       full ARQ data was used.  Correct?
   A.  Yes.
   Q.  Now we go to 6.46 and here you are saying that
       management information systems, the back end systems,
       were used to determine whether transaction corrections
       were to be applied, and you give as an example, over the
       page, POLSAP?
   A.  Yes.
   Q.  Now, ARQ data, the core audit store, that isn't back
       end, is it?  That's not held by Post Office and used by
       Post Office for its systems, it is an entirely separate
       process that's maintained by Fujitsu, isn't it?
   A.  Yes.
   Q.  So here in your first report you appear to be saying
       that there is a problem with the process by which
       Post Office decides transaction corrections because they
       are using Post Office management systems that might be
       unreliable?
   A.  Yes.
   Q.  But if you believed that when Post Office made those
       decisions actually they used the full ARQ audit data,
       that criticism would be utterly misconceived, wouldn't
       it?  So either you were telling the truth -- or either
       you believed when you did your first report that
       management systems, not ARQ data, was used, or it is the
       position that you were making a criticism of the use of
       management systems even though you believed that the
       full audit data was actually used, but it can't be both.
   A.  I think it can be both.  In my first report it was my
       perception that the range of management information
       systems and the ARQ data should be used, and in my
       second report -- well, before my second report
       I discovered that the ARQ data was not used.
   Q.  I see.  So let's look at paragraph 6.46 again.  In that
       paragraph, forgive me, but it appears to be yet another
       criticism of the method by which Post Office conducts
       its business and the criticism appears to be that
       Post Office is using a form of information which is
       unreliable {D2/1/119}.
           If you had wanted to be balanced in your approach to
       that criticism, would it not have been appropriate for
       you to say: but I do of course recognise that this is
       only one subset of the information that Post Office used
       and I do understand Post Office actually used the full
       audit data as well?  Would a need for balance not have
       required you just to make that point clear?
   A.  Certainly if I had included that it may have helped the
       reader, yes.
   Q.  If we could now move back to your second report, it is
       {D2/4.1/228}.  Actually I'm so sorry, I have taken you
       to the wrong page.  Could we go to 114 rather than 128
       {D2/4.1/114}.
           You say in paragraph 4.67 under the heading "ARQ
       Requests":
           "In her statement Ms Mather references the number of
       ARQ requests per year.  If it is correct that the
       contractual limit of 720 per year has never been
       exceeded except for this litigation, then in my view
       Post Office is not utilising the audit data sufficiently
       and certainly is not checking the audit data prior to
       issuing transaction corrections."
   A.  Mm.
   Q.  Then at 4.68:
           "in 2011/2012 using the figure that Dr Worden
       produces at his Table 9.3 (section 9.6, page 208) there
       were 107,583584 Transaction Corrections but only a
       fraction 213 of these were validated by the audit data."
   A.  Mm.
   Q.  So you are suggesting, are you, that full audit data
       should have been extracted from the database in at least
       107,584 occasions during that year?
   A.  Yes.  I believe that the audit data should be consulted
       every time there is a potential dispute and need to
       issue a transaction correction.  In light of now
       understanding the process of getting at the audit data
       and getting it extracted, I see that it wouldn't be
       possible to do that.
   Q.  It would also cost something like -- well, because of
       course you wouldn't only look at ARQ data when actually
       issuing a transaction correction.  Would I be right in
       thinking that your view would be that every time there
       is an exception, a discrepancy, a full audit data ought
       to be looked at as well?
   A.  Certainly a section of the audit data, yes.  You don't
       have to look at the full audit data.  On a lot of
       systems if you believe something happened between
       10 o'clock and 11 o'clock, you can go to a screen, put
       10/11 o'clock on a certain date, press go, and it will
       give you a list of all the actions that happened on that
       day.
   Q.  I don't believe the audit still works like that.  Is
       that something you are aware of?
   A.  No, I now understand the audit data doesn't operate like
       that and I now understand there are costs associated
       with it, but I don't believe that was understood --
   Q.  So let's say there are 107,000 TC decisions made a year.
       The number of decisions that don't result in TCs, let's
       pluck a figure out of the air, I have no idea, it could
       be an equivalent number, it could be much more actually.
       Let's say 250,000 decisions a year.  If each of those
       audit requests cost £200, we are looking at £50 million
       a year, aren't we?
   A.  Yes.
   Q.  Another point that interests me is: is it really the
       case that you are assuming that although the audit data
       was kept separate, it was nonetheless available in some
       kind of information stream in a similar way to POLSAP
       and to Credence?  Is that how you thought it worked?
       Because it bears no relation to how the audit store was
       actually operated and maintained.
   A.  I have got experience of working or designing audit
       stores for a number of different systems and they don't
       have to work in the way that they have been defined
       here.  Audit systems are often very easily accessible to
       be able to be read by certain users.  There's nothing
       inherently difficult about that.  I accept that the
       write aspect of it, you know, you wouldn't want people
       writing to an audit database.  But having the ability to
       read to it is something that's quite simple -- read from
       it is quite simple.
   Q.  And would you accept that the closer you get to raw
       data, the more you need specialist knowledge and skills
       to interpret that raw data?
   A.  You do, but the extraction or the report that's run from
       the audit data would typically handle the presentation
       aspect of it.  There might be lots of 0s and 1s at the
       back, but the report generator can often put it together
       in quite a usable form --
   Q.  So are you suggesting that when you were giving your
       first report, your first opinion, you believed that
       there was a process by which the core audit store that
       was kept separately by Fujitsu, and there are references
       in your report explaining how separately they were kept,
       that actually there was a system that extracted data
       from that audit store on a regular basis, perhaps on
       a continual basis, and packaged that information into
       easy to use reports that gave you all sorts of
       information that you would need in order to make
       transaction correction decisions, is that what you
       believed was happening?
   A.  I would not characterise it in the way you have done
       there, but my perception is if Post Office needed to
       hone in on a particular area, whether it be an hour or
       a day of what happened at a branch, that they would have
       a way of viewing that audit data for that period on
       a screen or by pulling a report.
   Q.  What I would like to suggest to you, Mr Coyne, is that
       if that had been your apprehension, if that had been
       your belief, you would have -- these would have required
       their own quite sophisticated systems and you would have
       been aware of those systems.  You have a vast amount of
       technical documents explaining all the different systems
       in operation and how they fitted together, including
       documents relating to the audit store, you would have
       seen that there was a system of that sort.  Indeed
       I rather imagine that Post Office would have been quite
       happy to come forward to explain how that process
       happened.  But instead, you are saying that you assumed
       that all this happened even though you had seen no
       documentation to support such a belief at all, and I'm
       asking you, Mr Coyne, can that really be right?
   A.  Well, the purpose of having an audit of what happens at
       branch counters is so that if there is a dispute over
       what has happened that somebody, presumably this will be
       Post Office, can have a very quick look at what happened
       and find out the truth.  That's the purpose of having
       an audit store.  There is no other reason for it other
       than looking back at what actually happened.  It is my
       perception that that look back was available to people
       at the Post Office.
   Q.  Could I suggest to you, Mr Coyne, that you knew very
       well that the system that Post Office used for the
       purposes of having what you describe as a quick look
       were its management information systems.  The hint is in
       the name, MIS, management information systems.  And you
       would expect, in the absence of being told to the
       contrary, that a business such as Post Office would use
       its management information systems for making business
       decisions of that sort?
   A.  And the audit database would be part of that management
       information system.
   Q.  Mr Coyne, if that were true you would certainly have
       seen documents explaining that amongst the management
       information systems of Post Office was an audit store
       that was maintained in an entirely separate facility
       that was owned by an entirely separate company and was
       maintained separately and had no connection with the
       outside world, would you not?
   A.  No, I don't believe that that was the case.  If you are
       making decisions based purely on a cut-down version of
       the data in a management information system you have to
       decide what data is going to be cut out of that.  So if
       everything is in the audit store and just a portion of
       that data is in your management information systems,
       then you are going to necessarily make a decision based
       on a subset of the data and not the whole of the data.
   Q.  I understand the logic of your position, Mr Coyne.  What
       I'm seeking to explore with you is whether it bears any
       relation to reality.
           Data comes out from the BRDB in streams.  It comes
       out -- copies of data come from the BRDB and go into
       Credence.  Copies of data come out from BRDB and go into
       POLSAP.  They go into other management information
       systems maintained by Post Office.  Entirely separately,
       and the word "separate" is in your first report,
       information goes out to a sealed audit store, the word
       "sealed" is in your report, where it is kept for seven
       years.
   A.  Yes.
   Q.  And what I'm suggesting to you is that there is no basis
       upon which you could ever have thought that the
       information in that audit store could be regarded as
       a Post Office management information system?
   A.  I believed that it was a system that Post Office would
       look at whenever there was a dispute about what happened
       at a branch counter.  I believed that they would have
       access to that.
   Q.  Did you see any technical documents indicating a route
       by which information from the sealed audit store was
       made available on a read only basis to Post Office?
   A.  No.
   Q.  Did you see any PEAK or any other document, any -- well,
       OCPs, it is too early for that.  But did you see any
       documents of any sort indicating or referring to the
       stream of data flowing on a continual basis out of the
       audit store into Post Office's management systems?
   A.  No, but that's not how things would work.  If
       Post Office wanted to get access to the data in the
       audit store they would go to a screen or go to
       an application on their computer and they would run the
       request for that data.
   Q.  Mr Coyne, I would like to suggest to you that it is
       completely unrealistic to think that a separate sealed
       core audit store of the sort we're talking about should
       be cracked open hundreds of times a day in preference to
       using management information systems which are designed
       for that precise purpose?
   A.  I think the word "sealed" is misleading and the concept
       of cracking something open to get access to it I think
       is misleading as well.
           Things in an audit store are only -- can be written
       to and only written to once, and the term that's often
       used is write once read many, WORM.  So the process is
       written to once, but people can read from that store on
       many occasions.
   Q.  But just to be absolutely clear, you had not and indeed
       you have not seen any documents suggesting that
       Post Office had the ability to gain access to the audit
       store on its own systems, had you?  There was no design
       facility, there was no -- there were no lines of
       communication between the audit store and Post Office in
       any document you had ever seen, correct?
   A.  No, it looks as if the majority of the references to
       audit database access was from Fujitsu personnel.
   Q.  And one final thing.  Would I be right in thinking that
       now that you understand how the audit store actually
       works and the costs and delays associated with
       extracting data on a large basis from the audit store,
       would you accept that it would be disproportionate to be
       using the audit store as a basis for making decisions on
       transaction corrections in every single case?
   A.  Yes, it would seem that it would be very expensive and
       very slow to access the audit store, and effectively for
       the number of transaction corrections you couldn't do
       that, and therefore you accept that you make decisions
       on the management information systems rather than the
       audit store.
   Q.  In your evidence yesterday we discussed your approach,
       remember, to whether and to what extent Post Office and
       Fujitsu did things on a cost benefit basis?
   A.  Yes.
   Q.  In the course of that evidence I recall you indicating
       that you regarded it as important to ascertain whether
       the possibility of error was reduced as far as possible.
       Do you remember that exchange that we had?
   A.  Yes.
   Q.  Was it your objective in your reports to address that
       question?
   A.  Was it my objective at the outset to address that
       question?
   Q.  To consider not whether the risk was reduced as far as
       reasonable or to consider whether the risk was reduced
       as far as practicable, but to consider whether the risk
       was reduced as far as possible, which is a much more
       exacting standard?
   A.  I believe that was the word that was used in the Horizon
       Issues.
   Q.  So would the answer to my question be yes, that when you
       produced both your reports you did so with the objective
       of applying that test when determining whether something
       constituted a problem in the system or not, whether it
       satisfied the test of reducing a risk as far as
       possible?
   A.  Yes.
   Q.  And not just with -- and did that inform -- does that
       inform actually the approach, the criticisms you make of
       the use or non-use of ARQ data in your second report?
   A.  Yes.
   Q.  But if you take a step back and consider questions such
       as proportionality and reasonableness, would you take
       a different view on that question and perhaps some other
       questions too?
   A.  As I understand it, the question was reduce as far as
       possible.
   Q.  Yes.
   A.  So that is the way I answered that question.
   Q.  Could we go to {C1/1/1}, please.  This is the Horizon
       Issues.  I don't want to take more time than is
       necessary, but I would like to give you an opportunity
       to tell me which of these Horizon Issues raises that
       question as a test.
           (Pause)
   MR JUSTICE FRASER:  Are you looking for it in your report?
   A.  I am, sir.
   MR JUSTICE FRASER:  The list of issues is at page 3,
       I think, of your first --
   A.  Thank you.
   MR JUSTICE FRASER:  You can only see one page at a time on
       the screen.
   MR DE GARR ROBINSON:  Absolutely, my Lord.
   A.  That's okay.
           (Pause)
           The reference at Issue 6 at 116 is reduced to
       an extremely low level, the risk.
   Q.  Yes.  So it is a factual question as to how low level
       the risk was.  It is not a question whether Post Office
       had reduced the risk to the lowest possible level, is
       it?  I'm just wondering, Mr Coyne, whether you may have
       applied in your entire approach to your reports the
       wrong test for the purposes of these proceedings.  Do
       you think that's possible?
   A.  No, I don't believe so.  I mean, it is not defined what
       an extremely low level is.
   Q.  But you do accept that as low as possible, that was the
       test that you used when approaching both your reports.
       I think you have already accepted that?
   A.  Yes.
   Q.  Thank you.  Let's move on to a different subject.
       Perhaps I can deal with this quickly.  I would like to
       talk about PEAKs and KELs.
           From what you said yesterday about your change of
       mind on robustness between the first joint statement and
       your first report, I imagine you would agree that the
       system of KELs and PEAKs that Fujitsu developed was
       quite a thorough system?
   A.  Yes.
   Q.  And that you formed the view that members of the SSC
       were very familiar with the Horizon system?
   A.  Yes.
   Q.  And they were very familiar with the PEAK and KEL
       system?
   A.  Yes.
   Q.  And with their training and experience and with using
       search facilities they were able to navigate that system
       quite well?
   A.  Yes.
   Q.  Notwithstanding the limitations that you have fairly
       identified.  And that using search facilities they were
       often able to find PEAKs or KELs addressing similar
       problems to the ones that they were facing?
   A.  Yes.
   Q.  And would you agree that the PEAKs show, generally show,
       the thoroughness with which they generally worked?
   A.  Yes.
   Q.  And they tended to keep a written record of what they
       did step by step in PEAKs, didn't they?
   A.  Yes.
   Q.  It wasn't comprehensive, no one is suggesting it is
       comprehensive, but it's quite a process-driven process,
       one doesn't often see something significant happening
       that isn't somewhere recorded or alluded to in the PEAK
       during the different processing steps that are described
       as you go down the PEAK from the top.
   A.  Yes.
   Q.  So in the scheme of things, compared with other systems
       with which you are familiar, you would accept, wouldn't
       you, that this is actually quite a well organised, well
       run system?
   A.  Certainly the way of recording the information in the
       PEAKs and KELs is a reasonably good system, yes.
   Q.  Thank you.  Now, I would like to ask you about something
       you say in your second report which is at {D2/4.1/176}.
       It is 5.186, Mr Coyne.
   A.  5 point what, sorry?
   Q.  5.186 at page 176 of the trial bundle.
   A.  Yes.
   Q.  You say:
           "At Dr Worden's paragraph 488 he suggests that
       serious bugs are rare in the KEL and PEAK records. I
       agree, they are rare in the KEL records because the
       purpose of KELs are to inform support personnel how to
       deal with historic problems, the PEAK's however do show
       many serious bugs as I have set out in Section 3 above."
           I would like to ask you what you mean by "the
       purpose of KELs are to inform support personnel how to
       deal with historic problems".  Could you explain
       precisely what you mean by that?
   A.  Yes.  So the purpose of a KEL is that it is to provide
       knowledge for people who -- for SSC support people who
       might be searching for things.  So if a problem arises
       they would search the KELs and either they will find
       a KEL that appears to be appropriate, and it shows that
       what has happened before has happened again and there
       might be instructions on how to deal with it.
       Alternatively if they don't find a KEL, they go through
       the process of creating a new one.
   Q.  One possible implication that might be drawn from
       paragraph 5.186 is that if there is a bug which has
       an effect on branch accounts you will often not find
       a KEL that addresses it.  But I would like to give you
       an opportunity to clarify whether that is what you are
       trying to imply or not.  I may be reading too much into
       it.
   A.  Because a KEL is a single source of a record of a bug,
       error or defect, what you will find is if the fault has
       occurred again they will refer back to the KEL and they
       will see that that fits their scenario.  They often will
       not put the details of this particular scenario, this
       particular bug, error and defect, back in the KEL.  The
       KEL is just the single source of knowledge for them to
       say, ah, it is that problem.
   Q.  I see.  So what you are saying is that if you get a bug,
       you generally speaking -- and of course there are
       always -- I'm not suggesting to you that anything is
       comprehensive -- but you are saying that generally
       speaking if you get a bug of that sort there will be --
       once it is detected, there will be a KEL which addresses
       it, yes?
   A.  Yes.
   Q.  But that KEL will generally address the first instance
       in which it arose?
   A.  Yes.
   Q.  And when there are other instances in which it arose the
       KEL won't necessarily address those?
   A.  Won't necessarily.  Sometimes you see it updated if they
       have got a slightly new manifestation of it, or there is
       some additional knowledge they have gained and they will
       add it to the KEL so that the next time it arises it
       might be helpful to the person, but there will only be
       one KEL.
   Q.  I understand.
   A.  There might be ten instances of that triggering and they
       will be in the PEAKs.
   Q.  That's very helpful.  So is this right, if there is
       a bug of that sort that's detected you are likely --
       there's likely to be a KEL which deals with it?
   A.  Yes.
   Q.  Which describes it, put it that way?
   A.  Mm.
   Q.  And if there are other manifestations of that bug that
       occur in a similar way, the KEL may well not refer to
       them?
   A.  Yes.
   Q.  But if the bug manifests itself in a slightly different
       way, in an odd way, then the KEL will be generally
       speaking -- there are always exceptions I am sure, it is
       a human system, but generally speaking there will be
       an amendment to the KEL to explain the new variance, to
       explain the new phenomenon, to enable the users to have
       a proper understanding of what they need to look for?
   A.  That is right.  Someone within SSC will decide whether
       to add to the KEL that's already there, or that it is
       significantly different so they will create a new KEL
       for it.
   Q.  I think we can agree that in some cases KELs give quite
       a lot of information and can be very useful?
   A.  Yes.
   Q.  In other cases one needs to look at PEAKs to have
       a proper understanding of some details that may be
       relevant to the inquiry that we are undertaking now?
   A.  Yes.
   Q.  So it depends?
   A.  It does depend.  Sometimes a KEL will say: please record
       every occurrence of this that you see in this KEL, other
       ones you don't have that information.
   Q.  That's very kind, Mr Coyne.
           My Lord, I wonder -- this is a convenient moment,
       I wonder whether we can break now and perhaps sit at
       1.50 pm.  Would that be acceptable to your Lordship?
   MR JUSTICE FRASER:  Yes.  Do I detect an undercurrent of
       concern about time on your part?
   MR DE GARR ROBINSON:  I always have an undercurrent,
       my Lord.
   MR JUSTICE FRASER:  We will come back at 1.50 pm.  Usually
       it will be a minimum for an hour.  It is not for me, it
       is not for you, it is for the witness.  But today,
       because it is the first time it has arisen and you asked
       so politely, we will come back at 1.50 pm.
   MR DE GARR ROBINSON:  I'm grateful, my Lord.
   MR JUSTICE FRASER:  Mr Coyne, usual arrangements.
   A.  Yes, my Lord.
   MR JUSTICE FRASER:  If you could come back for 1.50 pm
       I would be very grateful.
           Just one housekeeping point.  My file of PEAKs and
       KELs that were being referred to, obviously we can start
       a new one now for the experts, but I just wouldn't like
       that to be forgotten about.
           So 1.50 pm.
   (12.55 pm)
                     (The short adjournment)
   (1.50 pm)
   MR DE GARR ROBINSON:  My Lord, good afternoon.
           Good afternoon, Mr Coyne.
   A.  Afternoon.
   Q.  Let's see if we can agree some things.  First of all,
       can we agree that the parties aren't here spending all
       this time and money just to find out if Horizon could
       have been improved, yes?
   A.  Yes.
   Q.  That might be important to understanding the background
       to a particular claim by a particular claimant, but if
       no bugs in Horizon caused any non-transient losses to
       any claimants, we might as well go home, yes?
   A.  Yes.
   Q.  In those circumstances, would you agree with me that it
       is useful to know, in fact it is necessary to know, in
       this trial, about the likely impact of bugs that have
       occurred in Horizon and whether that impact is likely to
       have been transient or lasting?
   A.  Yes.
   Q.  That's what the Horizon Issues we discussed yesterday
       are all about, isn't it?  The extent to which it is
       likely or unlikely for bugs to cause shortfalls for
       which subpostmasters have been held liable?
   A.  Yes, I think the term is the extent to which it is
       possible or likely, yes.
   Q.  And for that issue to be meaningful don't we have to
       settle upon some metric for the likelihood of a bug
       causing a lasting shortfall of that sort?
   A.  The metric I have adopted is to find whether there was
       an actual bug, error or defect and see whether it had
       an impact.  I don't believe there is any other metric.
   Q.  Don't you need a yardstick to have a proper measure of
       extent?  For example, the likely impact of bugs in
       Horizon across all Post Office branches, the likely
       impact in a single branch in a single month, the
       likelihood impact across all claimant branches while
       they held those branches?  Wouldn't those be useful
       yardsticks for the purposes of deciding extent?
   MR GREEN:  My Lord, I'm hesitant to rise, but we
       specifically asked whether they were going to try to get
       the third report in by the back door.
   MR DE GARR ROBINSON:  I have no intention of asking any
       questions about that.
   MR GREEN:  That's one of the three questions that has just
       been asked.
   MR JUSTICE FRASER:  Mr de Garr Robinson, your last question
       is really a question for me, isn't it?
   MR DE GARR ROBINSON:  My Lord, it is a question about what
       this expert witness has done in approaching the --
   MR JUSTICE FRASER:  If you do it by reference to his witness
       evidence rather than by reference to submissions that
       really amount to arguing the case in front of me, that
       would probably be more useful.
   MR DE GARR ROBINSON:  My Lord, if I may, I would like to
       follow my own course with my cross-examination of this
       witness.
   MR JUSTICE FRASER:  You can, but your last question to that
       witness is an issue for me.
   MR DE GARR ROBINSON:  Now, Mr Coyne, can we also agree that
       there is no realistic prospect of you or anyone else
       examining, still less assimilating, every document
       that's been disclosed in this case in relation to
       Horizon and its operation over the last 20 years?
   A.  Yes.
   Q.  That left you and it left Dr Worden with a choice,
       didn't it?  You can make observations on the documents
       you found which may not advance matters very far: here
       are some bugs which caused shortfalls in some branch
       accounts, so it follows that it is not just possible
       that a bug has caused loss, it is a certainty.  That
       answers the question: is it possible or likely the bugs
       have a potential to cause shortfalls, but it is not
       an answer to the complete question, is it, because as we
       have seen the Horizon Issues include questions of
       extent?
   A.  Yes.
   Q.  To what extent is this likely or unlikely to happen in
       branch accounts in Horizon?  To what extent was the risk
       faced by a user in Horizon high or low?  Do you accept
       that those are the sort of issues that are raised in
       this trial?
   A.  Yes.
   Q.  And to address extent, you can look at a limited portion
       of the evidence that you can sensibly review.  You can
       assess its nature and scale and on the basis of those
       assessments you can arrive at overall conclusions that
       are generally useful, can't you?
   A.  Yes.
   Q.  An analogy with which we are familiar is an exit poll
       that is taken on the day of an election.  Only a very
       small number of people are actually asked how they
       voted, but based on that sample useful estimates can be
       made about how all the voters voted, do you agree?
   A.  Yes.
   Q.  So you can move by a process of sampling from a position
       of relatively uninformative certainty, you know, knowing
       with certainty how 5% of people have voted, to
       a position of much greater certainty or much greater
       interest, namely what the actual outcome of the election
       is going to be, yes?
   A.  I believe it would be an indicator, yes.  Yes.
   Q.  In order to be useful, do you agree that the sample you
       must choose needs to be an unbiased sample?
   A.  Yes, in that scenario it would need to be an unbiased
       sample, yes.
   Q.  So if you are trying to work out, for example, how
       voters have voted in an election it is no use just
       asking people coming out of voting booths wearing blue
       rosettes, is it?
   A.  No, that is true.  If you are going down the route of
       using sampling then you would have to make sure that
       that sample is unbiased.
   Q.  Once you have an unbiased sample it then becomes
       possible, doesn't it, to scale up.  So you can scale up
       the results that you have got from your unbiased sample
       and invite people, invite the court, to make judgments
       about what the overall likelihood of something happening
       or not happening is, do you agree?
   A.  Certainly in your election scenario that would scale up
       with a reasonable tolerance, yes.
   Q.  And that's what Dr Worden has done in his first report,
       isn't it?
   A.  Yes.
   Q.  And it is the sort of question that arises in a case of
       this sort where there are so many thousands of KELs and
       over 200,000 PEAKs and tens of thousands of OCPs, OCRs
       and MSCs, yes?
   A.  Yes, there is an inherent danger with that in that with
       such a large sample size, and in the likelihood that it
       is a small fraction of the entirety of the transactions
       or branches that have suffered failure, that you might
       sample and not find any.
   Q.  Yes, and there are well known statistical techniques for
       dealing with that problem, aren't there?  So if you
       take -- how is it -- ten samples.  The number of the
       samples you take will then determine how representative
       the results are that you are likely to get from
       attempting to scale up.  Are you aware of how this
       works?
   A.  Not really with what you have put to me there.  Would
       you explain that a little bit further?
   Q.  Let me see if I can put -- you will have to give me
       a moment, I'm afraid.  Would you give me one moment?
   A.  Certainly.
   MR GREEN:  My Lord, I know my learned friend is worried
       about time.  Mr Coyne has disavowed statistical
       expertise in his report.
   MR JUSTICE FRASER:  If Mr de Garr Robinson wants to use his
       time exploring basic elements of statistical analysis
       I'm not going to stop him.
   MR DE GARR ROBINSON:  My learned friend actually is quite
       right.  Mr Coyne has disclaimed any ability or expertise
       in this, so let me move on.
   A.  Certainly I have very little expertise.  I have a broad
       ability to understand the concept.
   Q.  So you have no expertise in what to do with samples and
       how to assess whether the sample can be
       effectively scaled up or not, is that right?
   A.  Yes, I can see how that could be effective in certain
       scenarios, but I don't believe the scenario that the
       application here will work.
   Q.  Well, isn't it actually what you are trying to do in
       your evidence as well, Mr Coyne?  Aren't you saying:
       I found a number of bugs, and aren't you suggesting that
       an inference should be drawn that there could be a great
       number of other bugs that you haven't found yet?
   A.  Yes, but it is from the basis of actually finding bugs
       and trying to identify how many branches may be impacted
       by those bugs, errors and defects.
   Q.  What I'm suggesting to you is that when you find certain
       hits in your sample, because any sample is necessarily
       limited, your ability to be able to say that the court
       should scale up and should infer that there are likely
       to be a certain number of other hits, or an uncertain --
       a certain scale of other hits, that is dependent upon
       the quality of the sample that you have chosen in
       a particular -- whether it is an unbiased sample, yes?
   A.  Yes, but in my report I haven't suggested any scaling up
       from particular bugs, errors and defects.  I have talked
       about specific bugs, errors and defects and how many
       branches they are recorded to have impacted.  There's no
       scaling applied to that.
   Q.  Have you done that?  Have you -- you say that you found,
       or I should say you say that there have been found 29
       bugs, yes?
   A.  Mm.
   Q.  Have you given any assessment of how many branches were
       affected by those bugs?
   A.  Yes, Dr Worden and I in I think it is the second joint
       statement, the longest joint statement, that has the
       bugs in, we have a column in there that attempts to
       identify the number of branches that were impacted.  And
       certainly some of the source documentation will indicate
       a number, it might not be the right number, but it does
       indicate whether it is 28 or 30 or 32.
   Q.  What I would like to ask you to do, please, is to look
       at your original second report, not your revised
       version.  This is {D2/4/43}.
   A.  I don't have a paper copy of that.
   MR JUSTICE FRASER:  You haven't?
   A.  Not a paper copy.  I will have the second.
   MR DE GARR ROBINSON:  I'm only going to take you to one
       paragraph.  It is 3.105.  This is the original version.
       You have changed it in the revised version.
   A.  Mm.
   Q.  You say:
           "The PEAKs analysed below are a small portion of the
       PEAKs I have identified as causing financial discrepancy
       in branch accounts outside of those bugs acknowledged by
       Post Office.  It should be noted there are potentially
       thousands more PEAKs that illustrate financial
       discrepancy arising in branch accounts, this is only a
       small selected sample from keyword searched PEAKs."
   A.  Yes.
   Q.  Now let's take this in stages.  You have changed the
       wording of the first sentence and I will go to that
       change but I want to ask you about what the original
       version means first.  What you are claiming there is
       that you have identified a large number of PEAKs
       recording bugs which cause branch shortfalls but you
       have only mentioned a small portion of them in your
       report.
   A.  Yes.
   Q.  That wasn't true, was it?
   A.  No.  By the conclusion of this report there was
       a substantial amount that was looked at.
   Q.  In fact what you had done is you had identified every
       single bug that you could and included it in this
       report, hadn't you?
   A.  Within the time available.  I mean it is probably quite
       possible that there could well be more but we have
       certainly had a good search.
   Q.  So what you say in the first sentence wasn't true, was
       it?
   A.  No, my concern with the way that it read is that --
       I was saying that I had only analysed a small portion of
       the ones that had caused financial discrepancy.
   Q.  What you are saying there -- what the words literally
       mean, Mr Coyne, is that you've analysed below a small
       portion of a larger group of PEAKs that you have
       identified as causing financial discrepancy?
   A.  Yes.
   Q.  And in fact that wasn't the case.  What you did in your
       report was you included every PEAK you could, every PEAK
       that you were aware of as causing financial discrepancy,
       didn't you?
   A.  Yes.
   Q.  So this is an important paragraph.  You will see that it
       is immediately under the heading "Horizon Issue 1
       PEAKs", so it is an introductory paragraph, it is
       introducing the reader to what comes next.  It is not as
       if you wouldn't have paid attention to what was in this
       paragraph.  I would just like to know what possessed you
       to make that extraordinary claim that wasn't true?
   A.  Well, all the PEAKs hadn't been analysed, but of the
       portion that had been analysed there was a number that
       identified financial discrepancy.
   Q.  Weren't you seeking to give an impression in that
       sentence that wasn't accurate?
   A.  No, not at all, but I did see that it wasn't as clearly
       worded as it should have been.
   Q.  And do you accept that this was corrected only after my
       instructing solicitors wrote to Freeths on 1st February
       asking for these other PEAKs to be identified?
   A.  Yes, clarification was requested.  Yes.
   Q.  And shall we look at Freeths' response.  It is at
       {C5/36/1}.  This is Freeths' response to my instructing
       solicitor's letter.  If we could go to page {C5/36/2},
       please.  Picking it up at the bottom, paragraph 3, it
       sets out the sentence that we are talking about.
       Post Office's request is:
           "Please identify the full set of PEAKs that Mr Coyne
       has identified as causing financial discrepancy in
       branch accounts outside of those bugs acknowledged by
       Post Office."
           The response comes:
           "See response 2.2 above, see the 'yes' entries in
       the column ..."
           Then if we can go over the page {C5/36/3}:
           "Mr Coyne has considered this paragraph in his
       Supplemental Report and notes that his opinion has not
       been articulated correctly in the sentence extracted in
       your client's Request ..."
           I'm sorry, my Lord?
   MR JUSTICE FRASER:  My screen has crashed yet again but I'm
       going to deal with it.  I still have the common screen
       and I still have LiveNote, that's good enough.
   MR DE GARR ROBINSON:  "As is clear from the sentence which
       follows it, Mr Coyne intended to refer to the fact that
       he has only reviewed a small proportion of the total
       PEAK documents disclosed.  As such, Mr Coyne now
       clarifies the meaning of this sentence as being: 'I have
       analysed a small proportion of the PEAKS, from that
       analysis, I have identified the following as causing
       financial discrepancies in branch accounts outside of
       those bugs acknowledged by Post Office'."
   A.  Yes.
   Q.  So was that your intention?  Was that what you had
       intended to say at paragraph 3.105 and by clumsy
       drafting you had said something very different?
   A.  Yes, what I meant to say was in the second draft.
   Q.  Well, let's go to the second draft.  That's at
       {D2/4.1/45}.
           I do believe I have the wrong page, I do apologise.
   MR JUSTICE FRASER:  No, you haven't.  It is {D2/4.1/45}, it
       is indeed page 45 but it is in the next version.
   MR DE GARR ROBINSON:  {D2/4.1/45}, please.  Here you say:
           "I have analysed a small proportion of the PEAKs,
       from that analysis I have identified the following as
       causing financial discrepancies in branch accounts ...
       It should be noted there are potentially thousands more
       PEAKs that illustrate financial discrepancy arising in
       branch accounts, this is only a small selected sample
       from [the keyword searches] ..."
           So knowing that it was being said that you have not
       found anything like the PEAKs that you would need to
       find even to begin to justify the claimants' claim,
       because that had been said by Dr Worden in his first
       report, hadn't it, you are now arguing in
       paragraph 3.105 that this is only a small sample from
       a large cohort and the small sample of bugs, the small
       number of bugs that you found should be scaled up to
       reflect the large size of the cohort, yes?
   A.  I'm saying that there is the potential.
   Q.  Could I suggest to you, Mr Coyne, that there's no basis
       for taking the fact that you found a small number of
       bugs in the PEAKs that you reviewed and inferring there
       could be any number of bugs in the PEAKs that you
       haven't reviewed because scaling up only works if you
       have taken an unbiased sample, I think you have already
       agreed that, yes?
   A.  Yes, and that's why I'm not giving a view on what the
       scaling might be.  I'm saying that the potential exists.
   Q.  So you accept that you can scale up from, say, 200 PEAKs
       or KELs to a cohort of 220,000 but only if the sample
       that you have looked at is chosen at random.  Would you
       agree with that?
   A.  No, I don't agree with that, because this scaling is
       starting from the basis that there was actual bugs,
       errors and defects that did cause financial
       discrepancies.  So you are scaling up from a positive
       position that there are records in there that do show
       that and there's the potential for other ones to show
       that.
   Q.  So what you are suggesting is that having found 29 bugs,
       it is possible to scale up, it is possible to infer that
       there are likely to be thousands more bugs in existence,
       is that what you are saying?
   A.  There's certainly the potential to be.
   Q.  There is the potential.  Because what interests me is
       this concept of scaling up.  The bugs that you found you
       didn't find by reviewing a randomly selected sample, did
       you?
   A.  No.
   Q.  You did the equivalent of asking everyone coming out of
       the polling station who was wearing a blue rosette,
       didn't you?  Because what did you was you were looking
       for bugs that had that effect.  You were positively
       searching for and excluding all others.  You were
       searching for bugs that had particular characteristics?
   A.  It doesn't work when you try and relate it back to the
       election scenario where the purpose of that is to try
       and establish what the percentage of people voting would
       be.  In this scenario here we are, as I understand it,
       trying to identify real bugs, errors and defects, so
       that is what is -- what was done during this exercise.
       And what I'm saying here is because I have only analysed
       a small number of the totality of the PEAKs, then there
       is the potential for many more to be in here.
   Q.  Could I suggest that the process that you have
       undertaken -- first of all, perhaps we can agree this.
       Do you agree that the sample, the small sample that you
       describe, the small selected sample that you describe in
       3.105, that isn't an unbiased sample, it isn't
       an unbiased sample from which it is possible to scale up
       anything?
   A.  It is certainly not an unbiased sample, that is correct.
   Q.  From which it is possible to scale up anything?
   A.  You wouldn't want to scale up from that and make any
       numerical assumptions based on the -- however many,
       I think it was less than 10,000, but however many there
       is, that that should be the number multiplied by that
       percentage of the total number of PEAKs, you wouldn't
       want to make that assumption.
   Q.  What I would like to understand, though, is why you
       chose to say there are potentially thousands more PEAKs.
       Why didn't you say dozens more or hundreds more?  Why
       did you choose to say thousands more PEAKs?  It seems to
       me, Mr Coyne, as if you are inviting the court to infer
       that because you found 29 bugs from a small sample, it
       is appropriate to think there could well be thousands
       more out there that you haven't found, and I would like
       to ask you why you think it is appropriate for the court
       to think that?
   A.  I don't believe there is any more appropriate number to
       put in there.  Nothing would provide any precision in
       there as to what the number could or should be.
   Q.  Are you suggesting that it is likely that there are
       thousands more bugs of this sort?
   A.  No.
   Q.  In fact are you suggesting that -- well, are you
       suggesting that it is more likely that there are
       thousands than there are hundreds or dozens, for
       example?
   A.  Yes, I don't -- I would think there would be more than
       dozens certainly.
   Q.  I'm interested that you should say that.  So you are
       suggesting -- you are now making a claim that having
       found 29 you think there will be dozens more.  Is that
       right?
   A.  I certainly believe there would be dozens more, yes.
   Q.  And are you prepared -- but you are not claiming that
       there could be more than -- that there are likely to be
       more than dozens, that is not a claim you are making?
   A.  I don't know what the number would be but there
       certainly is the potential for there to be thousands
       more.
   Q.  By "potential", you are saying it is not impossible, are
       you?
   A.  Sorry, say again?
   Q.  By "potential" you are saying -- sometimes when people
       say "potential" they mean it is a really viable
       possibility and close to a likelihood that something is
       going to happen, other times they simply mean it is
       simply not impossible.  Potentially I could become Prime
       Minister.  It is possible although it isn't going to
       happen.  Other times I could say potentially I could
       retire when I'm 65, and that would be -- it's not
       certain but it is a real -- now, when you say
       "potential" you mean not impossible, don't you?
   A.  It is certainly not impossible and it is on the basis
       that there are 200,000 PEAKs, whatever the exact number
       might be, that haven't been reviewed.  They weren't
       responsive to the search terms that were selected at the
       time.
   Q.  But here's what I don't understand, Mr Coyne.  As you
       explained very helpfully yesterday morning, you and your
       team have spent a great deal of time using the
       intelligent search functions that you have at your
       disposal, which I rather envy, I have to say, in order
       to find precisely these kinds of PEAKs and KELs?
   A.  Yes.
   Q.  And you are quite good at that.  You are certainly
       better than I would be.  It is the entire raison d'etre
       of your e-disclosure business, isn't it, that you can
       use what is in the trade called intelligent search
       functions?  In those circumstances isn't the fact that
       you have only been able to find 20 bugs significant?
   A.  No, I don't believe so.  Typically we would be and
       I would be assisted by somebody within the organisation
       who created the documents, that would be helpful with
       things like particular error codes that might indicate
       financial discrepancy, certain terminology in code and
       things like that.  None of those were provided.  So
       I was starting from the basis of just trying to come up
       with words and phrases that may be used, that may
       indicate a discrepancy, that may indicate a defect,
       an imbalance, but it is entirely possible that there are
       words and phrases used that I'm not aware of.
   Q.  But I presume that by the phrase "intelligent search
       functions" you include concepts such as once you have
       started finding some and you see how they draft
       themselves, you realise there are word patterns or
       symbol patterns or report patterns, TPSC256, TPSC268A,
       all those automatic report numbers which indicate
       receipts and payments mismatches, or CRC failures, or
       all the other issues that could be indicative of a bug.
       That as you go along and as your knowledge of the PEAKs
       and the KELs increases, you become better and better at
       finding terms and symbols which will enable you to zero
       in on the PEAKs and KELs that you are actually looking
       for.  Isn't that how it works?
   A.  That is how it works.  One of the documents that
       I requested very early on was a document that would
       indicate where a financial impact has been discovered.
       That was one of the very first documents.  And that
       simply wasn't provided, that request was --
   Q.  I'm sorry, why are you making that point?
   A.  Because that would have assisted with the searching that
       I was doing at the time.
   Q.  Mr Coyne, what document are you talking -- what document
       did you request?
   A.  There was an RFI request that I put in, I think it was
       in June or July --
   Q.  Right, and this RFI request was a request which was
       designed to assist your search functions, was it?
   A.  Yes.
   Q.  And what did the request seek?
   A.  We have probably got the document, but it asked for
       documents that were returned that would indicate that
       a financial impact has been discovered.
   Q.  I have to say, it may be my fault, but I'm scratching my
       head metaphorically, I don't recall that particular
       request, but let's not waste any time on it.
           Let me ask you a different question.  My suggestion
       to you is the fact that you have only been able to find
       29 bugs, it is significant, because with or without the
       document you just referred to you have found bugs which
       you think do disclose a lasting shortfall for which
       subpostmasters were made liable, is that right?
   A.  Yes.
   Q.  So you have seen how PEAKs and KELs relating to those
       bugs are worded, yes?
   A.  Yes.
   Q.  So it is not as if you have been prevented from making
       choices about the form of words and symbols and so on,
       not been prevented from alighting upon the kind of
       searches you need to do in order to find the bugs you
       are looking for, yes?
   A.  The process would have been far easier if the documents
       that I had asked for would have been provided at the
       time.
   Q.  Okay.  Let's not talk about the process, let's talk
       about what you have actually done.
   A.  Yes.
   Q.  Imagine for the sake of argument that your team's
       intelligent service system were absolutely brilliant and
       that it were perfectly qualified, having become really
       familiar with the system and with all the reports that
       are generated in the system and all the phrases that
       tend to be used and so on, they were absolutely
       brilliant at capturing evidence of bugs of the sort you
       were looking for.  Now if that were the scenario, the 29
       bugs that have been found could be very close to the
       absolute total number of bugs that are to be found in
       the cohort of documents, couldn't they?
   A.  It is possible that that's the case.  I think it is
       highly unlikely but it is possible that that's the case.
   Q.  Then let's assume that you weren't quite that good.  It
       might be that you only got half of them or perhaps
       a third of them.
   A.  Yes.
   Q.  But doesn't, with all the intelligent search functions
       that you have at your disposal and the people you have
       to help you, doesn't the fact that only 29 have been
       found strongly suggest that there aren't thousands more
       bugs lurking in the PEAKs that you have been unable to
       uncover?
   A.  It is certainly possible that the number is a lot less
       than that, yes.
   Q.  I would suggest to you, Mr Coyne, that it is obvious
       that if you have only been able to find 29 PEAKs there's
       no way in the world that there are 100 times that number
       of PEAKs out there that you have been unable to find?
   A.  There are hundreds of thousands of PEAKs and the impact
       is not always described in the PEAK.
   Q.  Could you just answer my question before we can move on.
       You have segued into the question of impact and I am
       going to come to that.
   A.  Sorry, forgive me.
   Q.  But just focus on my question for a moment.  Isn't it
       obvious that however difficult the process has been,
       that your team has not been so clumsy and incompetent
       that it has only managed with all its intelligent
       searches to find 1 in 100 of the bugs that are actually
       disclosed in the PEAKs and KELs that you have reviewed?
   A.  No, I do not think that's inconceivable.
   Q.  Can you say that again?
   A.  I don't think it is inconceivable that that might be the
       case.
   Q.  You don't think it is likely, though, do you?
   A.  It would be a complete guess.
   Q.  Yes.  Here's what interests me, Mr Coyne.  Before we
       broke for lunch I asked you some questions about the KEL
       system and the PEAK system, and I'm afraid I don't have
       the transcript in front of me, but I think you helpfully
       agreed that if a bug was detected which had branch
       impact it would -- the chances were, I think I may be
       putting it slightly too low, that it would be likely to
       be addressed in a KEL somewhere, yes?
   A.  A PEAK somewhere.
   Q.  A KEL.  We were talking about KELs at that time.
   A.  Sorry, could you put your question again then.
       I thought you --
   Q.  Perhaps I should look at the transcript.  Could we go
       back to just before we broke.  Would you give me
       a moment, please.
   A.  Certainly.
   Q.  If we could go to page {Day15/94:21} of today's
       transcript, please.
   MR JUSTICE FRASER:  Of which page, sorry?
   MR DE GARR ROBINSON:  94, my Lord.
   MR JUSTICE FRASER:  94, line 21.
   MR DE GARR ROBINSON:  I ask a question.  I say:
           "I see.  So what you are saying is that if you get
       a bug, you generally speaking -- and of course there are
       always -- I'm not suggesting to you that anything is
       comprehensive -- but you are saying that generally
       speaking if you get a bug of that sort there will be --
       once it is detected, there will be a KEL which addresses
       it, yes?
   A.  Yes.
   Q.  And you said "Yes".
   A.  Yes.
   Q.  You are not withdrawing that evidence, are you?
   A.  No, no.
   Q.  So we have this situation.  We have if there is a bug
       that has been detected, the chances are it is going to
       be described in a KEL somewhere, yes?
   A.  Yes, if it has been detected, yes.
   Q.  And there aren't 220 KELs, are there, there are
       something like 9 and a bit thousand, is that right?
   A.  Yes.
   Q.  And we know that even by the time of your first
       report --
   MR JUSTICE FRASER:  Just give me literally 30 seconds.
   MR DE GARR ROBINSON:  Of course, my Lord.
           (Pause)
   MR JUSTICE FRASER:  It is all right.  Go ahead,
       Mr de Garr Robinson.
   MR DE GARR ROBINSON:  At the time of your first report we
       know you had already reviewed yourself 5,114 KELs, yes?
   A.  Mm.
   Q.  And members of your team had, would I be right in
       thinking, I may have specifically asked you this, and if
       so I'm sorry to be going over it, but members of your
       team would have reviewed other KELs in addition to that?
   A.  Yes.
   Q.  And since your first report you would have looked at
       further KELs on top of the --
   A.  Yes.
   Q.  And I think I asked you how many you had reviewed and
       you said you were unable to tell me.  That's my
       recollection but forgive me if I'm wrong.
   A.  No, I think that is right.
   Q.  Right.  So you have looked at more than 5,114 KELs.
       Would I be right to infer you probably looked at more
       than 6,000?
   A.  No, it probably is between 5 and 6,000.
   Q.  And your team had looked at others as well.  Are you in
       a position to give any kind of assessment of how many
       other KELs they are likely to have looked at?
   A.  It will likely be probably 1,000 more or something like
       that.
   Q.  Okay.  So you and your team collectively have looked at
       something like between 6 and 7,000 KELs out of a total
       of, what, 9,500, that kind of figure?  Yes?
   A.  Mm.
   Q.  And yet all that has been found is 29 bugs, correct?
   A.  Yes.
   Q.  Isn't that a significant indication, Mr Coyne?  Doesn't
       it suggest quite strongly that the chances are that the
       total number of bugs that are out there are not going to
       be in the thousands.  In fact there are unlikely to be
       more than perhaps twice as many in the entire cohort of
       KELs, would you agree?
   A.  That have an impact on branch accounts, yes.
   Q.  Yes.  And if one then were to assume that some error
       factor, let's assume there are some PEAKs that don't
       make it through to KELs, that's not going to account for
       very many, is it, because most of the time you would
       expect the PEAK to have resulted in a KEL as soon as the
       problem was discovered?
   A.  Yes.
   Q.  So actually, just ignoring the PEAKs for a moment, in
       that you have reviewed, you and your team, over 6,000
       KELs, so more than two-thirds of the available KELs --
   A.  Yes.
   Q.  -- and you found, between you and Dr Worden, 29 bugs --
   A.  Yes.
   Q.  -- wouldn't it be fair to infer that the total number of
       bugs to be found in those KELs is likely to be less than
       40?
   A.  Yes, that sounds reasonable.  I'm just concerned that
       the confusion here is between KELs, which document
       a single instance of a bug, and a PEAK which is what we
       are referring to here, which shows how many times that
       bug has had an impact on the Horizon system.
   Q.  So let's take this in stages.  Two questions raised.
       The first one is how many bugs?  The second one is how
       many impacts and what's the nature of their impacts?
   A.  Yes.
   Q.  Both questions are important building blocks in order to
       arrive at an assessment of extent, yes?
   A.  Yes.
   Q.  Thank you.  I think you have agreed that the chances
       are, as a result simply of your examination of KELs,
       that there are likely to be no more than 40 bugs which
       have been found to have branch impact, yes?
   A.  Bugs, errors and defects, yes.
   Q.  So we move on to the next question which is what was the
       impact of these 40 bugs?  What you say is -- and I fully
       understand it -- that you can't get that from KELs, you
       get that from PEAKs.
   A.  Yes.
   Q.  But here's the interesting thing: because you have got
       the KEL, actually the KEL operates as a really useful
       way of looking for PEAKs to which the KEL is relevant,
       doesn't it?
   A.  Yes, you will often see the link between the two.
   Q.  Quite often, as you fairly say, there is actually
       a specific link made between the PEAK and the KEL, and
       you can search for PEAKs and the PEAKs will -- this
       would be right, wouldn't it: PEAKs invariably refer to
       the KELs to which the problems they address are
       relevant, yes?
   A.  Yes.  There is a better quality of link from PEAKs to
       KELs.  There's not always that link KEL back to PEAK.
   Q.  But this is where the beauty of intelligent searching
       comes in, isn't it?  Because you can intelligently
       search through the body of PEAKs, having identified all
       the relevant KELs, and there may be a number of them,
       there may be one KEL and perhaps two or three others
       that are also relevant, you search for all the PEAKs
       which refer to those KELs, don't you?
   A.  Yes.
   Q.  And by that means you are going to get actually quite
       a good sense of -- you are going to get a good hit rate.
       You are going to find most, probably more than most, of
       the PEAKs considering problems which those KELs address,
       yes?
   A.  It is entirely possible to do that, yes.
   Q.  I'm not asking you whether it is possible, I'm
       suggesting that it is likely that if you undertake that
       search you will actually find all the PEAKs that are
       relevant -- that exist that are relevant to the problem
       addressed in the KEL?
   A.  Yes.
   Q.  So you have identified 29 bugs, you do your searches for
       all the PEAKs, and by that means you can identify all
       the PEAKs which record manifestations of the bug.  It's
       likely, I'm going to use the word "likely".  I'm not
       suggesting it can be entirely comprehensive, Mr Coyne.
       So can we take it as read that obviously there are going
       to be gaps at the margin, aren't there?
   A.  Yes.  A PEAK is typically created when the bug, error or
       defect gets to SSC within Fujitsu.  So there may be
       others that don't get there, but once they get to third
       line support the PEAK is created, so yes.
   Q.  So by this means you were in a good position both to
       identify bugs that have been detected in the system --
   A.  Yes.
   Q.  -- quite reliably, so with a fair degree of confidence
       that there won't be that many more bugs in the system?
   A.  As long as they hit the search terms that I have used,
       yes.
   Q.  Remember we are talking about the KELs now.  The
       starting point for the process that I have described is
       the KELs.
   A.  Yes.
   Q.  And you've physically reviewed those KELs, haven't you?
   A.  Yes.
   Q.  So it is not as if you need intelligent search terms to
       find the right ones, you have actually looked at them,
       haven't you?
   A.  Yes.
   Q.  So by physically looking at them you found however many
       bugs you found, I think it was -- it must be somewhere
       below 20 because of course of the 29 there were also
       bugs that were accepted by Post Office and there were
       the bugs that were found by Dr Worden as well.  So you
       found -- if I suggest to you that you found around in
       the late teens, would that figure sound about right to
       you?
   A.  I believe that the figure in the second report was 28,
       was it?
   Q.  28.  Well, that would include the bugs that have been
       found by -- that have been admitted --
   A.  Yes, it did --
   Q.  -- by Post Office, yes?
   A.  -- the three --
   Q.  And it would include bugs that have been identified by
       Dr Worden in his report?
   A.  Yes.
   Q.  Let's not quibble about numbers.  The point is by that
       process I think you have accepted that it is
       a relatively reliable process by which you will have
       successfully identified the large majority of the bugs
       that have been identified by the SSC during the
       operation of the PEAK and KEL system?
   A.  Yes.
   Q.  What's more, because having done that you can then look
       through the PEAK system, you have the ability to
       identify all the PEAKs which represent manifestations of
       those bugs?
   A.  Yes.
   Q.  And those PEAKs identify -- where there are branch
       effects those PEAKS generally will identify the branches
       affected, yes?
   A.  Generally but not always.
   Q.  Generally speaking, if a branch has a problem then the
       PEAK will identify the FAD code of the branch, correct?
   A.  No, that's not always the case.  They will typically say
       this might have an impact on branch accounts.  They will
       sometimes list a FAD code.  They will sometimes refer to
       a branch by name or location.
   Q.  I see.  Let me park that issue until another day.  But
       in any event by this means, by moving from the PEAKs to
       the KELs, you have quite a good method of identifying
       the number of instances in which branches have been
       affected by the relevant bug, yes?
   A.  Yes.
   Q.  Perhaps half an hour ago you gave me an answer which
       surprised me.  I wasn't expecting it.  You indicated
       that in the bug table in joint statement 2 there was
       a column which indicated what the extent of each of your
       bugs were, in other words the number of branches that
       were affected by each of your bugs.  Did I misunderstand
       your answer?
   A.  It is either in the joint statement 2 or it is in the
       report 2.
   Q.  You see, I'm struggling to understand what you are
       talking about.  It may be my fault.  But I'm not aware
       that anywhere -- that you have anywhere given any
       evidence as to how many branches you think were affected
       by any particular bug.  Have I misunderstood your
       evidence?
   A.  It certainly is in here.  There is an attempt to
       identify how many branches may have been impacted by
       that defect.
   Q.  It would be helpful if you could tell me because it
       might save some time tomorrow or the day after.  Should
       I be looking at the joint statement?
   A.  Yes, if you look at joint statement 2.  So the page
       where the actual table starts, index 1 on the left-hand
       side.  Then it says receipts and payments mismatch, the
       year, and then it says:
           "Coyne's opinion as to branch account impact ...
       identified approximately 60 branch accounts ..."
   Q.  That's because Post Office had identified 60 branches.
       But if we go --
   A.  If you go down for example to number 5, 14 examples of
       branch impacted.
   Q.  Oh, I see, and we have got 57 branches impacted.  For
       6.2 it's one branch impacted.
           That's very helpful, Mr Coyne, and by saying it is
       very helpful I'm of course conveying that I obviously
       haven't read this table properly.
           What's interesting about it, though, is that the
       numbers of branch impacts you are talking about are
       relatively low, aren't they?  I mean even 60 branches.
       So if we take the receipts and payments mismatch, 60
       branches affected.  60 branches out of 30 million
       monthly accounts, that represents a 1 in 5 million
       chance of a bug impact on a given set of accounts,
       doesn't it?
   A.  You have used two different variables there.  You have
       used branch accounts --
   Q.  Yes.
   A.  -- and you have applied that to branches.  If 60
       accounts are impacted should that not be compared
       against the total number of branches?
   Q.  Are you suggesting that when 60 branch accounts are --
       so what you have done here is you have indicated how
       many branches were impacted, but are you suggesting that
       lots of these branches were impacted on lots of
       occasions, is that --
   A.  No, I was responding to your -- you gave me an example
       of why it was low.
   Q.  Yes.  My suggestion to you, Mr Coyne, is that in
       circumstances where you have got a bug that affects 60
       branches?
   A.  Yes.
   Q.  And let's say it affects 60 branches on one occasion
       each, sometimes it will be more, but let's assume it is
       on one occasion each.  If you look at the entire corpus
       of monthly branch accounts, which is 3 million over
       20 years, then of the 3 million branches that will have
       been affected -- or that could have been affected,
       I should say, 60 will have been affected.
   A.  Yes.
   Q.  Which means it has a 1 in 5 million chance of hitting
       any given branch that you are looking at at any --
   A.  On any given month.  That may well be right, yes.
   Q.  Okay.  Thank you.
   MR JUSTICE FRASER:  But that's where his evidence was to
       where he had identified the numbers.
   MR DE GARR ROBINSON:  My Lord, I see that.
           So moving back to my line of questioning, I was
       asking you whether finding only 29 bugs strongly
       suggests that there aren't thousands more bugs in the
       system in the way that you seem to suggest at paragraph
       3.105?
   A.  No, but that's referring to "potentially thousands more
       PEAKs".  PEAKs are occurrences of bugs.
   Q.  I see.  But I thought -- I may have misunderstood your
       evidence then, Mr Coyne, because I thought you just told
       me that once you had a KEL identifying a particular bug
       it is relatively easy to find all the PEAKs to which
       that KEL is relevant, yes?
   A.  Yes.
   Q.  And I think you established with me that you had done
       that and that the result of that process is the column
       in joint statement 2 that we have just been looking at?
   A.  Yes.
   Q.  Could I ask you, Mr Coyne, if you were to add up all the
       PEAKs that are referred to in that second column in
       joint statement 2, will it demonstrate that there are
       thousands more PEAKs that are relevant?
   A.  I would have to go through and add them up.  It is not
       something that I have done at this stage.
   Q.  You are making a claim here, Mr Coyne, that there could
       be -- you don't just say -- yes, "potentially thousands
       more PEAKs".  But even when you wrote that sentence you
       had it in your power to actually work out how many more
       PEAKs we are talking about, didn't you?
   A.  No, I have worked out the number of PEAKs that relate to
       the KELs and the numbers will be in here.  But there
       could well be many more PEAKs that don't have the
       references to the KELs.
   Q.  Well, aren't you contradicting some evidence you gave to
       me a few minutes ago, Mr Coyne, in order to preserve
       your position on that second sentence?  Because before
       I think you told me that once you identify a KEL it is
       actually quite easy to then find all the PEAKs to which
       the KEL is relevant.
   A.  Yes, that is right.
   Q.  Right.  So the simple fact is that in order to
       demonstrate the truth of the statement:
           "... there are potentially thousands more PEAKs that
       illustrate financial discrepancy ... in branch accounts
       ..."
           All I have to do is go to column 2 of joint
       statement 2 and add up all the numbers.  Do you think
       there is a remote possibility of all those numbers
       coming to thousands, plural?
   A.  No, I don't believe -- I'm just looking through now to
       see if there's any that has an impact on ... It is
       possible that there is a PEAK that has an impact on, for
       example, 100 branches but I don't know whether that's
       the case or not, we would have to look through.
   MR JUSTICE FRASER:  One says 88, I think.
           Mr de Garr Robinson, do you want the witness to do
       a more detailed arithmetical analysis of that column, is
       that what you are asking him to do?
   MR DE GARR ROBINSON:  My Lord, I'm asking the witness to
       accept that in actual fact, having done all the work
       that he has done, and having produced the results which
       are now recorded in the second column which I freely
       admit I haven't fully understood and I'm grateful for
       Mr Coyne's clarification, it doesn't get anywhere near
       thousands more PEAKs.
   MR JUSTICE FRASER:  So, Mr Coyne, that was a question to
       you.
   A.  I'm content with the answer that I have given that there
       is the potential for there to be up to 1,000 PEAKs
       because you don't -- you only need to find a few more
       KELs that have impacts such as the bug, error or defect
       that impacted 88 branches to get to 1,000.
   Q.  Here's what's interesting, Mr Coyne.  What we have is
       what might be called an example of evasion.  The claim
       you make in 3.105 is that there are potentially
       thousands, plural, more PEAKs, and do you see what you
       did with your answer?  You went down to up to 1,000 in
       order to maintain your position.
           What I'm suggesting to you is on the very process
       that you have described to this court, on oath, and your
       explanation of the documents and how they work and the
       kind of things that they show and the things that they
       are likely to contain, it stands to reason as night
       follows day that there are not thousands more PEAKs in
       the PEAK corpus that illustrate financial discrepancy
       arising in branch accounts, would you accept that?
   A.  I don't accept that.  I don't accept that that position
       that there are potentially thousands more is incorrect.
   Q.  Well, I'm not sure I can go any further with this at
       this point, Mr Coyne.
           Let's analyse other examples where you use words
       suggesting that things happened on a large scale where
       in fact that might not be the case.  Can we go to your
       first statement {D2/1/38}, please.  At paragraph 3.18 --
   A.  Sorry?
   Q.  Of your first report.
   MR JUSTICE FRASER:  28.  I think you said 38.
   MR DE GARR ROBINSON:  I'm so sorry.
   MR JUSTICE FRASER:  Let's go to 28.
   MR DE GARR ROBINSON:  {D2/1/28}.  At 3.18:
           "Many ... (KELs) identify that not all errors were
       understood even by Fujitsu.  In the circumstances, it is
       highly unlikely that a Subpostmaster could interpret or
       identify the causes of any bugs/errors or defects when
       Fujitsu themselves often did not understand the cause of
       such or their full effects."
           So you have got another use of the word "often", do
       you see that?
   A.  Yes.
   Q.  And no indication is given of the scale of the word
       "many" or the scale of the word "often", what kind of
       numbers you are discussing.  Would you agree with me,
       Mr Coyne, that 10, for example, is hardly material given
       the scale of Horizon and the long period over which it
       has been in operation, yes?
   A.  10 KELs?
   Q.  Yes.
   A.  When you say it is material?  I don't --
   Q.  Let me change the question.  Use of words like "many"
       and "often" are capable of being really dangerous,
       aren't they, because they are capable of giving
       an impression unless you are clear about the scale of
       the occasions which you believe to be shown by the
       evidence, yes?
   A.  Yes.
   Q.  And what scale -- when you say "many" and when you say
       "often", what scale of numbers are you referring to in
       that paragraph, can you recall?
   A.  No, I can't recall.  There is a number.
   Q.  In paragraph 5.64 at page {D2/1/71} of the same
       document, the last sentence in that paragraph:
           "I have noted that hardware replacement often seemed
       to be a 'fix' of last resort where no other explanation
       could be given, and therefore there is certainly a
       possibility that hardware was at fault."
           Now you see there is a footnote there.  Do you see
       that, footnote 88?
   A.  Yes.
   Q.  Perhaps we could have a look at that, it is {F/178/1}.
       This is a KEL, it is dated -- it was raised on
       18th November 1999, so it is very early in the life of
       Horizon.
   A.  Yes.
   Q.  It was last updated in January 2004.
   A.  Yes.
   Q.  Under "Solution" -- and I see again that Atos is
       referred to, which is very curious because Atos wasn't
       on the scene for years until after 2004 and I do rather
       think that there must have been some word-processing
       change made at some point, I don't know how.
           It says:
           "This appears to be either the PM is typing ahead of
       themselves and the system suddenly catches up, or
       keyboard or screen fault generating spurious key
       presses.  Recommend that the PM tries not to type ahead
       of the system (or to press the same key a number of
       times if there appears to be no response from the
       system) or to replace the keyboard and/or screen.  PM
       should select existing. This functionality has been
       removed from CI4 ..."
           Which I think would be a release.  So there came
       a point at which it was no longer a problem, yes?
   A.  Yes.
   Q.  Now, that's one example of a possible hardware fault,
       yes?
   A.  Yes.
   Q.  How can that justify the claim that you make here that
       hardware replacements are often a fix of last resort?
   A.  Well, there's a number of other examples.  There is the
       phantom transactions example where hardware was changed
       because it was -- they were trying to work out whether
       it was environmental issues or not that were causing
       erroneous transactions.  There are a number of examples.
       There is only one cited here but there are a number of
       examples throughout the report.
   Q.  Are we talking about five examples or are we talking
       about 100 examples?
   A.  There will certainly be five examples that --
   Q.  I see.  So there "often" means something in the region
       of five, does it?
   A.  Yes.
   Q.  Then if we move on to page 97, paragraph 5.161
       {D2/1/97}:
           "Whilst both Horizon and Horizon Online contain a
       number of measures and controls designed to check system
       integrity, these mechanisms have been shown to have
       failed.  This is a point agreed upon in the Joint
       Statement.  It has been identified that known
       issues/bugs were often deferred and dealt with on a
       cost/benefit basis."
           That is a paragraph that we have discussed before
       and I have already suggested to you that the evidence
       gives no basis for claiming that it happened "often".
           But more importantly, I would like to ask you what
       you mean by "often" in that sentence.  Again you give no
       scale, no sense of how many times we are talking about.
       Are we talking about five times, ten times, a hundred
       times?
   A.  Quite a few times.  There is a document that looks at
       defect deferment, so a number of defects have been
       identified but they are deferred to be dealt with later,
       although they acknowledge that they could have an impact
       on branch.  This was the point that we discussed
       yesterday.
   Q.  This is the handful of PEAKs that you provided to me
       this morning, is it?
   A.  Yesterday we were taken to an incorrect reference in the
       document and --
   Q.  Do you mean a document you had incorrectly referred to
       in your statement or did I take you to a wrong document?
       Because if it is the latter, I'm terribly sorry.
   A.  Sorry, I think it was as a result of -- Mr Green
       interjected and directed to a document, but the document
       wasn't provided at the side of Magnum, so we should have
       gone to the footnote.  I would like to go there if
       that's possible.
   Q.  I'm afraid I don't actually know what you are talking
       about.  Does Mr Green?
   MR GREEN:  My Lord, I didn't want to press my intervention
       yesterday.  I will explain it in re-examination.  But
       5.166 is the paragraph which gives context to what
       I mentioned yesterday.
   MR DE GARR ROBINSON:  The document we looked at just before
       we broke?
   MR GREEN:  No, it is the release note.
   MR DE GARR ROBINSON:  Oh, I see.
   MR GREEN:  In the context of which there is an example.
   MR DE GARR ROBINSON:  Very good.
           So in answer to my question what's the scale of
       "often" in that sentence, what would your answer be?
       What sort of number of cases are we talking about?
   A.  A number, it will be the number that's actually
       contained within that document at footnote 156.
   Q.  Okay.  So the answer is to be found in that footnote, is
       it?  Thank you, I will look at that later.
           Let's move on to your second report at {D2/4.1/14}.
       If we could go to page 14.  At paragraph 3.13 you say:
           "For example, it appears that PEAKs are often closed
       or suggested to be closed if analysis has paused or has
       not uncovered a full diagnosis despite the Subpostmaster
       and/or Post Office not having a conclusion.  It is also
       not always clear whether a Subpostmaster was informed
       ..."
           Then at the end you say:
           "I have seen PEAK records that are closed despite
       support not being able to diagnose a root cause whilst
       acknowledging that there clearly is some form of
       error ..."
   A.  Yes.
   Q.  You will I think recall that my instructing solicitors
       wrote to Freeths to ask which PEAKs were being referred
       to.
   A.  Right.
   Q.  If we could look at {C5/36/1}, please.  Go to page
       {C5/36/2}, paragraph 1.  Here is Freeths' answer to that
       question.  The question is:
            "Please identify the ... PEAKs [referred to in that
       paragraph]."
           And the answer comes, there are nine PEAKs referred
       to.  Do you see that?
   A.  Yes.
   Q.  I don't have time to take you to them, there are
       subtleties in those documents which might otherwise be
       put.  But are you aware that of those nine PEAKs there
       have been only two examples since 2002?
   A.  I don't understand the question.  So within those
       PEAKs --
   Q.  Of the nine, seven of them occurred between 1999 and
       2001.
   A.  Right.
   Q.  One is from 2012 and one is from 2017.
   A.  Right.
   Q.  So the truth is that over the past 17 years this has
       happened, or there are PEAKs showing this as having
       happened twice, yes?
   A.  Yes, okay.
   Q.  Bearing in mind you are making a claim about these
       things often happening, could I suggest to you it might
       have been helpful and balanced for you to have indicated
       that that was the position?
   A.  Yes.
   Q.  That most of these occasions occurred during the very
       early years of the original Horizon.  Do you accept
       that?
   A.  Yes, it would have been helpful to include that.
   Q.  If I can take you to another example just before we
       break.  At 5.108 at page {D2/4.1/156} it is said:
           "At paragraphs 251 to 257 of his report, Dr Worden
       refers to the concept of 'User Error Correction'
       enabling the facility of correcting many software
       errors.  It should be noted that this would not apply to
       any bugs/errors and defects unbeknownst to Fujitsu or
       the Subposmaster.  It is evident from the PEAK analysis
       that often bugs lay undetected for weeks, months or
       years."
           My instructing solicitors wrote in the letter that
       we have just discussed, wrote a letter asking which
       PEAKs show that happening, and the response came at
       paragraph 9.2 of the letter, that's {C5/36/5}.
           The answer was:
           "Please refer to Mr Coyne's Supplemental Report
       (particularly paragraph 3.26 to 3.54) which sets out
       commentary in respect of various bugs ..."
           Now paragraphs 3.26 to 3.54 are the section which
       I think is headed "Acknowledged Bugs" but it is actually
       the four bugs, the four main bugs, that you focus on in
       your report, namely the three bugs identified by
       Post Office plus Dalmellington.  Do you remember that?
   A.  Yes.
   Q.  No particular reference is made to any other bug in this
       context.  How many other bugs do you say lay undetected
       for weeks, months or years?
   A.  Well, there was certainly Dalmellington.
   Q.  Yes, that's one of the bugs referred to in those
       paragraphs.  Let's not worry about weeks because one can
       understand why it may take time for a bug to be
       detected, but months or years.  In your list of 29 how
       many other bugs lay undetected for months or years, can
       you tell me?
   A.  I can't.  I would have to go through and --
   Q.  Could you give me a scale?  Between one and four?  Ten?
       Any idea at all?  It is just that by using the word
       "often" it suggests to me that you have a clear idea in
       your head, Mr Coyne.
   A.  Well, I will have had a clear idea in my head, I just
       don't know what the actual number is now that you are --
   Q.  An approximate number, a scale.  Can you give me any
       indication?
   A.  No, I would prefer to work it out properly and give you
       a proper answer to that.
   Q.  That is entirely fair.
           Let me ask one last question before the break.  In
       relation to bugs that are detected after a period of
       time there's evidence showing, I am sure you all agree,
       that investigations are undertaken by Fujitsu to ensure
       that all the branches that are affected in the meantime
       are identified, are you aware of that evidence?
   A.  Yes.
   Q.  It is said that this is a standard process undertaken by
       Fujitsu when they identify a bug that could affect
       branches, yes?
   A.  Yes.
   Q.  Do you have reason for thinking that that has not
       happened in any number of cases?
   A.  I am aware of one where the data was no longer available
       to investigate it.
   Q.  Which bug was that?
   A.  I would have to find the example.  It is in the report.
   Q.  I see.
   A.  I would have to find the example.  I know that on
       Dalmellington there was I think at least two occurrences
       where Fujitsu weren't able to identify what the impact
       actually was.  They were able to identify the number of
       branches --
   Q.  We will come to Dalmellington.  So it is Dalmellington
       and one other, those are two examples you are aware of,
       is that right?
   A.  I've certainly got an example of another, yes.
   Q.  Are you aware of any other examples of this not
       happening?
   A.  No.
   MR DE GARR ROBINSON:  My Lord, I don't know whether this
       would be a convenient moment?
   MR JUSTICE FRASER:  We will come back at 3.15 pm.  10
       minutes.
   (3.05 pm)
                         (A short break)
   (3.15 pm)
   MR DE GARR ROBINSON:  Mr Coyne, I would like to suggest to
       you that ultimately the main number in this case is the
       impact, the financial extent of bugs in Horizon, do you
       agree?
   A.  Yes.
   Q.  But it is interesting that you don't address that in
       your reports, do you?
   A.  The actual amount of the impact?
   Q.  Yes.
   A.  No.
   Q.  You have not been prepared to venture any figure which
       estimates the total impact in financial terms of bugs in
       Horizon, yes?
   A.  That is right, yes.
   Q.  Might I suggest that with your experience of IT risk
       analysis, you are perfectly capable of setting out at
       least in broad terms a view of the extent of the losses
       caused by the bugs that you have identified?
   A.  The best that I could do would be to go through the
       PEAKs and write down the numbers that were said to be
       wrong but I do not think that would be a very
       satisfactory way of doing it.
   Q.  What I suggest to you is that if you were to do that
       process or perform any judgment at all as to financial
       impact, you wouldn't find a number which remotely
       supports the claimants' case, would you accept that?
   A.  I haven't looked at the detail of the claimants' case.
   MR JUSTICE FRASER:  Just hold on one second.
           Mr de Garr Robinson, are you pursuing those
       questions as part of what ought to have been done on
       particular Horizon Issues or just as a general umbrella
       question?
   MR DE GARR ROBINSON:  Both.
   MR JUSTICE FRASER:  As a general umbrella question it is not
       a question for the witness.  But if you want to pursue
       it in terms of: to answer this Horizon Issue properly
       you ought to have done that, then you should do it by
       reference to the Horizon Issues.
   MR DE GARR ROBINSON:  Well, my Lord, I'm putting my case to
       the witness and that case is based upon what's in
       Dr Worden's expert report.
   MR JUSTICE FRASER:  No, I --
   MR DE GARR ROBINSON:  To which I will be coming in due
       course.
   MR JUSTICE FRASER:  Mr de Garr Robinson, this witness is
       giving evidence on the Horizon Issues.
   MR DE GARR ROBINSON:  Yes.
   MR JUSTICE FRASER:  I'm not saying don't put it, I'm saying
       if you are doing it in terms of: in order to address the
       Horizon Issue correctly, you need to identify it by
       reference to a Horizon Issue.
   MR DE GARR ROBINSON:  My Lord, I have a case to put which is
       based upon my expert's report.  The questions I am
       putting are establishing the building blocks for putting
       that case and I would be obliged if your Lordship would
       let me do that.
   MR JUSTICE FRASER:  But Mr de Garr Robinson, the two experts
       have approached each of their separate -- the same
       exercise, they have approached it in different ways.
   MR DE GARR ROBINSON:  Yes.
   MR JUSTICE FRASER:  That's the point.
   MR DE GARR ROBINSON:  And I have to put my case to this
       witness, my Lord.
   MR JUSTICE FRASER:  Yes, but he is called to give evidence
       on the Horizon Issues.  I'm not saying you can't put the
       question, what I'm saying is you need to put it by
       reference to the Horizon Issues that he is giving on
       rather than just arguing your case.
   MR DE GARR ROBINSON:  My Lord, I don't want to argue my case
       with this witness which is why I would like to ask him
       questions rather than engage in argument about
       particular Horizon Issues.  If your Lordship is saying
       that I have to argue with him about particular Horizon
       Issues then I will do that, but I would respectfully be
       obliged if your Lordship would let me put my case.
   MR JUSTICE FRASER:  Mr de Garr Robinson, you are slightly
       misunderstanding two things.  You are misunderstanding
       that cross-examination is just arguing with him, which
       it isn't.
   MR DE GARR ROBINSON:  Well, exactly.
   MR JUSTICE FRASER:  What I'm saying is by reference to
       Horizon Issues upon which he gives evidence, if you are
       going to put to him: in order to have fulfilled that
       properly you should have done X, Y and Z, and you
       haven't, then you can put the question, but you need to
       peg it back to the Horizon Issues.
   MR DE GARR ROBINSON:  My Lord --
   MR JUSTICE FRASER:  I think I have now said that three
       times.
   MR DE GARR ROBINSON:  You have.
   MR JUSTICE FRASER:  And I do not think it is a controversial
       point.  If you want to make it a controversial point in
       due course in submissions you can.
   MR DE GARR ROBINSON:  My Lord, let me explain to
       your Lordship where I'm going.  I'm loath to do that in
       front of the witness, but let me explain.
   MR JUSTICE FRASER:  Hold on a minute.  A couple of minutes.
           Just for the transcript, I'm asking the witness to
       step outside.
                 (In the absence of the witness)
   MR DE GARR ROBINSON:  I am going to take the witness to
       section 8.5 of Dr Worden's report.
   MR JUSTICE FRASER:  Yes.
   MR DE GARR ROBINSON:  In which he says that in order to have
       sufficient bugs to even begin to justify the claims
       being made by the claimants there would need to be in
       the region of 40,000, and I'm going to put it to him
       that that's right.
   MR JUSTICE FRASER:  Yes.
   MR DE GARR ROBINSON:  Now, is your Lordship going to permit
       me to do that?
   MR JUSTICE FRASER:  Yes, of course.  But the question you
       asked him was:
           "If you were to do that process or perform any
       judgment at all as to financial impact, you wouldn't
       find a number which remotely supports the claimants'
       case ..."
   MR DE GARR ROBINSON:  Yes.
   MR JUSTICE FRASER:  Now, by "number", I take that to mean
       financial number as in pounds, not shillings anymore,
       and pence.  Is that right?
   MR DE GARR ROBINSON:  Yes, that's --
   MR JUSTICE FRASER:  But that is not part of one of the
       Horizon Issues.
   MR DE GARR ROBINSON:  That's the point that's put in
       section 8.5 of Dr Worden's report, my Lord.
   MR JUSTICE FRASER:  Well, if you are going to suggest to
       him -- I'm going to deal with it on this basis.  As
       I made clear, I think, I'm going to let you put the
       question and pursue that line by reference to
       Dr Worden's evidence, but if you are going to maintain
       to this witness that in order properly to have addressed
       each of the Horizon Issues he should have come back --
       or, sorry, he should have arrived at an financial impact
       figure, which is a point I'm saying I'm allowing you to
       put, you need to do it by reference to which Horizon
       Issue you say he could only address properly if he did
       that exercise.  Is that clear?
   MR DE GARR ROBINSON:  My Lord, yes.
   MR JUSTICE FRASER:  All right.
   MR GREEN:  My Lord, just before he comes back in, the
       question that was put at {Day15/147:18} was:
           "Might I suggest that with your experience of IT
       risk analysis, you are perfectly capable of setting out
       at least in broad terms a view of the extent of the
       losses caused by the bugs that you have identified?"
           And the suggestion in cross-examination we hadn't
       done that anywhere, there are various examples.  At
       {D2/4/29} there is a table, and so forth.
   MR JUSTICE FRASER:  Mr Green, you can --
   MR GREEN:  I know it is a separate point.
   MR JUSTICE FRASER:  Mr Green, it is a separate point and,
       with respect, you can wrap those points up in your
       re-examination.  If Mr de Garr Robinson is putting
       a point which is clarified by a reference, the time at
       which to do that is in re-examination.
           Just before we have the witness back in, I think
       I've made it quite clear I have a fairly light touch in
       terms of your cross-examination.  I'm neither steering
       you one direction or another, I'm giving you virtually
       free rein, but there is a limit to the free rein because
       it has to be done by reference to the Horizon Issues
       which is what the trial is about.  The Horizon trial is
       completely different to the individual claimants' cases
       or their financial losses, for example.
   MR DE GARR ROBINSON:  Absolutely, my Lord.
   MR JUSTICE FRASER:  Right.  Can we have the witness back in
       please.
                 (In the presence of the witness)
   MR DE GARR ROBINSON:  Now Mr Coyne, could we first of all go
       to bundle {C1/1/1}, please, which is the Horizon Issues.
       I can take any of them, but I would like you to look at
       Horizon Issue 1 when it comes up on the screen.  I am
       sure you have read it many times.
   A.  Yes.
   Q.  It begins with the words:
           "To what extent was it possible or likely ..."
           That's a question which is asking the experts to
       explore the likelihood of bugs in Horizon causing
       shortfalls in postmaster branch accounts, correct?
   A.  Yes.
   Q.  As I suggested to you I think this morning, but it may
       have been after the luncheon adjournment, there are many
       different ways or there are various different ways in
       which an assessment could be made of extent of
       likelihood, yes?
   A.  Yes.
   Q.  One of those ways that might be thought to be very
       helpful in the context of this case is to consider
       whether the likely extent of bugs in Horizon has any
       chance of justifying any significant part of the claim
       that is made by the claimants, would you accept that?
       That would be a useful measure to adopt in a case of
       this sort?
   A.  Sorry, could you put the question again?  I do not think
       I understand the question.
   Q.  Horizon Issue 1 requires the experts to consider the
       extent of the -- and I'm using shorthand now --
   A.  Yes, that's good.
   Q.  I hope it is not controversial.  The extent of the
       likelihood of bugs in Horizon causing shortfalls in
       branch accounts.
   A.  Yes.
   Q.  And I'm suggesting to you that one useful yardstick for
       measuring extent is whether the likelihood in this case
       is of any sort which could begin to justify the claims
       that these proceedings are designed to decide.
   A.  Yes.
   Q.  And do you accept that that could be a useful yardstick
       for measuring extent in the context of this case?
   A.  I can see how it might be one of the contenders for
       that, yes.
   Q.  Thank you.  Now you know, because Dr Worden said so in
       his first joint statement, that Dr Worden was going to
       look at extent in that kind of way, yes?
   A.  Yes.
   Q.  He was going to look at financial impact?
   A.  Mm.
   Q.  And see whether the likely financial impact of bugs,
       which there were likely to exist, would have any chance
       of justifying the sort of claim, the sort of assertions
       that are being made in the context of these overall
       proceedings?
   A.  Yes.
   Q.  So you saw that he was going to do that and you have
       refrained from doing that, correct?
   A.  Yes.
   Q.  What I suggested to you before you very kindly stepped
       out was you could have done that.  You could, using your
       skills as an IT risk analyst, you could have, by
       reference to the KELs and the PEAKs that you had
       identified, formed an assessment as to the likely
       financial impact in each of the instances that you had
       identified?
   A.  Yes, I could.  If I could please answer that by way of
       illustration to the problem with that approach.
   Q.  Please do.
   A.  We will often see a bug, error or defect with a very
       wide range of impacts and the impact is typically
       whatever the counter was doing at that point in time.
       So if the counter was doing a foreign currency
       transaction for just £50 and something goes wrong, the
       discrepancy may be £50.
           By knowing that there is a bug, error or defect in
       the Horizon system that leads to problems with foreign
       currencies, you can't then say it is only a £50 defect
       because that isn't incorrect.  If another person was to
       be subject to that defect and they were doing a £10,000
       foreign currency transaction, they would likely have
       that same level of defect.
           Now that's an illustration to say you can't value
       bugs and their potential impact by looking at what has
       happened historically.  You can't value it in that way.
   Q.  But what you can do is form an estimate having regard to
       the totality of the PEAKs that you have seen, can't you?
   A.  But it is a fundamentally flawed approach.  If you have
       seen three branches that have had an impact because they
       were doing three foreign currency transactions, a £20,
       a £50 and a £100, but then there is a fourth person that
       believes that they have been subjected to that but that
       isn't recorded, you can't simply look at the three where
       it has been recorded and say the fourth couldn't
       possibly have occurred because we know there is a defect
       there.
   Q.  I'm not suggesting that you could arrive at a certain
       conclusion of an absolute cast iron number, but I repeat
       my question.  Can't you form an estimate having regard
       to the totality of the PEAKs that you have seen?
   A.  Yes, but your estimate would have to be based on the
       three people where it has been recorded to have
       occurred, so you would say £20, £30 and £50, and the
       best you could possibly do is come up with an average of
       that, and you would say that the value of that defect is
       whatever that is.
   Q.  There is another way that you could do it, isn't there,
       which is that you could look at the three bugs, the
       receipts and payments mismatch, the suspense account bug
       and Callendar Square, the ones that have been thoroughly
       investigated, and you could form inferences from the
       scale of those bugs, yes?  Would that be reasonable?
   A.  For those types of bugs, potentially yes.
   Q.  Those are quite large bugs, aren't they?  They are not
       small bugs in the scheme of things.  That's why they
       were identified in the letter, because these were major
       bugs of which even Post Office was aware?
   A.  Yes.  And carrying on from the one we were not told
       about until later, the Dalmellington one, there were
       some which were only pounds, just a few pounds, and
       I think there was at least one that was £25,000.
   Q.  But just fixing our attention for the moment on those
       three bugs, would you accept that the evidence indicates
       that altogether the total financial impact of those
       three bugs is no greater than £100,000.  Do you agree
       with that?
   A.  No, I haven't done that so I would not know whether that
       was right or close to right.
   Q.  You are aware of the calculations that were made by
       Fujitsu and the documents setting out the financial
       impact of those three bugs, and you are aware, aren't
       you, that that calculation has resulted in a figure that
       is less than £100,000?  Would you not accept that for
       the purposes of forming an estimate and scaling up from
       some examples, it would be useful to take that kind of
       figure as an example of sizeable bugs and start doing
       calculations on the basis of that sort of figure?
   A.  No, I don't accept that that's --
   Q.  Do you have any evidence to suggest that the financial
       impact of those bugs was more than £100,000?
   A.  I don't know.  Looking at Dalmellington there was
       a £25,000 in there, but I don't know what the outcome of
       that was, how that was fixed.
   Q.  Do you not accept -- sorry, you don't know -- you just
       made a point about Dalmellington.  You don't know how
       Dalmellington was fixed?
   A.  The document talks about -- I think there was only two
       that they didn't know how it was fixed.  The majority of
       them had been fixed by some form of correction to --
   Q.  All but -- you will remember there are two Dalmellington
       documents, the first one covered -- identified 118
       branches affected and dealt with 114 of them?
   A.  Yes.
   Q.  And as it happens all of those branches had been
       actually made good already because of the
       countermeasures that existed in the Horizon system, yes?
   A.  Yes.
   Q.  So of the 114 branches that were affected by the
       Dalmellington bug, in fact there was no lasting impact
       on any of those 114 branches, was there?
   A.  I think there was two that Fujitsu wasn't aware of.
   Q.  I have asked you about the 114.  You are trying to talk
       about the other four, aren't you?
   A.  Yes.  They were corrected, yes.
   Q.  So we can lay the 114 to one side, because the answer to
       the question: what was the lasting impact of the
       Dalmellington bug on those 114 branches?  The answer is
       zero, correct?
   A.  Yes.
   Q.  So we then move to the other four branches which were
       dealt with with another document, which I can tell from
       your answers you are completely familiar with.  There
       were four branches, weren't there?  Two of them appeared
       to have large impacts?
   A.  Yes.
   Q.  And the other two had impacts of around £1, I think one
       of them was pennies?
   A.  There were some very small ones.
   Q.  So shall we lay the pound and the penny to one side for
       a moment?
   A.  No, because if that indicates a defect then it depends
       what transaction it was doing, so we can't lay any of
       them to one side.
   Q.  Let's talk about the other two which I think were the
       two you wanted to talk about, yes?
   A.  Yes.
   Q.  Those two, it turned out, weren't examples of
       the Dalmellington bug and they didn't suffer any loss,
       did they?  There is another document which contains
       an analysis which demonstrates that, do you recall?
   A.  I don't recall, no.
   Q.  Very good.  Well, let's lay those two aside.  So we have
       got two bugs, one of which is for about £1 and one of
       which is for pennies, and we have 114 branches that were
       affected, all of which were made good by the
       countermeasures that existed in the system, yes?  And
       are you suggesting that the two branches that hadn't
       been hunted down, that had a £1 deficiency and a
       deficiency of two pennies, are you suggesting that it is
       to be inferred that those two branches uniquely weren't
       corrected by those countermeasures?
   A.  No, it should be the case that it all was corrected.
   Q.  Very good.  So would you accept that in relation to the
       Dalmellington bug, the overwhelming likelihood is that
       that bug caused no lasting deficiencies in branch
       accounts?
   A.  By "lasting" -- well, can you help me by saying what you
       mean by "lasting"?
   Q.  A deficiency that wasn't made good either by the SPM
       actually reversing the remming error, because of course
       Dalmellington was a bug which caused people to rem in
       amounts more than once when they -- which is why it
       looked like human error, and why it was picked up by the
       system and fixed as time went on.
   A.  Yes.
   Q.  That's why it took so long.  It is one of your examples
       of a bug that took a long time to identify.
   A.  Absolutely.
   Q.  But the reason why it took a long time to identify was
       because it looked exactly like a human error, didn't it?
   A.  Right.
   Q.  Right.  All of those instances were picked up by the
       system and either the SPM himself reversed the rem in
       some way and made himself good, or it was picked up by
       Post Office and TCs were sent, correct?
   A.  Yes.
   Q.  So that's what I mean by saying in relation to the
       Dalmellington bug there were no lasting deficiencies, no
       lasting shortfalls for which SPMs were ultimately made
       liable?
   A.  Yes, I think that is right.  I think once everything was
       detected everything was made good.
   Q.  In fact, Dalmellington is quite a good example of how
       countermeasures need to be brought into account and how
       it is important to look at financial impact to make
       an assessment of the extent question that's raised in
       Horizon Issue 1, isn't it?  Because it is no good saying
       118 branches were affected and some of them were
       affected by large amounts, when in fact we all know that
       even before the bug was detected the branches involved
       were in fact made whole anyway, yes?
   A.  It does suggest that, but it also suggests that there
       were deficiencies with the countermeasures because
       I believe it was five years after the first occurrence
       of the bug that the defect was finally discovered.
   Q.  Because, as you have already agreed, to an outside
       observer it looks exactly like a human error because it
       was a human error.  The bug wasn't causing losses, it
       was causing errors to be made and those errors were
       picked up, would you agree with that?
   A.  No, it wasn't a human error, it was a defect of the
       system.  It made it look like a human error.
   Q.  That's for another day.
           Going back to these three bugs.  Do you have any
       reason for thinking, any evidence to suggest, that the
       receipts and payments mismatch bug, the suspense account
       bug and the Callendar Square bug had a total net impact
       of more than £100,000?
   A.  As I said previously, I have not gone through and noted
       all the impacts of those so I could not give you
       a number.
   Q.  Would you accept that the best evidence is that the loss
       was under £100,000?
   A.  That still requires me to work out what the number is.
       All I can tell you is that it would appear that the
       receipts and payments mismatch, it was about 60 branch
       accounts that were impacted, but without going through
       and looking at the numbers I don't know.
   Q.  Well --
   A.  It was 30 per Callendar Square.
   Q.  Dr Worden has done that in his expert report, you
       recall, yes?
   A.  Mm.
   Q.  Are you aware of any evidence to suggest that
       Dr Worden's analysis is wrong?  Is there any evidence to
       challenge his calculations in relation to those three
       bugs?
   A.  No.  I understand the process that Dr Worden has gone
       through, he has looked at the numerical values that are
       recorded in the PEAKs for the branches that are
       available that are recorded in there and he has added
       those up, so I do not think his maths is going to be
       wrong.
   Q.  So if we were to take those these bugs as some kind of
       indication of fairly sizeable bugs that might appear in
       the system, it is fair to say, isn't it, that £100,000
       is quite small compared to the £19 million that's
       claimed by the claimants in this case.  It is less than
       1%, correct?
   A.  Just on the pure numbers, yes.
   Q.  So these three bugs, which are the ones we know most
       about, do not by themselves even begin to support the
       claimants' case, do they?
   A.  But the numbers that are given are only the numbers that
       are in the PEAKs, and the PEAKs only reflect where
       Fujitsu have become involved and have started to
       investigate the impact of those bugs from the branches
       that they are aware of.
   Q.  So assuming that £100,000 represents a fair assessment
       of the impact of those bugs, what would you say?  You
       would say, well, they are the tip of the iceberg.  There
       are many more bugs that are capable of producing the
       kind of financial loss that would justify the claimants'
       claim, would you say that?
   A.  Well, it is my position that there's many more than the
       three and I have set out these here.
   Q.  In paragraph 3.105 of your report you said potentially
       thousands, but I think you have moved from that now.
       Now you are saying perhaps up to 40, yes?
   A.  On the logic that we went through before, yes.
   Q.  Well, you accepted that logic, didn't you?
   A.  My position as stated in the report is that there is the
       potential for more.
   Q.  Well, I won't go back over the answers you have already
       given in cross-examination, Mr Coyne.
           Do you accept that 40 bugs are plainly nowhere near
       enough to have caused the claimants the shortfalls that
       they are seeking to recover?
   A.  I don't accept that position.  Because the bug impacts
       the transaction that would be in effect at the time, if
       it was a large transaction at the time then the impact
       of that bug would be a lot larger.  In the alternative,
       they may have experienced the bug a number of times but
       on smaller transactions.
   Q.  Well, let's see if we can agree some steps here about
       what can properly be done to scale up in this case.
       Could I ask you to go to bundle {D3/1/148} at page 148.
   A.  Yes, I have that.
   Q.  This is Dr Worden's first report and he says at
       paragraph 6.19:
           "Over the period 2000-2018 the Post Office network
       has consisted of more than 11,000 branches.  The mean
       number of branches in all years over the period has been
       about 13560."
           And he explains how the figure is derived.
           "If this evidence is accepted, the number of 'branch
       months' (a single branch, trading for a single month)
       has been ... 3,091,680. This is the number of monthly
       branch accounts that have been produced."
           I believe you have agreed that figure with Dr Worden
       in one of the joint statements, yes?
   A.  Yes.
   Q.  That leaves Dr Worden to formulate what he describes as
       a scaling factor between on the one hand the number of
       bugs on all branches over the lifetime of Horizon and
       that's what he calls scope A, and on the other hand the
       number of bugs on one claimants' branch in any one
       month, and that scaling factor is 3 million.  Do you
       accept the basic logic as a starting point?
   A.  I don't believe I do, no.
   Q.  You do refer in your report, Mr Coyne, to technical
       flaws in Dr Worden's analysis?
   A.  Yes.
   Q.  If you look at paragraph 620 is there a technical flaw
       in what Dr Worden says there {D3/1/148}?
   A.  No, I don't believe so.
   Q.  There is no technical flaw?
   A.  No, if the 3 million has come from the 3 million branch
       months.
   Q.  Then what Dr Worden has done is he has considered where
       there is evidence showing the need to adjust that factor
       to account for the possibility that the claimants'
       branches may not be typical when compared with the
       general body of sub-post office branches?
   A.  Yes.
   Q.  They might be more or less likely to be hit by bugs than
       your average Post Office branch?
   A.  Yes.
   Q.  And he sees no such evidence other than in relation to
       their size, do you accept that logic?
   A.  I don't accept that logic, no.  We see from the bugs
       that are reported that bugs appear to impact branches
       differently and one bug may hit a branch a number of
       times and only hit a handful of branches, and there
       doesn't appear to be any -- well, there will be
       an underlying technical reason for that but it is often
       down to the way that that branch is configured or the
       processes that they follow.
   Q.  Do you accept -- no, let me ask you this.  Are you aware
       of some special factor applying to the claimant branches
       which marks them out as very different from the rest of
       the branch network making their susceptibility to bugs
       very different from the wide range of branches in the
       rest of the network?
   A.  I'm not aware of it, no, but it would require -- in
       order to look at that it would require a detailed
       understanding of the particular processes of the
       branches.  And I think that's illustrated by what we
       know of Dalmellington, that impacted -- I think one
       branch was impacted five times, where there was only
       a handful of branches -- I think, was it 88 that was
       impacted in total?
   Q.  Mr Coyne, it is quite important not to get confused
       between particular instances of things happening and the
       law of averages.  What I'm asking you about is what's
       likely to happen over a large number of cases, do you
       understand the difference?  So of course if people are
       all walking down the streets, some of them are going to
       get hit by lightning.  Everybody won't get hit by
       lightning in the same way.  It will be a very small
       number of people who will be hit by lightning and
       they'll usually be in particular situations when it
       happens.  But one can make a generalisation about the
       likelihood of people being hit by lightning, do you
       understand, because of the law of averages?
   A.  I do understand, but in that illustration it would be
       very unusual for one person to be hit by lightning five
       times.
   Q.  But it has happened.  I remember, goodness gracious me,
       watching a programme, I think it may have been
       Blue Peter when I was a child, where precisely such a
       thing had happened.
           That's the point, that one has to step back from the
       particular.  When undertaking statistical analyses one
       has to step back from the particular and look at the
       broad range of consequences applying the law of large
       numbers and what's conventionally known as the law of
       averages.  You do understand that?
   A.  I do understand that, yes.
   Q.  And are you suggesting that you are aware of any factor
       relevant to the claimants' branches that means that
       there's some material, significant -- that they have
       some significant feature that takes them away from --
       that makes them completely different from the broad
       range of branches that exist in the Post Office network,
       large and small?
   A.  No, I'm not aware of it, but my suggestion is that in
       order to conduct that you should ensure that they are
       representative before using the law of averages.
   Q.  Well, one thing that Dr Worden has done is he has
       considered the size of the branches.  Unlike you, and
       I'm not making this by way of criticism, but he has
       thought about this very question to see whether there
       are any real differences between the broad range of
       branches in the network and the claimant branches and he
       has spotted that claimant branches tend to be smaller,
       okay, in the sense that they do fewer transactions per
       day on average than the average Post Office branch.
       Yes?
   A.  Yes.
   Q.  He does that at paragraph 623 on page {D3/1/149} of this
       document, and he reaches the conclusion that he sets out
       in paragraphs 624 and 625.  Could I ask you to read
       those three paragraphs, please.
           (Pause)
   A.  Yes, okay.
   Q.  Do you accept that these calculations that he does there
       in principle?  There is no technical flaw?
   A.  I am sure the maths is correct, but again I'm not sure
       that transactions is wholly the correct unit to use.
       Has it been considered what the value of those
       transactions are; whilst they do half the amount, do
       they do high value transactions, does that have
       an impact?
   Q.  Isn't that brought into account by -- isn't in forming
       a calculation the first question you have to ask
       yourself is: how likely it is that this branch is going
       to be impacted?  And then you have to form an assessment
       of what the size of impact is likely to be, yes?
   A.  Yes.
   Q.  At this point in the equation what Dr Worden is trying
       to do is to work out how likely it is branches are going
       to be affected?
   A.  Yes.
   Q.  He reasons that branches with more transactions, doing
       more transactions, are statistically over a large number
       of occasions more likely to be hit by a bug than
       a branch doing fewer transactions and would you agree
       the principle underlying that observation?
   A.  I think that's probably a reasonable principle.  If we
       look at the defects we found though, there is probably
       other factors that could be brought into that.
   Q.  But you are not aware of any, you are not in a position
       to suggest a single factor which you have any evidence
       to think actually applies in relation to the claimants
       as compared with the rest of the branch network?
   A.  I don't know the make up of the claimant, but one
       example might be, do they have an outreach branch?
   Q.  You are aware, aren't you, how small the number of
       outreach branches there is, how insignificant that is in
       the context of the Post Office network, aren't you?
   A.  Well, it wouldn't be insignificant to the people that
       are impacted by the defect and I don't know whether any
       of the -- because I haven't studied the claimants,
       I don't know whether any of the claimants have that or
       not.
   Q.  What you are doing is you are speculating that some of
       the claimants might have outreach branches?
   A.  I'm not.  I'm illustrating the potential problems with
       the approach that you are taking just simply using the
       unit of number of transactions.
   Q.  Mr Coyne, what I'm suggesting to you is that, in order
       to suggest that there is a marked difference in
       susceptibility to bugs as between the claimants and the
       general -- other Post Office branches, one needs
       a reason for doing it.  It is not enough to sit in
       an armchair and think of possible factors where you have
       no basis for knowing whether the factors apply or don't
       apply.
   A.  Yes.
   Q.  What you must do is simply work with the evidence that
       you have and there is no evidence to suggest that there
       is any particular preponderance of outreach branches
       amongst the claimants as compared with the general
       Post Office network, is there?
   A.  No, I'm not aware of the make up of the claimants.  No.
   Q.  So when it comes to the calculation that's set out --
       where is it?
   MR JUSTICE FRASER:  Are you looking for the smaller than
       average?
   MR DE GARR ROBINSON:  Yes.
   MR JUSTICE FRASER:  Because it is at 6 --
   MR DE GARR ROBINSON:  It is at 629 on page {D3/1/150}.
       Thank you my Lord.
           On the basis of the evidence that we actually have
       in relation to the claimant branches, the process that
       Dr Worden goes to in order to arrive at a scaling factor
       of 0.37, there are no technical flaws in that, are
       there?
   A.  No.
   Q.  Then in his second report, and perhaps we could go to
       that, it is at {D3/6/30}, it is paragraph 113, Dr Worden
       alighted upon a methodology that was more accurate,
       an improvement.  Could I ask you to read paragraphs 113
       and 114.
   A.  Mm. (Pause).  Yes.
   Q.  So here he refines his approach by aggregating the size
       of all claimant branches across three years for which
       data is available and then comparing that to the
       aggregate size of all branches across all three years
       and his view is that that is a more reliable approach,
       would you agree?
   A.  It would appear to be a more reliable approach than the
       approach that was being taken for before yes.
   Q.  That results in an increased factor of 0.45, yes?
   A.  Yes.
   Q.  You accept that that's a better approach in principle?
   A.  It is a better approach than the approach that was being
       taken before, yes.
   Q.  And do you accept that on the information of which you
       are aware, a scaling factor of 0.45 is in the right
       ballpark?  One can speculate that there might be other
       information out there producing a different result but
       taking account of the information of which you are
       aware, that scaling factor is in the right ballpark,
       yes?
   A.  Scaling based on number of transactions, yes.
   Q.  Thank you.  What that means is that because they have
       fewer transactions a claimant branch is less likely to
       be hit by a bug than an average branch, you accept the
       logic, yes?
   A.  No, I don't accept that logic, no.
   Q.  So do you suggest there should be a different scaling
       factor?
   A.  No, I don't accept that because they do less
       transactions they are less likely to be hit.
   Q.  Isn't it rather like going outside -- if we go back to
       the lightning analogy.  If people spend most of their
       time inside a building, they are much less likely to get
       hit by lightning than people who spend most of their
       time out in the countryside, yes?
   A.  Yes.
   Q.  Doing a transaction is a bit like going out into the
       countryside, isn't it?  It makes you vulnerable to the
       elements, yes?
   A.  Yes, but it is the types of transactions.
   Q.  You say types of transactions, Mr Coyne.  All of these
       branches did a wide range of transactions?
   A.  They do but some will do more of a particular type of
       transaction than others.
   Q.  That's also true of the general body of the Post Office
       network?
   A.  Yes.
   Q.  If there were some really curious difference between the
       kind of business done by the claimant branches and the
       kind of business done by the vast -- I mean you do
       accept, don't you, that the non-claimant branches in the
       Post Office network go from the very small to the very
       large and can do a vast range of different kinds of
       business?
   A.  Yes.
   Q.  If there were some feature that marked out the claimant
       branches as different from the general Post Office
       network branches, do you not think we would have
       identified that by now?
   A.  It isn't a piece of work that I have done so I wouldn't
       have identified it but I don't know whether anyone else
       has looked at that or not.
   Q.  My suggestion to you, Mr Coyne, is that you are raising
       an armchair objection on the basis not that you believe
       this would actually produce a different result in this
       particular case, but merely as an attempt to object to
       performing an estimate which it is in fact open to you
       and to Dr Worden to perform?
   A.  My perception is that it is a flawed process because the
       units, the inputs to the process I don't believe are
       correct.  I'm not arguing with the maths, it is the
       principles behind it that I'm not happy with.
   Q.  Let's move on.  If you are genuinely saying it is
       impossible to arrive at any judgment on these matters,
       if you had to make a business decision about risk and
       this was the only information available to you, you
       wouldn't just sit there and say you couldn't make
       a decision, would you?  You would perform a judgment on
       the best information you have, yes?
   A.  Yes, I agree with that, if a decision had to be made and
       there was no other information available, then I would
       use the best information that was available to me.
   Q.  And you would build in a margin of error, wouldn't you,
       to account for the possibility that there may be
       unknowns out there that could throw your figures out?
   A.  Yes.
   Q.  In the way that Dr Worden has done, correct?
   A.  Yes.
   Q.  Let me ask you this, you are suggesting that
       susceptibility of bugs may change depending on the kind
       of business that's done.  Is it your view, having regard
       to your close study of the 29 bugs that are in the JS2
       bug list, is it your view that those bugs are such that
       there is some feature in them which makes it likely that
       the claimants are going to be more or less susceptible
       to them than anybody else?
   A.  Not with regard to claimants because as I said I have
       not looked at those but there are bugs in there which do
       seem to be susceptible to particular branches.
   Q.  Now let's go back to Dr Worden's first report.
       {D3/1/150} please and paragraph 630.
   A.  Yes.
   Q.  On Dr Worden's approach:
           " ... Claimants' branches, being generally smaller
       than Post Office average, have fewer transactions per
       month and so are less likely to be hit by a Horizon bug
       in a given month."
           Here he refers to his 0.37 scaling factor.
           "The factor 0.37 increases the scaling factor above,
       between scopes (a) (see paragraph 617.1) and (c) (617.3)
       from about 3 million to about 8 million."
           Yes?
   A.  Right, yes.
   Q.  You accept that logic, don't you?
   A.  I accept Dr Worden's mathematics based on his logic,
       yes.
   Q.  So if you take a bug which has occurred 16 times over
       the lifetime of Horizon?
   A.  Yes.
   Q.  With a mean financial impact of £1,000 and that's quite
       a significant bug compared with most of the bugs you
       found, would you agree?
   A.  It is certainly significant in its impact, yes.
   Q.  It is in the top five, yes?
   A.  Quite possibly.
   Q.  And then you select a claimant branch a month at random,
       the chance of that bug occurring in that branch in that
       month is 16 in 8 million, correct?
   A.  If bugs affect branches equally, yes.
   Q.  And that's around 2 in 1 million and you have just
       accepted the logic, thank you.
           And the probabilities are obviously additive.  So if
       there is a second similar bug, the chances become 4 in
       1 million and so on.  So if there are 100 bugs it is 1
       in 5,000, yes?
   A.  But all this is predicated on accepting that bugs affect
       branches equally.
   Q.  Well let's do this Mr Coyne, let's assume that bugs
       don't affect branches equally.  I presume you are not
       saying it is impossible to say -- I assume you are not
       saying that the claimant branches are likely to be 100
       times more likely to be susceptible of bugs?
   A.  As I say I don't know the make up of claimant branches
       so I don't know about that.  All I do know is that for
       the bugs that I looked at they don't appear to impact
       branches equally.  Dr Worden referred to a rainfall
       across a field, that isn't a concept that I accept.
   Q.  Which is why he changed it in his second report to --
   A.  Lightning.
   Q.  Do you remember?  Or maybe it was the joint statement,
       I can't remember.
   A.  Yes.
   Q.  What I'm suggesting to you, Mr Coyne, is that you don't
       just throw up your hands and say it is theoretically
       possible that a particular claimant branch or particular
       set of claimant branches may do a particular kind of
       business making them more or less susceptible to bugs
       than other branches.  You need to have -- it is
       possible, isn't it, to have a sense of what the maximum
       likely impact of that phenomenon is, yes?  You are not
       suggesting that the claimant branch is 100 times more
       likely than any other branch bearing in mind that the
       rest of the Post Office network contain branches from
       large to small the entire spectrum of branches that
       exist in the network?
   A.  The way I would approach this would be possibly the way
       that Fujitsu did when they were investigating
       Dalmellington by way of example, in that they know it
       impacted 88 different branches but a number of those
       were impacted multiple times.
           I would have a look at the make up of those -- the
       business process that was followed in the branches that
       were impacted multiple times to find out why one branch
       was impacted three times and nobody else has been
       impacted at all.
           Once you understand what it is within that business
       process, you can then see from the knowledge you have
       got within the Post Office and Fujitsu who else across
       the estate operates in that same way.
   Q.  Mr Coyne, if we were just talking about one bug that
       affected a particular event or transaction, then I would
       understand what you are saying.  But we are not, are we?
       We are talking about a whole range of bugs operating and
       being triggered in a whole range of different
       circumstances?
   A.  Yes.
   Q.  So we have bugs which themselves have a scatter gun
       effect and we have a scatter gun of branches all over
       the country, some of which peppered all over the country
       are claimant branches?
   A.  Yes.
   Q.  What I'm suggesting to you is that when forming
       a judgment, trying the best you can do to make
       an assessment on the information available, what you are
       going to do in the absence of a strong indication that
       there is some factor that justifies the inference that
       all of the claimant branches are very different from all
       of the branches in the rest of the network, is you adopt
       the sort of approach that Dr Worden has adopted.
   A.  It isn't an approach that I would take.  I wouldn't feel
       confident in undertaking that type of approach.
   Q.  If you were making a business decision I think you
       accepted that you would form a judgment that would be
       the best judgment you could on the information you had,
       correct?
   A.  Yes. I think that is in a different scenario, isn't it?
       You are making a business decision.
   Q.  What it shows is that it is possible to make estimates
       that have some materiality and could be of use when
       forming judgment about human conduct, yes?
   A.  Yes.
   Q.  That's what Dr Worden has done and the calculation he
       arrives at is that, as we have said and I think you have
       accepted the logic of the position, that the chance of
       the bug we have discussed occurring in a claimant branch
       in a particular month are 2 in a million, and if there
       are 100 such bugs the chance would be 1 in 5,000.  And
       that's what Dr Worden says at paragraph 634 on page 151.
       Again do you accept the logic?
   A.  I accept the maths, yes.
   Q.  So for a bug to have even a 1 in 10 chance of hitting
       one claimant branch in one month there would need to be
       tens of thousands of such bugs, wouldn't there, yes?
   A.  Based on those mathematics, yes.
   Q.  Dr Worden says 50,000 in his first report but this
       becomes 40,000 in his supplemental report because he has
       chosen a different scaling factor, correct?
   A.  Yes.
   Q.  If it were to be a 5 in 10 chance there would need to be
       200,000 bugs, correct?
   A.  That's what he says.
   Q.  If one were to assume that the scaling factor were very
       different, you would still need an enormous number.  You
       would still need thousands of bugs, wouldn't you, to
       even begin to have a chance of justifying the sort of
       claim that is being made in this case.  It is a matter
       of commonsense, isn't it?
   A.  All I'm doing here is effectively just confirming
       Dr Worden's maths, but I don't accept the process.
   Q.  You don't accept the process because you are seeking to
       suggest there might be some factor you can't identify
       which means that the claimant branches have a different
       susceptibility to bugs when doing a transaction than the
       branches in the rest of the Post Office network, yes?
   A.  Yes, quite possibly.
   Q.  What I'm suggesting to you, Mr Worden --
   A.  Coyne.
   Q.  -- is that even if you factored in some enormous number,
       a substantial number, assuming the claimants were much
       more likely than your average Post Office branch to be
       susceptible to bugs, it would still be in the thousands.
       There would still need to be thousands of those bugs
       wouldn't there in order to justify anything like the
       claim made by the claimants.
   A.  I don't have the detail of the claims by the claimants.
       So I don't know the make up of them.
   Q.  Do you have any evidence that causes you think that
       there are in fact that scale of bugs in Horizon,
       thousands of bugs in Horizon?  I'm thinking as a result
       of the evidence you gave after lunch you don't, do you?
   A.  I don't -- there is the potential for that many in
       Horizon but I think it will be likely somewhere lower
       than that.
   Q.  Mr Coyne, the overwhelming likelihood is that there
       aren't more than 40 bugs of the sort that you have
       identified in Horizon, correct?
   A.  Yes, but what you have got to remember is each of those
       bugs can have an impact on multiple branch accounts.  It
       is not just one bug, one impact.
   Q.  So are you suggesting that some of those bugs have
       massive impact?  Remember, we are not talking about bugs
       that affect only the claimant branches.  We are talking
       about bugs that are in operation over the entire
       Post Office network.
   A.  Yes.
   Q.  So on any view those bug impacts -- the total bug
       impacts is going to be very, very much greater than the
       specific impact that might affect the claimants, yes?
   A.  Yes.
   Q.  So the total impact of the kind of bug you are
       suggesting, to justify the £19 million claim just made
       by the claimants alone, the total impact of the kind of
       bug you are hypothesising would have to be vast,
       wouldn't it, in the tens and tens of millions?
   A.  If we are using the law of averages to show that it has
       impacted everybody equally rather than just impacting
       the claimants' branches, yes.
   Q.  I see.  You are now suggesting that there might be bugs
       which have only affected the claimants and haven't
       affected the wider Post Office network, is that what you
       are claiming?
   A.  I'm saying it is entirely possible because we know from
       the bugs that have actually happened that they have only
       impacted a small number of branches.
   Q.  Is it really entirely possible though, Mr Coyne?  You
       are hypothesising a bug which has had many, many, many
       effects in order to justify the suggestion that it has
       generated a large number of losses.  That's what we are
       talking about?
   A.  Yes.
   Q.  Those bugs occur at branches because of factors that you
       are not prepared yet to identify, yes?
   A.  I haven't been given the information to enable me to
       identify.
   Q.  You don't have a specific bug in mind which has
       a particular feature which means that it only affects
       certain kinds of branches.  You are not telling me that.
       You are saying that it is possible there might be such
       a bug?
   A.  Yes.
   Q.  With this theoretical bug you are suggesting it is quite
       possible that it could affect just the claimants'
       branches and not affect the branches in the wider Post
       Office network.  Is that really your view?  Do you
       really think that is a likely outcome?
   A.  It may well have affected other branches in the branch
       network.
   Q.  That's my point, Mr Coyne.  It is all very well to say
       "I don't know, it is a really difficult judgment to
       make", but there are certain features which are just
       matters of commonsense.  What I'm suggesting to you is
       that, bearing in mind the claimants represent such
       a small fraction of the total Post Office network over a
       period of 20 years, it stands to reason as night follows
       day that if there are bugs which justify their claims
       those bugs would also have incurred losses in the wider
       Post Office network.  Would you accept that?
   A.  Yes.
   Q.  Would you accept, therefore, that the wider losses that
       would have been caused in the Post Office network would
       be substantially greater than the £19 million --
   A.  That's likely, yes.
   Q.  Yet nowhere do we see any sign of any bugs of the sort
       of scale that would be necessary to justify 100 billion,
       200 billion of losses caused by these bugs.  Do you not
       find that surprising?  Do you not draw any inferences
       from that fact, Mr Coyne?
   A.  But there are bugs that we have found that have impacted
       many branches.  So what if there is another -- I keep
       using the example of Dalmellington -- but what if there
       is another Dalmellington that has impacted 88?
   Q.  We have talked about Dalmellington and we have agreed
       I think that there is no net lasting impact from that
       bug.
   A.  But that was only cleared up in its entirety after five
       years.
   Q.  So you are talking about a bug which affects 88
       branches?
   A.  A number of which were impacted on multiple occasions.
   Q.  So was it 114 impacts on 88 branches, is that right?
   A.  I think that is right.
   Q.  You are suggesting that that phenomenon would justify
       the conclusion that there doesn't need to be thousands
       of bugs in order to justify the claimants' claim, is
       that right?
   A.  What I'm saying is using the profile of that type of bug
       illustrates how there can be a large number of branches
       impacted for a relatively long amount of time before it
       is dealt with and with a large range of branch impacts.
   Q.  Mr Coyne, so far you have found 29 bugs and you have
       very helpfully accepted that there are unlikely to be
       more than 40 bugs if one were to read each and every
       document, for which I'm obliged.
           The supposition is that 40 bugs are capable of
       causing £19 million worth of loss in the claimant
       branches without causing any greater loss in the Post
       Office network.  Is that what you are suggesting?  Are
       you suggesting that's a viable scenario which might
       explain what's happened?
   A.  You are asking me questions about the actual claim and
       I haven't studied the claim.  I'm aware of that headline
       number that you are talking about but I have not looked
       at the detail of the claim.
   Q.  I see.
   A.  So there is two possible scenarios.  There is probably
       many more.  But that it has impacted the wider branch
       network as well or that it has only impacted the
       branches which are claimants'.
   Q.  Sorry could you say that last point again?  I didn't
       hear.
   A.  What I'm saying is that there is a range of scenarios
       that go from additional bugs or the bugs that we found
       impacting just the claimant branches, is the start of
       the spectrum, up to bugs, errors and defects not only
       affecting the claimant branches but the wider Horizon
       estate.
   Q.  I suggest to you, and my Lord bearing in mind the speed
       with which this cross-examination has come, with
       your Lordship's indulgence I'm not proposing to put
       Dr Worden's section 8.7 analysis to him unless you
       indicate to me --
   MR JUSTICE FRASER:  To Mr Coyne?
   MR DE GARR ROBINSON:  Yes, unless your Lordship indicates
       that I really need to.
   MR JUSTICE FRASER:  No, it is entirely up to you.  It can't
       be said against you, I don't think, if you don't put it,
       and as I made clear at the pre-trial review and in a
       time limited trial generally a point like that would be
       ambitious -- are you concerned that if you don't go
       through it chapter and verse it will be said it hasn't
       been properly put?
   MR DE GARR ROBINSON:  Yes.  My Lord, important points should
       be put in my respectful submission.
   MR JUSTICE FRASER:  Yes, but this is really a methodological
       difference, isn't it?  So I'm not going to require you
       to put every single --
   MR DE GARR ROBINSON:  I'm obliged my Lord.
   MR JUSTICE FRASER:  Is your intention literally not to put
       any of it or just to put two or three headline points or
       do you want to think about it?
   MR DE GARR ROBINSON:  Would your Lordship give me one
       moment?
   MR JUSTICE FRASER:  Of course.
   MR DE GARR ROBINSON:  My Lord, I suspect that I will not be
       putting any questions but I'm going to take instructions
       overnight to see whether there might be some very short
       number of points.
   MR JUSTICE FRASER:  Mr de Garr Robinson that is entirely
       understood and entirely sensible.
   MR DE GARR ROBINSON:  My Lord, unless you have any further
       points this may be a convenient moment.
   MR JUSTICE FRASER:  All right.  That's fine.  I mean just to
       be clear Mr de Garr Robinson to make it -- well, to try
       and help you when you take instructions on the way you
       might want to or not to deal with 8.7, one way which is
       often adopted is simply to boil it down to two or three
       propositions and deal with it like that.  But I'm not
       saying you have to do that because I think both the
       experts have been quite clear about the different way in
       which they have and haven't approached the exercise.
       I don't know if that's helpful.
   MR DE GARR ROBINSON:  It is very helpful my Lord and I'm
       obliged to your Lordship.
   MR JUSTICE FRASER:  I have one very minor housekeeping point
       which may sound like a joke but isn't.  Yesterday was
       Day 14 and today is Day 15, what happened to Day 13,
       does anyone know?
   MR DE GARR ROBINSON:  No idea.  I was wondering that.
       I think it is Day 11 actually.
   MR JUSTICE FRASER:  It is definitely not Day 11.
   MR GREEN:  My Lord, wasn't the handing down of the recusal
       judgment which technically was counted as a Horizon
       day --
   MR JUSTICE FRASER:  Is that what --
   MR GREEN:  I think it may have, I will check --
   MR JUSTICE FRASER:  I'm not suggesting we re-number at all,
       as long as we are all working on the same numbering I do
       not think it matters.  All right.  Thank you very much.
       So Mr Coyne you are going to come back tomorrow.
   A.  Yes.
   MR JUSTICE FRASER:  Can I just raise one point about timing
       for both of you to think about.  Ordinarily in any time
       limited trial, which almost all the trials in this
       building, certainly in the QB part of this building are
       time limited, re-examination is kept to a very, very
       minimum and I think I did tell you Mr de Garr Robinson
       at the pre-trial review that although at that stage we
       were looking at two days cross-examination of this
       witness, which was then changed to four, that you would
       have the vast bulk of that time.
   MR DE GARR ROBINSON:  Yes.
   MR JUSTICE FRASER:  Just in terms of the court staff, not
       for my convenience, can you just liaise between
       yourselves about approximately the time you will finish
       on Friday afternoon.  I'm not making any indications one
       way or the other, but it would just be sensible for you
       to have a dialogue because Mr Green today I think has
       threatened re-examination on at least two and maybe more
       occasions.
   MR DE GARR ROBINSON:  My Lord, yes.
   MR JUSTICE FRASER:  So tomorrow 10.30 Mr de Garr Robinson?
   MR DE GARR ROBINSON:  My Lord, yes.
   MR JUSTICE FRASER:  10.30 tomorrow morning.  Thank you very
       much.
   (4.25 pm)
        (The court adjourned until 10.30 am on Thursday,
                          6 June 2019)