Monday 11 March 2019

Horizon trial: day 1 transcript

This is the unperfected transcript of the first day of the Horizon trial, held in court 26 of the High Court's Rolls Building. Enjoy.

                                          Monday, 11 March 2019
   (10.30 am)
   MR GREEN:  May it please your Lordship.  One of the matters
       your Lordship was going to deal with first thing today
       I think was the question of when the common issues
       judgment might formally be handed down and whether the
       claimants may have permission in the event that they
       receive a draft to disclose the draft to the steering
       committee of the claimants.
   MR JUSTICE FRASER:  Yes.  I think Mr Warwick's email said
       two members of the steering committee.  I don't know how
       many people are on the steering committee.
   MR GREEN:  Yes.
   MR JUSTICE FRASER:  This is the draft which is currently
   MR GREEN:  Indeed.
   MR JUSTICE FRASER:  Right and my position on this is as
       follows.  I have no difficulty with two members of the
       steering committee seeing the draft, subject to
       conditions.  One is the names of those two people have
       to be notified to my clerk in an email and that email
       should come from one of the claimants' legal advisors.
       It should identify who they are by name and the fact
       that they have been specifically told the terms of the
       embargo and that to breach the terms is a contempt and
       what the possible consequences of the contempt are.
   MR GREEN:  My Lord, I'm grateful.
   MR JUSTICE FRASER:  If those three elements are satisfied
       then they can be shown the draft --
   MR GREEN:  I'm most grateful.
   MR JUSTICE FRASER:  -- under terms of the embargo.
   MR GREEN:  I'm most grateful, my Lord.
   MR JUSTICE FRASER:  Now, handing it down, that ball is
       rather in your court and when I saw "your" court I mean
       the Post Office's court, although I know it is not
       Mr De Garr Robinson's side of it.
   MR GREEN:  Side of it yes.
   MR JUSTICE FRASER:  I was expecting or had invited typos to
       be provided by Wednesday.
   MR GREEN:  Indeed.
   MR JUSTICE FRASER:  Depending on the extent and scope of
       those I intend to hand the judgment down on Friday, but
       if there are an enormous quantity and it was a judgment
       that was sent out more in draft than I would ordinarily
       hope for, on Thursday morning I might say to you it's
       not going to be Friday, it's going to be the beginning
       of next week so I can use the weekend just to deal with
       the typos.
   MR GREEN:  Most grateful, my Lord.  Would it be possible for
       us to deal with any consequential matters the following
       Friday rather than this Friday?
   MR JUSTICE FRASER:  No, you can have longer than that.
   MR GREEN:  I'm most grateful.
   MR JUSTICE FRASER:  The important thing on consequential
       matters is that when I hand it down I make the order in
       the relevant terms for the parties to give them the
       necessary extension of time so that time doesn't run.
       I don't intend to put the parties in the position that
       time is running for the purposes of any consequential
       applications during this trial.
   MR GREEN:  My Lord, I'm most grateful.
   MR JUSTICE FRASER:  So you don't even necessarily need to
       come when it is formally handed down, depending on when
       that is.
   MR GREEN:  I'm most grateful.
   MR JUSTICE FRASER:  Is that clear?  Yes, good.
   MR GREEN:  My Lord, in terms of use of the time for
       claimants' oral opening, what I propose to do -- because
       I know your Lordship has obviously had sight of the
       expert reports and joint reports already.
   MR GREEN:  I propose to highlight a few key areas of
       agreement at a fairly high level, highlight a few key
       areas which we will say at this stage are likely to be
       important battle grounds for the trial, highlight the
       importance of the difference of approach of the two
       experts and the approach we're going to respectfully
       invite the court to adopt and then, which we hope will
       be useful for the court, take your Lordship through one
       worked example of a bug.
   MR JUSTICE FRASER:  One worked example of ..?
   MR GREEN:  Of a particular bug, so that your Lordship can
       see practically how a KEL works and practically how it
       relates to a PEAK -- or one or most PEAKs or one or more
       KELs because your Lordship will have probably
       appreciated that you can have multiple KELs referring to
       one or more PEAKs and vice versa -- the practical
       realities of the information flow between Post Office
       and Fujitsu in their system and the extent of the impact
       that one bug can have, the limitations on the sources of
       information available to the experts.
   MR JUSTICE FRASER:  You seem to be counting on your figures
       but I don't know what number you've got to.
   MR GREEN:  I'm sorry.  I will give your Lordship just
       a narrative overview to start with.
           The limitations of the sources of information
       available to the experts and what knowledge Post Office
       had of that bug and when, to inform the court as to its
       approach to these bugs and that will hopefully give
       the court at least at the outset of the trial both
       a practical understanding, in addition to what the court
       has already gleaned from the experts' reports, of how it
       actually worked, upon which we are going to place
       I think greater emphasis possibly than the defendant,
       and to provide a sort of microcosm illustration of what
       we, the parties and the court, now have, what those
       documents do and do not do, what the claimants have had
       to do, what Post Office doesn't do and what Post Office
       has known in relation to that bug all along.
           So with that brief introduction --
   MR JUSTICE FRASER:  We also at some point, and I would like
       to do it before lunch if possible because I don't want
       to cut into Mr De Garr Robinson's time, need to address
       in outline terms whether either or each of you want an
       adjustment to part 2 of the timetable to deal with
       experts' cross-examination, which was an email I sent
       out about two weeks ago.
   MR GREEN:  My Lord, yes.
   MR JUSTICE FRASER:  So we just mustn't forget that.
   MR GREEN:  At the moment I think our position is we are not
       seeking one and we are content to deal with it in that
   MR JUSTICE FRASER:  "We" being you?
   MR GREEN:  The claimants.  But there may be a different
   MR JUSTICE FRASER:  Well, I'm in a much more developed
       position than I was at the pre-trial review and I'm
       obviously raising the point for a reason but we can deal
       with it in outline terms at about 5 to 1, and please
       don't forget that the shorthand writers need a short
       break between about quarter to 12 and 10 past.
   MR DE GARR ROBINSON:  My Lord, I wonder whether now would be
       a good time to discuss that question.  It has been
       something that's been bearing on my mind.
   MR JUSTICE FRASER:  All right.
   MR DE GARR ROBINSON:  Your Lordship will be aware from the
       PTR that my judgment is four days is needed with
       Mr Coyne.  That remains my judgment.  Now that
       your Lordship may have had an even greater opportunity
       to read the expert reports and their length and assess
       the complexity of their contents and their length,
       your Lordship may have a different reaction to that
       submission than the one that you had during the PTR.
   MR JUSTICE FRASER:  Well, my position -- I will tell you
       what the situation is and then we can revisit at the end
       of the day if necessary, although because today is silks
       day there is a ceremony for the new silks at quarter to
       5 which means we will have to stop bang on half past 4.
           My position at the pre-trial review was fairly
       clearly based on anticipation of future expert
       agreements because I knew that they were still
       continuing to meet.  I knew what your position was about
       four days and I was fairly robust about that.
   MR JUSTICE FRASER:  Robustness being the word of the moment.
           Given what has happened since then, I am now more
       open minded to you having longer than two and a half
       days if you consider you need longer than two and a half
       days.  I know originally Mr Green said he wanted I think
       three but he seems broadly content with two and a half.
       There's also something of a difficulty brewing in the
       wings in respect of other litigation on Friday 5 April
       because as you know this court sits usually on Fridays
       on other business, so part 2 of the trial, by which
       I mean the expert evidence onwards, seems to me,
       Mr De Garr Robinson, if you tell me you need four days,
       I am going entirely to rejig the second half of this
       trial so you can have your four days.
   MR DE GARR ROBINSON:  I'm much obliged to your Lordship.
   MR JUSTICE FRASER:  So if you just want to mull that over
       and we obviously need to set the timetable in more
       outline terms, though nothing is set in stone, either
       today or tomorrow and that might mean we need to
       readdress when the closing arguments are going to be.
   MR DE GARR ROBINSON:  Yes.  My Lord, I was thinking more
       time with the experts and then perhaps oral closing
       arguments at the beginning of next term, but I have
       heard what your Lordship said and I'm grateful for the
   MR JUSTICE FRASER:  And that's not impossible either, that
       way forward.  It would break those two weeks, which are
       currently the first week after the break, expert
       cross-examination, and the second week, closings; it
       would break them into a week for the claimants' expert
       to be cross-examined by you, however many days of the
       next period Mr Green would want to cross-examine
       Dr Worden and whether that's two and a half or three,
       there's not really much difference, and then whatever we
       do about closings.  That's to give you my outline
   MR DE GARR ROBINSON:  I'm grateful.
   MR JUSTICE FRASER:  All right, Mr Green.
                 Opening submissions by MR GREEN
   MR GREEN:  Most grateful, my Lord.
           My Lord, there is, as your Lordship has seen, some
       fairly high level agreement.  Obviously the Horizon
       architecture is broadly uncontroversial and Horizon
       support, at least the outline of it, how it was provided
       is uncontroversial.
   MR JUSTICE FRASER:  Are there four statements now?
   MR GREEN:  There are four joint statements.  I think the
       fourth doesn't take it that much further.
   MR JUSTICE FRASER:  But there are four?
   MR GREEN:  There are four.
           And the experts agree that the PEAKs and KELs, the
       known error logs KELs and the PEAKs, which are the
       higher level record created in relation to a KEL, it
       goes up the support line; the experts agree that the
       PEAKs and KELs together form a useful source of
       information about bugs in Horizon, but while they are
       the best available resource to the experts, they provide
       a limited and incomplete window on what happened and so
       cannot paint a comprehensive picture and that's the
       agreed view of the experts, just to give your Lordship
       a brief outline of the landscape of the evidence they
       predominantly relied upon.
           It is also true that there are a number of other
       documents they have had reference to, as your Lordship
       would expect, I think various emails, Fujitsu
       presentations and other things which we will come to,
       how they provide greater insight.
           The PEAKs are created by Fujitsu's third and fourth
       line support teams.  So your Lordship has in mind how it
       works, the first line support seeks to identify whether
       there is an existing known error log entry that might
       relate to a support call.  If not, the second line
       support will create a new KEL, a new known error log
       record, and then it is the third and fourth line of
       support teams that create and investigate PEAKs.  And if
       your Lordship wants a reference just as a description of
       PEAKs there's a helpful one in joint 2 at 0.5, the
       reference of which is {D1/2/27} which is where the
       experts agreed:
           "PEAKs record a timeline of activities to fix a bug
       or problem.  They sometimes contain information not
       found in KELs about specific impact on branches or root
       causes - what needs to be fixed."
           Then there is some agreement about bugs and errors,
       robustness and potential for errors and measures and
       controls.  It is agreed that there were multiple bugs
       with the potential to have a financial impact on branch
       accounts and your Lordship will probably have seen from
       joint 2, paragraph 1.15 {D1/2/29} the range of views is
       between 12 and 29 bugs that the experts have identified
       as distinct for which they have seen strong evidence of
       the bug causing a lasting discrepancy in branch
       accounts.  So that's the range of opinion at the moment
       and for your Lordship's reference the table where
       Mr Coyne refers to those is {D2/4/16}, which is in
       Mr Coyne's supplemental report, which if we can just
       bring that up on Magnum, your Lordship will see there is
       an "Evidence of branch impact" column and where it says
       "Yes", that's where the relevant bugs that he has
       counted up to 29 are said to have strong evidence of
       potential impact on branch accounts.
           The areas of robustness where there is at least
       outline agreement, the experts agree that overall the
       system is relatively robust.  They agree that computer
       systems are considered more robust if access to the
       back-end database is restricted tightly and they agree
       that in 2012 Post Office's auditors observed that there
       were inappropriate system privileges in this regard and
       that's found at paragraphs 3.2 to 3.3 of joint 3, which
       is at {D1/4/3}.  And the source -- so that's
       paragraphs 3.2 to 3.3, your Lordship will see at the top
       there.  And the source for that is the Ernst & Young
       management letter which is at {F/869/32}, which over the
       page at 33 {F/869/33} identifies:
           "There are inappropriate system privileges assigned
       to the APPSUP role and SYSTEM_MANAGER role at the Oracle
       database level on the branch database server ...
       inappropriate privileged access ..."
           And so forth.  On page 34 over the page {F/869/34}
       they say:
           "Unrestricted access to privileged IT functions
       increases the risk of unauthorised/inappropriate access
       which may lead to the processing of unauthorised or
       erroneous transactions."
           And then at the bottom they say:
           "We noted that there is currently no process to
       review POLSAP user accounts or HNGX back-end user
       accounts on a periodic basis to determine that user
       access is appropriately granted given the job
       responsibilities.  As a result, our review revealed the
       following ..."
           And over the page they set out examples of what they
       have found and the bottom entry:
           "Whilst we noted that there was a monitoring control
       in place for privileged access to POLSAP whereby
       accounts associated to the SAP_ALL profile are reviewed
       and monitoring of failed and successful login attempts
       for SAP*, DDIC and BASISADMIN accounts is performed,
       this control does not include accounts associated to the
       SAP_NEW privileged profile."
           And so forth.  So that was a document which the
       experts identified in relation to that point.
           Then at the level of undetected errors, in joint 3,
       paragraph 3.6, which is {D1/4/3}, the experts agree
           "PEAKs show that some defects have lain undetected
       in Horizon for extended periods without being diagnosed
       and fixed."
           And your Lordship will anticipate we will look at
       that carefully through the prism of Dr Worden's
       countermeasures which he relies on for robustness and
       detection of errors and correcting them in a timely way.
           Then as to the effectiveness of the countermeasures
       themselves, the experts agree that the effectiveness of
       various countermeasures has changed throughout the life
       of Horizon.  The existence of the countermeasures has
       changed as well and your Lordship finds that at joint 3,
       paragraph 3.11 and 3.20; that's {D1/4/4}.  Your Lordship
       will see there 3.11:
           "The effectiveness of various countermeasures
       changed throughout the life of Horizon."
           And if we go over the page {D1/4/5}, at 3.20:
           "As Horizon has changed throughout its lifetime, the
       existence and effectiveness of any countermeasures has
       too.  To have considered the time dependence of all
       robustness countermeasures over 20 years would have made
       the expert reports impossibility lengthy.  There was not
       the time to do so."
           There is a final point on this page, while I'm
       there, which is also we say highly pertinent to one of
       the difficulties which is at the core of this
       litigation, which is that the experts are agreed at 3.22
           "Many software bugs can have the same effects as
       a user error (as illustrated, for instance, by the
       Dalmellington bug, which produced a remming error)."
           Your Lordship will anticipate how that will feed
       into an analysis of attribution of fault when
       a subpostmaster raises something and the experts now
       agree that many software errors will look like user
           Then moving on, as to reconciliation and transaction
       corrections, about which the court has heard quite a lot
       of evidence already in the common issues trial, the
       experts agree on the mechanics at least of
       reconciliation and TCs and just briefly as to that, it
       is clear that the process of data comparison between
       third party client data and Post Office held data, much
       of that appears to be automated to identify where there
       are discrepancies and some of those comparisons are
       carried out within Horizon itself, as defined, while
       others may not be, so there's a difference I think in
       terms of system about exactly where the automated
       comparison of external third party client data and
       Post Office held data is carried out and we can see that
       at {D2/4/22} which is Mr Coyne's second report at 3.37
       to 3.38 which is essentially agreeing with Dr Worden's
       comments about reconciliation.
           Then in relation to the manual allocation of
       responsibility and TCs, essentially the experts agree
       that there is a human or manual process which is
       inevitably subject to errors and in a sense to some
       extent that is as far as Dr Worden goes, subject to
       a point which I will return to as to the role that
       transaction corrections play in this trial.
           In relation to other quick topics where there is
       broad agreement, issues 2 and 14, which is Horizon
       alerting and reporting facilities, there are two big
       agreements about that.  One is that Horizon doesn't
       alert SPMs to bugs and errors itself and that's joint 2,
       paragraph 2.1 at {D1/2/38}.  And then in relation to
       functionality for SPMs obviously it's agreed that
       Horizon's functionality allowed subpostmasters to check
       cash and stock by counting the same and inputting those
       figures.  The experts generally agree on the steps and
       processes in dealing with discrepancies and they agree
       that it was not possible to record a dispute on Horizon
       itself and that's joint 2, paragraphs 14.1 to 14.3 at
           And the two last areas of high level agreement are
       Horizon shortfalls data and reporting for
       subpostmasters, which is 8 and 9.  There is fairly high
       level agreement about available data and reporting.
       They agree factually about what transaction data and
       reporting functions were available to subpostmasters and
       Post Office respectively and that's joint 3 at 8.2
       {D1/4/10}.  And they agree, importantly -- although
       your Lordship has already heard some evidence about this
       in the common issues trial -- that Post Office had
       access to several sources of information to which
       subpostmasters were not privy, which is joint 3 at 8.1,
       which is {D1/4/10}.
   MR JUSTICE FRASER:  Right.  I'm a bit puzzled at some of
       your references although you were going quite quickly so
       it might be that I missed some of them.  You said
       joint 2, paragraph 14.1 to 14.3.  I'm not sure if you
       actually turned that up, did you?  Was that turned up on
       the common screen?
   MR GREEN:  {D1/2/40} is 14.1 to 14.3, my Lord, at the bottom
       of page 40.
   MR GREEN:  So that's where the expert agreement is and the
       reference is -- they haven't set out.
   MR JUSTICE FRASER:  No, I was looking for a different --
       I was looking for subparagraphs of the earlier number 14
       which was euro discrepancies which was on page 12, which
       is also a paragraph 14 but it doesn't have
   MR GREEN:  Indeed.
   MR JUSTICE FRASER:  Yes, so ...
   MR GREEN:  I'm sorry if I'm going too fast but this is
       just --
   MR JUSTICE FRASER:  Well, I don't know -- when you refer me
       to these I don't know whether you are expecting me
       hurriedly to look at them on here before you go on to
       the next one or not.
   MR GREEN:  I will perhaps indicate if I invite your Lordship
       to look carefully at it.
           In relation to remote access and editing of
       transactions, your Lordship will appreciate this is
       going to be something of a hot potato during the trial.
       There is a measure of agreement that certain categories
       of access or tool have been available at various times
       and at the moment I don't want to go into that in too
       much detail save to identify what the types are, if
       I may.
           So there is Post Office remote access which was at
       a minimum read only access to all branch transaction
       data, which is referred to at joint 3, paragraph 7.2
       which is {D1/4/10}.  I'm not asking your Lordship to
       look at it particularly.
           There is the Fujitsu ability to insert transactions,
       one such tool being the balancing transaction tool used
       by Fujitsu's SSC, system support centre, third line
       support, which has been used on at least one admitted
       occasion {D1/5/5}, which is joint 4 at 10.3.
           There is privileged users with the capability to
       edit or delete transaction data and there are some logs
       relating to this access but what's recorded in those
       logs has changed significantly over time.  That's
       joint 4 at 11.3 {D1/5/10}.
           There is accounts altered without consent which is
       changes to an SPM's branch transaction data made without
       a subpostmaster's consent which your Lordship will find
       for example at Dr Worden's first report, paragraph 1136
       which is {D3/1/248}.
           And there is also reference there to the rebuilding
       of branch transaction data --
   MR DE GARR ROBINSON:  I'm sorry, I don't understand
       my learned friend's reference to paragraph 1136.  It may
       not be helpful but I wonder if he could explain what he
       means about Dr Worden agreeing.
   MR GREEN:  So if your Lordship sees, there is a table at
       1136 where Dr Worden sets out what he says can be done
       and it has got "Data amendment" as the first column and
       he says:
           "Whether Post Office has had the ability to:
           "Inject/inject: yes, global users have had that
           "Edit/delete: no."
           And then "without the knowledge of subpostmaster:
       no"; "without the consent of subpostmaster: yes".
           So I'm just trying to summarise the effect of what's
       in that column.
   MR JUSTICE FRASER:  That's how I read it.
   MR GREEN:  I hope that's helpful to my learned friend.
   MR DE GARR ROBINSON:  I'm grateful.
   MR GREEN:  Then the final one is the transaction repair tool
       which is referred to at {D1/5/3-4}.  It is at the bottom
       of that page:
           "The earlier transaction repair tool affects only
       [this is Dr Worden] back-end copies ..."
           And then he lists some of the others.
   MR JUSTICE FRASER:  Sorry, where were you just looking at?
   MR GREEN:  Sorry, the bottom bullet point on that internal
       page 3, which is actually --
   MR JUSTICE FRASER:  The one that begins "The earlier
       transaction repair tool ..."
   MR GREEN:  Exactly:
           "... affects only back-end copies of transaction
       data [this is Dr Worden's view] and so does not enable
       remote access."
           And then he refers to other things.  On that,
       my Lord, I just invite your Lordship to note on page 7
       {D1/5/7}, at 10.11, the experts agree:
           "We have not been provided with logs or audits from
       the transaction repair tool (TRT)."
           So just trying to pick out the broad landscape,
       where the big disputes will be, one of the big areas of
       difference between the parties is as to robustness and
       has a couple of facets to it: one is its utility in
       terms of determining the individual other issues in the
       trial and we have referred to that on page 9 of our
       written opening {A/1/9} including at, for example,
       paragraph 17.1.
           The issue of robustness has been a fairly focal
       point of the defendant's case throughout.  They first
       raised the contention that the Horizon computer system
       is robust in their response to the Panorama documentary,
       which is at {F/1422/1}.  Your Lordship will probably
       remember this document from it before, but the
       Post Office was responding to the Panorama programme on
       17 August 2015 and it says that the allegations are
       based on partial, selective and misleading information:
           "The Post Office does not prosecute people for
       making innocent mistakes and never has.
           "There is no evidence that faults with the computer
       system caused money to go missing at these Post Office
           "There is evidence that user actions, including
       dishonest conduct, were responsible for missing money."
           And so forth and then it refers to the allegations
       and then the last paragraph before "Background facts" is
       where they say:
           "The Horizon computer system is robust and effective
       in dealing with the six million transactions put through
       the system every day by our postmasters and
       employees ..."
           And so forth:
           "It is independently audited and meets or exceeds
       industry accreditations."
           Then the letter of response makes four references to
       "robust".  I won't take your Lordship to them.  The
       generic defence then reflects that language at
       paragraph 16, paragraph 50 and paragraph 153, which are
       respectively -- we don't need to turn them up, but are
       respectively for reference at {C3/3/5}, {C3/3/21} and
           And then the third CMC, Post Office themselves
       particularly proposed issue number one, "Is Horizon
       robust and extremely unlikely to be the cause of
       shortfalls" and for reference -- we don't need to turn
       it up -- that's the skeleton argument at {C8.4/2/25}
       and then that becomes the focal point of Dr Worden's
       report and we respectfully say feeds into his approach
       to the evidence as a whole.  So what he does, in
       contra-distinction to Mr Coyne, is Dr Worden looks at
       the idea of robustness in a particular way, which I will
       break down in a moment, and then through the prism of
       robustness then looks at the other issues and we
       respectfully will invite the court to adopt an approach
       more consistent with Mr Coyne's approach which is to
       look at what actually happened with particular examples
       and trace them through to a reasoned and careful
       conclusion and then from the ground up, as it were, draw
       inferences upwards rather than from an overarching
       hypothesis downwards.
           Your Lordship has probably seen that in addition to
       that we have questioned the extent to which something
       being said to be robust provides any solace to an
       individual postmaster and that's at page 10 in -- it is
   MR JUSTICE FRASER:  Page 10 of?
   MR GREEN:  In our written submissions.  And the short point
       there, my Lord, is what's different about this is this
       is not a provider of an IT system and the customer and
       the customer being able to aggregate and deal with
       things in the round with the provider.  We're looking
       one hop over and it is no solace to an individual SPM
       who faces an unexplained shortfall which might be a lot
       of money for them, even though a small amount of money
       relatively for Post Office, to be told "But don't worry,
       it's a rare occurrence", and so we respectfully say that
       the entire universe of transactions is not the right
       comparator for answering at least the other questions
       and possibly the question of robustness in this trial.
           We have made clear from the latest our generic reply
       that we respectfully say that a low chance multiplied by
       a high number of transactions is totally consistent with
       the levels of claims being advanced by the claimants in
       this case.  My Lord, if I can just very briefly take
       your Lordship to that, {C3/4/21}.  It's at paragraphs 36
       and 37 of the reply.
   MR JUSTICE FRASER:  Which paragraphs?
   MR GREEN:  It is paragraphs 36 to 37:
           "Notwithstanding the various checks and
       controls ..."
   MR JUSTICE FRASER:  It is really 35 to 37, isn't it?
   MR GREEN:  It really is, my Lord, your Lordship is right.
           Then in 37, second line:
           "In fact, the relatively small chance of errors
       admitted by the defendant, would be likely to produce
       the very picture reflected in the claimants' case.  The
       existence of money in the defendant's
       suspense account(s) ... shows that errors generate
       financial consequences by which the defendant receives
       money to which it does not believe it is entitled, and
       credits those sums to its profits."
           That's a reflection of the suspense accounts being
       credited profits after three years which we will turn to
       when we get to Dr Worden's evidence.
           "This too is consistent with the claimants'
       case ..."
           And if we go over the page {C3/4/22}, it identifies
       the publicly maintained position and if we can go
       forward to page 29 of that document which is {C3/4/29},
       we look at paragraph 52:
           "... the claimants repeat its primary case pleaded
       in GPOC and aver that the terms identified are onerous,
       oppressive ..."
           Et cetera and that relates back -- that's the
       position on the common issues reflecting the 52.5, the
       identification of some errors and so forth.  So that's
       how that fed back into the trial your Lordship has
       already heard.
           In relation to robustness there's the sort of second
       big area, but it's absolutely dovetailed with
       Dr Worden's definition of robustness, which is his
       definition of countermeasures which your Lordship will
       have seen.  He has identified these countermeasures and
       he has drawn inference about their effectiveness and as
       a result of those inferences concluded that the Horizon
       system is robust and the most extraordinary of these is
       the treatment of transaction corrections as
       a countermeasure, because Post Office's position, we
       respectfully say, is illogical and effectively a one-way
       ratchet in Post Office's favour.  So the approach is to
       say that the process of making transaction corrections
       is relevant as a countermeasure but not within this
       trial, that's effectively the position adopted, and it
       is we respectfully say an unusual position.
           If we look at {A/2/44}, this is Post Office's
       written opening for the Horizon issues trial, and we can
       see at paragraph 114 the opening line of that:
           "Post Office's reconciliation processes are outside
       the scope of the Horizon issues."
           So that's the prism through which the Post Office
       invites your Lordship to be constrained for the purposes
       of this trial.
           Now, we compare that to what Dr Worden refers to at
       {D3/1/71} and we see at paragraph 257 that Dr Worden
           "In my opinion, the countermeasure of UEC was so
       essential in Horizon, and it was effectively
       implemented.  Because of this, many software errors
       resembling user errors were also corrected."
           Now, there what he is talking about is user error
       correction, which is not the correction of user errors,
       it's subpostmasters spotting software errors and drawing
       them to the attention of Post Office and that feeding
       through to Fujitsu.  And we will look carefully at what
       the information flow in real life looks like, but UEC is
       not just an important countermeasure, in it Dr Worden's
       opinion UEC, which is SPMs spotting system errors and
       pointing them out successfully, UEC was "so essential in
           If we then go to page 78 in that document {D3/1/78},
       I would invite your Lordship to look carefully at
       paragraph 294, flatly contrary to Post Office's position
       in its written opening:
           "The processes of reconciliation, TAs and TCs are
       a very important part of the robustness countermeasures
       built into Horizon - particularly for UEC."
           So, my Lord, that ties back to that paragraph we
       have just looked at.
           Then just as an extra point, when Dr Worden is
       presented with KELs in which it appears there might have
       been a financial impact on branch accounts, he
       frequently infers that TCs would have been issued to
       correct the position and therefore it's robust.  We will
       explore some examples of that with Dr Worden in due
       course.  And therefore it would be certainly helpful if
       my learned friend could clarify what Post Office's
       position is about the relevance of TCs.  If they do not
       form part of the Horizon system, robustness of which
       they have insisted is in issue in issue 1, then to the
       extent that that robustness is dependent on transaction
       corrections, the system is not robust.  If it is in then
       it is dependent on that process of transaction
       corrections and your Lordship will obviously appreciate
       from the previous trial what that involved in practical
           So ultimately, the approach of Dr Worden more
       globally we respectfully say has an element of
       circularity about it because what we find is that where
       bugs are detected -- so where we have been able to find
       evidence of bugs -- Dr Worden prays that in aid to show
       that "Well, they can be detected, so the system is
       working, so it is robust", where bugs have not been
       detected by the system that's prayed in aid as the
       absence of bugs and where bugs have been found that have
       gone undetected for a long time, but are finally found
       several years later, that is still said to be evidence
       that the system and controls are all robust.  So in fact
       the position is that it is not possible to identify any
       evidence which could ever lead to the conclusion the
       system is not that robust on that basis and we
       respectfully invite the court to scrutinise that
       approach with some care.
           The big difference which I have already indicated
       between the approach of Mr Coyne and the approach of
       Dr Worden is Mr Coyne has started with the primary
       source material and the PEAKs and KELs -- and we will
       illustrate this in a moment in relation to
       Dalmellington -- and has tried to build up the picture
       from those, whereas Dr Worden has looked at everything
       through the prism of the robustness countermeasures that
       I have just been describing.
           The third big area of dispute is remote access.
       Your Lordship will appreciate that the position on
       remote access has developed somewhat to say the least.
       It is worth just briefly recapping how that position has
       developed.  Your Lordship will remember that Post Office
       originally made its public statement in response to the
       Panorama programme which denied any ability remotely to
       alter branch data.  That's at {F/1422/2} and you see
       there halfway down the page:
           "There is also no evidence of transactions recorded
       by branches being altered through 'remote access' to the
       system.  Transactions as they are recorded by branches
       cannot be edited and the Panorama programme did not show
       anything that contradicts this."
           Your Lordship will remember Mr Roll appeared on the
       Panorama programme.
           We have set this section out at page 22 of our
       written opening {A/1/22} and we have traced through the
       development of the position.  It is clear that what
       Post Office told the public in response to the Panorama
       programme was not true.
           At paragraph 65:
           Following the claimants' letter of claim ...
       Post Office admitted (in their 28 July 2016 letter of
       response) that in fact Post Office and/or Fujitsu did
       have some limited capacity remotely to access and edit
       transactions ..."
           Then at the GLO hearing my learned friend said to
       the master -- wanted to deal with the remote access
       point upfront and explained that the people who gave
       that response thought it was correct.  He said:
           "... Horizon system is a very complicated system.
       It involves lots of departments in ... both in Fujitsu
       and in the Post Office.  And the people who are
       responsible for the correspondence didn't know that in
       fact there were these two other routes.  Very few people
       at Post Office knew that there were these two other
       routes.  They were ... they were routes that are
       under ... essentially under the control of Fujitsu who
       is the expert independent contractor ..."
           And then at the bottom:
           "... the point having been discovered, the
       Post Office wasted no time in ... in bringing the
       truth ... the accurate ... and accurate set of facts to
       the knowledge of the claimants."
           And at the end of that section:
           "... the fact that there is a possibility to alter
       remotely itself is ... not in issue, it now having been
       discovered and the fact that there are these clever
       routes by which they can be done."
   MR JUSTICE FRASER:  I think you have gone over the page
       to 23.
   MR GREEN:  I have indeed, my Lord, yes.
           So that was what the court was told at the GLO
       hearing, that it shouldn't be in issue because it had
       now been admitted, and then the generic defence, we
       identified what was admitted at that point paragraph 57
       of the generic defence which is {C3/3/25}.  In fact can
       we just go back one page to 24 {C3/3/24}.  There's an
       admission at the bottom there:
           "... bugs or errors ... have resulted in
       discrepancies and shortfalls or net gains ..."
           Go over the page and then that deals with those bugs
       and errors.  You come down to the "Remote editing of
       branch transaction data" heading at 57 and then
       subparagraph 1 says:
           "Neither Post Office nor Fujitsu has the ability to
       log on remotely to a Horizon terminal in a branch so as
       to conduct transactions."
   MR JUSTICE FRASER:  What's the name of the Post Office or
       the person who signed the statement of truth on the
       generic defence?  There is just a signature on it but
       I can't see what the name is.  {C3/3/75}.
   MR DE GARR ROBINSON:  My Lord, it is Jane MacLeod.
   MR JUSTICE FRASER:  Jane MacLeod, right.  Thank you.
   MR GREEN:  Your Lordship will see that the case is then set
       out in the subsequent subparagraphs:
           "A Post Office employee with 'global user'
       authorisation can, when physically present at a branch,
       use a terminal ..."
           And then:
           "Any transactions effected by a global user are
       recorded against a global user ID and are readily
       identifiable ..."
           Then 3 {C3/3/26}:
           "Fujitsu (and not Post Office) has the ability to
       inject transactions into branch accounts (since the
       introduction of Horizon Online in 2010, transactions of
       this sort have been called 'balancing transactions')."
           Then halfway down 3:
           "They may be conducted only by a small number of
       specialists at Fujitsu and only in accordance with
       specific authorisation requirements.  They are rarely
       used.  To the best of Post Office's information and
       belief, only one balancing transaction has ever been
       made so as to affect a branch's transaction data, and
       this was not in a branch operated by a claimant.
       A balancing transaction is readily identifiable as
           And so that's the account that we get which appears
       on its face to conclude the allegation which had been
       originally made by Mr Roll in Panorama, subject to --
   MR DE GARR ROBINSON:  My Lord, just to be clear -- I'm sorry
       to interrupt, but just to be clear, your Lordship may
       recall an argument in the pleadings, this is talking
       about Horizon Online, paragraph 57 is principally about
       Horizon Online not about Legacy Horizon and it is
       important not to swerve between the two different forms
       of Horizon in relation to which different forms of
       remote access were possible.
   MR GREEN:  My Lord, I agree it is important not to swerve
       between them.
           The development of the case in terms of evidence
       essentially followed the service of Mr Roll's witness
       statement, Mr Roll's first witness statement {E1/7/3}.
       Now, in fact that statement was dated 11 July 2016 --
       your Lordship will understand why a witness statement
       was taken at that stage before a pleading of this
       type -- and it was served on 28 September 2018 and at
       paragraph 18 of that witness statement, for example,
       Mr Roll says:
           "The ability to remotely access the Horizon system
       at branch level was extensive, in that we were able to
       change not only data and transaction information, but we
       also had the ability to insert transactions and transfer
       money remotely without the subpostmaster knowing.
       Obviously this was not done by me, however I can recall
       thinking that a third party may have been able to do
       that if they could have remotely accessed the system in
       the way that we could (which may or may not have been
           So he is being qualified there about him not having
       done it himself.
           We then on the same day are served the witness
       statements of Mr Godeseth and Mr Parker, Mr Godeseth
       first witness statement on 28 September 2018.  He says
       that any remotely injected transaction would be easily
       identifiable as such because it would have a counter
       number greater than 32 and the reference to that is
       {E2/1/17}, 58.10.  And it is right to distinguish
       between Legacy Horizon and Horizon Online:
           "In Legacy Horizon any transactions injected by SSC
       would have used the computer server address as the
       counter position which would be a number greater than
       32, so it would be clear that a transaction had been
       injected in this way."
           So that's served on the same day as Mr Roll's first
           Then we get the 16 November 2018 statement of
       Mr Parker, Parker 1, {E2/11/5}, at paragraph 22.  He
       says -- notwithstanding that he is responding to
       Mr Roll, he says:
           "It is correct that the 'remote access' described
       above could have been carried out without the permission
       of a subpostmaster.  However, any additional
       transactions inserted remotely would be identifiable as
       such from the transaction logs that are available to
       subpostmasters from Horizon."
           So the effect of those two witness statements, one
       at the same time as Mr Roll and one responding to
       Mr Roll, is to preclude a transaction appearing as if it
       has in fact been done by a subpostmaster when actually
       it hasn't and to do so clearly.
           Then Dr Worden says that Post Office's evidence
       accords with his experience with support staff more
       generally and we see at {D3/1/244}, paragraph 1114, he
       effectively takes a view on the conflict of evidence
       between Mr Roll and Mr Godeseth and sides with
       Mr Godeseth saying that it accords with his experiences:
           "... that support staff should have a facility like
       this, so that branch accounts could be corrected in
       exceptional circumstances - without resorting to DBAs."
           And if we go forward to page {D3/1/245},
       paragraph 1119, he said:
           "Mr Roll worked in the SSC and I established
       above ... certain SSC users had the ability to transact
       injections ..."
           I think it might be the other way round, inject
           "... although these would have become visible to
       subpostmasters.  So in my opinion Mr Roll could not have
       made these changes to branch accounts 'without the
       subpostmaster knowing'."
           So that's Dr Worden's conclusion on that, that's
       7 December and then if we -- your Lordship will remember
       that Mr Roll's second statement explained how those were
       to be made and in response to that Mr Parker serves
       a witness statement confirming that Mr Roll is right and
       we see that at {E2/12/9} which is paragraph 27, where he
           "In paragraph 20 of Roll 2 [which we will look at in
       a second] Mr Roll describes a process by which
       transactions could be inserted via individual branch
       counters by using the correspondence server to
       piggy-back through the gateway.  He has not previously
       made this point clear.  Now that he has, following
       a discussion with colleagues who performed such actions
       I can confirm that this was possible.  I did not mention
       this in my first witness statement because, when faced
       with a less clear account in Mr Roll's first statement,
       my recollection was that if it was necessary for the SSC
       to inject a transaction data into a branch's accounts,
       it would have been injected into the correspondence
       server (injecting via the server was the default option
       which was followed in the vast majority of cases)."
   MR JUSTICE FRASER:  So apparently it is all Mr Roll's fault.
   MR GREEN:  Apparently.
   MR JUSTICE FRASER:  Well, I dare say that will be pursued in
       due course.
   MR GREEN:  Indeed.  And the final reference, just so
       your Lordship has a reasonably complete picture about
       this, is {D3/6/21} which is Dr Worden's second report,
       which postdates Mr Parker's second statement, at
       paragraphs 83 to 85.  If we go to {D3/6/21} and if we
       look at 83 to 85:
           "It seems to me that I require further factual
       information before I can comment on this evidence.
       Which 'specific person'?  Under what circumstances?  How
       frequently?  Until I have that information, it remains
       possible in my view that any transaction which 'would
       appear to the subpostmaster as though it had been
       carried out through the counter in branch' might only be
       a transaction that he had given his consent for, as the
       'specific person' - and which had in effect been made on
       his behalf.
           "Therefore, Mr Roll's new evidence does not cause me
       to alter the opinion expressed at paragraph 1119 of my
       first report, when commenting on Mr Roll's first witness
       statement, that he could not alter branch accounts
       without the subpostmaster knowing."
           "In paragraphs 27-34, Mr Parker provides detailed
       and specific commentary on Mr Roll's paragraph 20, using
       his knowledge and the appropriate contemporary
       documents, where they have been found.  Here he
       acknowledges that Fujitsu could insert transactions into
       branches by a piggy-back process.  I am not yet able to
       comment on Mr Parker's evidence or the documents he
           So that's the manner in which Dr Worden engaged with
       the picture that emerged in that way through those
       witness statements.
           So, my Lord, that's the background to those big
       issues and the sort of key features of them.  I don't
       know whether it is worth taking a very slightly earlier
   MR JUSTICE FRASER:  I would maybe go on a bit.
   MR GREEN:  Shall I push on a bit?
   MR GREEN:  Because I'm about to begin just tracing through
       the Dalmellington bug to see how that arises.
           The Dalmellington bug is a shorthand name for
       a particular bug and it takes its name from an iteration
       of the bug which occurred in October 2015 and it is in
       the Horizon Online system, particularly, but probably
       not exclusively, affecting transfers of cash between
       core branches and outreach branches, which are small
       part-time branches that use mobile equipment.  Cash is
       sent from Post Office to a core branch and then it is
       transferred from the core branch to an outreach branch.
       When the barcode for the cash was scanned at the
       outreach branch, the number of remittances would
       multiply so that Horizon showed far more cash on the
       system than was physically present.  And just so
       your Lordship knows, or has an idea of how it seemed to
       the postmistress Ann Ireland who was affected in
       Dalmellington, what in fact appears to have happened is
       that at the remming in stage the post mistress was
       presented with the "Enter" button, she presses it, that
       should take her on to a new screen, but it stays there
       and it is pressed again and again, and it appears that
       there is a reasonable inference that the multiple of
       cash reflects the number of times -- that looks to be
       how it works.  So she ends up with trying to rem in
       an £8,000 cash remittance into the outreach branch and
       actually the Horizon system recording her account as
       having remmed in 32,000.  So four enters effectively, so
       she is over by £24,000.
           My Lord, we say that this is quite an interesting
       bug to look at because of the precise way in which it
       arose because unusually, in a lot of other cases the
       postmistress or postmaster is trying to work out what's
       happened, without knowing, but in this case they are on
       both ends of the transaction because they see a receipt
       that they have printed out going out and they are at the
       outreach branch and they see what they get at the
       outreach branch and they can literally hold them up and
       go "That cannot possibly be right".  So this is a good
       example of a very particular bug which was at literally
       one end of the spectrum of visibility to subpostmasters,
       in contrast to a number of others, and we will see that
       in the documents that we go through.
           My Lord, can I begin by taking your Lordship to
       {F/1389} which is the PEAK.  I say "the PEAK" and I will
       explain in a little more detail in a minute.  So this is
       PEAK PC0246949 which for convenience I will refer to as
       the 949 PEAK.  And just to show your Lordship how the
       PEAK document works -- it may be absolutely apparent
       already -- "Call type", "Live incidents/defects" and
       then there's a "Priority", priority C, "Non-critical".
       And then there's a summary "Horizon -- transaction
       discrepancies", as we come down, and then "All
       references" your Lordship will see as a heading, "Type"
       and "Value" and there under SSCKEL, that is system
       support centre known error log, your Lordship can see
       the acha621P KEL.  Just as a practical matter, in terms
       of knowing what a KEL is about, the way the KEL system
       was set up, the KELs are named after the person who does
       them, not the issue.  So you can't tell anything about
       the KEL from the name.
           Then your Lordship then sees "Clone call", PC0246997
       which for convenience I will refer to as the 997 PEAK.
       So that's an associated PEAK.
           Then just following this document through, the
       progress narrative has a very brief summary of the
       fuller data above, so you can see "Call type: L", means
       "Live", "Call priority: C", cross-references to
       non-critical and so forth.  And we then come down.  The
       date of the record in the sort of yellowy-beige section
       is 13 October at 2.46 in the afternoon, 14.46.
   MR JUSTICE FRASER:  Where are you looking?
   MR GREEN:  Just halfway down the page, my Lord, when it goes
       from a duck egg blue colour to a yellowy colour.
   MR JUSTICE FRASER:  I see, but date time is the day before
       at 5.10.
   MR GREEN:  Exactly.
           So the record there in yellow is 13 October and the
       transfer note begins:
           "Please can PEAK investigate this discrepancy issue.
       NBSC has confirmed that following discussions and checks
       with the user that this is not a user error issue ..."
           So confirmed not a user error issue, which in this
       particular case was very easy for reasons I will show.
           "... but an issue within the system requiring
       Fujitsu investigation."
           And then there's the message:
           "Hi Eden, need to raise an incident for below issue
       and ... provide Fujitsu re ... it's been confirmed with
           Then you get the same text effectively again about
       confirming that it is not a user error.  And if
       your Lordship comes down to just under
           "User has discrepancies when transferring cash from
       one branch to another (specifically between their main
       branch to their outreach branch); OUTREACH BRANCH ISSUE.
           "User said instead of the system logging it as
       £8,000 transaction, it recognises it as £32,000
           "User already contacted NBSC and was rightaway
       directed to us ..."
           And the amounts are logged at the bottom: star
       £8,000 over £32,000.  I think the amount should be
       actually over 24,000 but I think they are recording what
       it actually did is 32,000.
   MR JUSTICE FRASER:  Well, it says it should have been £8,000
       but it has been recorded as 32,000.
   MR GREEN:  Exactly.
           And the subsequent pages of the PEAK record the
       dealings with the PEAK in a bit more detail.  I will
       come back to some more specific aspects of that.  If we
       can just go back to the first page for a second
       {F/1389/1} and we look at "Clone call" and the 997 PEAK,
       if we go to {F/1389.1} this is the 997 PEAK, which
       appears to be a PEAK cloned from the other one.  So
       "Release PEAK", there's a reference to another PEAK
       there and the clone master is the 949 PEAK that we have
       just looked at.  So this is a clone of the master PEAK
       and it refers to another release PEAK and it also refers
       to the KEL, the acha621P KEL which is the same KEL
       referred to in the previous one.  And your Lordship will
       see under "Progress narrative", this 997 PEAK is opened
       on 14 October by someone called Anne Chambers who we see
       quite a bit and then it appears to paste in what's
       happened on the 13th below it, so it's not actually in
       chronological order going downwards.
           And we get a bit more information there in the
       middle set out quite helpfully:
           "Branch has a discrepancy of £24,000.  4 other
       branches have had a similar problem in the last
       2 months."
           And your Lordship will see in a minute that she is
       unable to go back more than two months at that stage to
       see what else has happened.
           "3 of these resolved by remming out the excess.
       Since these were all branch to branch rems and there is
       no cross-branch accounting within Horizon this removes
       the discrepancy."
           And then if we can then go to {F/1393}, this is
       PEAK 207, which is just for your Lordship's note not the
       other release PEAK that was mentioned in the previous
       document, because that ends 024, it is a third one.  And
       here we can see halfway down that the date is 21 October
       we're in now and there is email correspondence involving
       Mr Parker as the originator -- your Lordship will see
       "Originator" about halfway down the page, just below
       where it goes from sort of green to duck egg green or
       duck egg blue, 21 October 2015, 13.35.01.  Four lines
           "Originator: Parker Steve ..."
           Which we understand to be Mr Stephen Parker who is
       giving evidence in this trial.  And we can see that
       there is correspondence from Tony Wicks to Anne Chambers
       about Anne Chambers having done most of the
       investigation into this.  If we look at the bottom we
       see Tony Wicks on 20 October 2015 at 15.25 to
       Sandie Bothick saying:
           "Hi Sandy,
           "Looking at PEAK [949] ..."
           Which --
   MR JUSTICE FRASER:  That's the one we started with.
   MR GREEN:  Exactly, so that's effectively the master -- it
       is treated as the clone master PEAK:
           "Looking at PEAK [949] it appears to be derived from
       I7991774 and I found TfS incident ..."
           And there is a number there:
           "There is no problem record raised for this, however
       PEAK [997] was used by development to investigate this.
       A code fix has been developed, but requires official
       testing and releasing.  I've made enquiries and
       unfortunately LST are unable to take the fix for testing
       and release 12.88 without significantly impacted that
       release to live.
           "As the condition can be avoided by postmasters,
       ie by making them aware of the condition and advising
       them not to press enter multiple times, I propose that
       this is KELed and included in the counter release
           And just a bit further down {F/1393/2},
       Sandie Bothick is replying.  She is saying:
           "Hi POA DM ..."
           Duty Manager:
           "Have PEAK make you aware of this issue?
           "Do you have a PR open ..."
           And so forth.  And then below is what she sent to
       Atos earlier:
           "Hi Katie, I'm coming in blind on this.  Looking at
       the incident this is our update from PEAK TfS
       connector ..."
           And so forth.  Then there are further references to
       the 997 PEAK:
           "We are continuing to investigate the problem ..."
           Just over halfway down:
           "... but any fix will not retrospectively change the
       branch accounts."
   MR GREEN:  "So we are aware of the issue and are continuing
       to investigate but NBSC should be able to sort the
       discrepancy out in the meantime."
   MR JUSTICE FRASER:  I think if you are going to move on to
       another document I think we should take a break.
   MR GREEN:  Shall we take a break there?
   MR JUSTICE FRASER:  Yes.  Come back again at 12.
   (11.54 am)
                          (Short Break)
   (12.00 pm)
   MR GREEN:  My Lord, can I just highlight a couple of
       remaining points in this PEAK.  If we can begin at
       {F/1393/4} please.
   MR JUSTICE FRASER:  This is still in PEAK 207, is it?
   MR GREEN:  It is still in PEAK 207, exactly.
           Now, your Lordship will see at the bottom of that
       page, Monday October 19th, 2015 at 5.33 pm.
   MR GREEN:  "IT-Solutions R SPM Post Office incident
           "Hi Ibrahim, as this incident is not getting
       resolved can we have a concall set up between NBSC &
           "The site had transaction discrepancy.  As per
       Fujitsu, they have found 4 other instances (outreach
       branches ..."
           It gives the name:
           "... and all but the last removed the discrepancy by
       completing a rem out for the excess ..."
           Then if we go back to the previous page {F/1393/4},
       there is an email which is on 20 October 2015 at 9.53 at
       the top of the page, part of it, and if we come down
       four paragraphs -- well, let's maybe take it from the
       second line:
           "She was concerned as she had never seen this
       before.  She balanced core and it was correct, but
       outreach was £24,000 short.
           "Although the core had sent only one lot of £8,000,
       the outreach had accepted 4 lots of £8,000 in one
           "She has spoken to NBSC ref 1358666 who told her it
       was a technical issue.
           "She then phoned the IT help desk ref I7972295.  She
       was unconvinced they understood the problem although
       they said they could probably 'rectify remotely'.  After
       waiting till the end of day she called back and
       escalated to option 7 and spoke to Rich who told her to
       phone NBSC.
           "I don't think the helpline understood what's
       happened.  I can understand that as you would think it
       is not possible.  But incredibly Anne's outreach Horizon
       now shows £24,000 short and it does not exist.  As you
       can imagine, Anne is concerned and I have told her not
       to touch the outreach unit until this is resolved for
           "The incident was passed to Fujitsu who have advised
       that in order to resolve the issue the branch/NBSC must
       'complete a rem out for the excess to correct the cash
       holding' which Fujitsu are unable to do.  The NBSC has
       subsequently advised that they cannot assist as this is
       an IT issue, however Fujitsu are also advising that they
       cannot assist.  As a result, the issue has been passed
       back and forward for over a week."
           Then if we go back a page {F/1393/3} to 20 October
       at 10.42, Kendra Dickinson at the Post Office says:
           "Could I enlist your help and support on the below
       issue please?"
   MR JUSTICE FRASER:  This is underneath "Hi Rod/Dawn"?
   MR GREEN:  Exactly.
           "Whilst I am happy for NBSC to try and support where
       they can, the concern I have with the below is that we
       have no process for managing this type of issue and we
       are unable to see any of the back-end accounting for
       this branch.  Therefore, any advice that we try and
       provide could end up making matters worse - this is
       already showing a 24K loss.  I am not happy for NBSC to
       give advice on something that is not a process that
       exists within the knowledge base.
           "Similar to a disconnected session, NBSC would have
       no understanding as to the implications on branch
       accounting if they were to advise the branch as
       suggested below."
           And then if we go up there, there is more about the
       difficulty of NBSC giving help and advice with
       insufficient information, top of that page, which is
       under Tuesday 20 October at 11.57:
           "... an explanation of the root cause to be supplied
       by Fujitsu via Atos so that both our Finance Service
       Centre and NBSC colleagues can be assured that the right
       advice is given, there is no impact to the branch
       account and a full audit trail is available.  It does
       not feel right for Atos and Fujitsu to be giving
       instruction to NBSC to speak to branches with advice
       with insufficient information.
           "If this has happened in this case it would be
       useful to see that in this email trail.
           Then if we go back a page to {F/1393/2}, halfway
       down the page -- this is in the Sandie Bothick email
       that appears to start at the top and appears to include
       what she sent to Katie, who I think I referred
       your Lordship to already.  If we come halfway down does
       your Lordship see "POA-Horizon".
   MR JUSTICE FRASER:  No.  Oh yes, I see it.
   MR GREEN:  Just there and then you have "Provider ref" and
       a reference to the 949 PEAK.
   MR JUSTICE FRASER:  This is the one we have seen already.
   MR GREEN:  Indeed.  It says:
           "Resolution details: update by Anne Chambers:
       category 70 - final - avoidance action supplied: we have
       found that if there is a log-out before a user has fully
       logged on, then subsequently a pouch is remmed in
       manually (most likely at an outreach branch), then after
       the rem in slip has been printed, the same screen is
       redisplayed and the user is likely to press enter again
       and duplicate the remittance, possibly several times.
       A different screen should be displayed which would
       prevent this happening.
           "A rem in slip is printed each time, showing the
       same details but different session numbers, and
       a transaction log search confirms the repeated rems."
           And then importantly, we would respectfully say, she
           "This is not an area that has changed for several
       years so it is likely to have happened before but we
       have no record of it having been reported to us.  I can
       only check back two months; I've found 4 other instances
       [of the outreach branches] ... and all but the last
       removed the discrepancy by completing a rem out for the
       excess ..."
   MR JUSTICE FRASER:  Those four must be in the last two
       months, mustn't they?
   MR GREEN:  Precisely.  So pausing there --
   MR JUSTICE FRASER:  On the face of the document they must be
       the last two months.
   MR GREEN:  Exactly.  So pausing there, your Lordship will be
       beginning to get a feel for the fact that it is quite
       difficult to identify with confidence from a particular
       KEL that you have necessarily found any related KELs, or
       from a particular PEAK that you have necessarily
       identified all of the PEAKs that relate to that problem
       and there is obviously some information flow
       practicalities, both in relation to what information
       flows to the affected subpostmaster and whether there's
       a clear process for that, and also internally about the
       reporting of these issues by Post Office through its
       processes back to Fujitsu because the reporting of those
       issues is obviously essential to one of Dr Worden's
       countermeasures in terms of fixes being applied to
       correct the system errors.
           From there can I invite your Lordship just to see
       {F/1389} and {F/1389} is -- that's 949 that we have
       identified as the PEAK we have looked at and then when
       we look at the reference to the additional receipts
       being produced we need to look at a couple of
       documents --
   MR JUSTICE FRASER:  Do you mean on the same PEAK?
   MR GREEN:  Yes, this is still -- this is all cross-referring
       to the same ones.  We've got {F/1386}, and, my Lord,
       this makes good the advantage to a subpostmaster --
       "advantage" is a relative term, but compared to some
       other software issues.  By being on both ends of the
       transaction she is able to produce what has gone out and
       in.  So we've got the remming in money which should be
       £8,000 and you've got on the right-hand side four
       separate remming ins of £8,000, all effectively in the
       same -- within one minute of each other, those four
       separate receipts coming down the right-hand side and
       one slightly on the left, numbered 1 to 4.
           If we then go to 1392 we've got the events log which
       the subpostmistress was able to see showing rem out slip
       on the left-hand side where she has written "Rem out"
       and then on the right-hand side she has identified
       multiple receipts being printed which are the four
       receipts we have just looked at.  So on this particular
       bug this subpostmistress at Dalmellington had both ends
       of the data, as it were, and was able to say "Look, this
       cannot possibly be right."
   MR GREEN:  And that is obviously reflected in the
       recognition internally within Post Office of what's then
           Then if we just briefly go back, your Lordship will
       notice that the date of the event itself is 8 October
       and then we get the record created at 12th and notes at
       13th and then the later ones where they are arguing
       about the fact it has been bounced back between helpline
       and Fujitsu are dated the 20th, just to get the timeline
       clear in that respect.
           Can I take your Lordship now to the underlying KEL
       which is at {F/1426}.  So this was a KEL raised by
       Anne Chambers on 15 October 2015, so on the face of it
       looks like prior to that date the acha6 -- it is
       difficult to see, prior to that date of 15 October 2015,
       to what extent there was an existing KEL identifying
       this on the face of the documents.
   MR GREEN:  And the symptoms are described:
           "A cash pouch was received at an outreach branch and
       scanned into Horizon.  The manual process was followed
       and 2 Delivery Receipts printed.  Then the clerk pressed
       Enter to complete the process, and a Rem In slip was
       printed.  They were then able to press Enter again and
       another Rem In slip was printed - and the same amount of
       cash was recorded a second time.  They may have repeated
       several times before using Cancel to escape, resulting
       in much more cash being recorded on the system than they
       actually have."
           And "Solution - ATOS" at the bottom -- they refer to
       the problem they have identified which is a problem
       your Lordship has already seen.  "Solution":
           "Known problem should now be fixed so any further
       occurrences need to be investigated - send call to PEAK.
           "Outreach branches can avoid the problem by making
       sure that they do not press Enter again after they have
       printed both Delivery Receipts and the Rem In slip - if
       they find the Rem In screen is still displayed they
       should press Cancel to get out of it, ignoring the
       message that not everything has been printed."
           So actually what's reflected there, my Lord, just
       pausing, is that what's shown on the screen is actually
       positively misleading and that is in fairness to some
       extent reflected in the next sentence on {F/1426/2},
       where they say:
           "However, they are unlikely to notice immediately
       that they are on the wrong screen, and will probably
       have duplicated the rem in before realising something is
           Because if you successfully press the enter button
       it's meant to move on to the next screen, so when it
       doesn't go away they just keep pressing it.
   MR JUSTICE FRASER:  That says in the third paragraph:
           "The cause of the problem is being investigated ..."
   MR GREEN:  Indeed.
   MR JUSTICE FRASER:  And can we go back to the previous page,
       page 1 {F/1426/1}:
           "Known problem should now be fixed ..."
           What does "known problem" mean?
   MR GREEN:  My Lord, this is one of the points -- there are
       two tasks I'm trying to do, which is to try to
       interrogate the documents as helpfully as I can for
       the court, but at the same time try to identify to
       the court the difficulties that one is -- anyone is
       faced with when looking at the incomplete picture that
       the experts have referred to.
   MR JUSTICE FRASER:  All right, understood.  Let's go on to
       page 2 again please {F/1426/2}.
   MR GREEN:  And your Lordship will see again there:
           "... but it will not retrospectively correct the
       accounts at affected branches."
           So that's noted again there.
           Now, what in fact happened was there was external
       third party interest in this problem and that resulted
       in a Fujitsu presentation on 10 December 2015.  To be
       fair to Post Office, Post Office wanted answers but
       wanted an identification of root cause and there's also
       some pressure from a blogger at the time which we will
       see later, and at {F/1415}, there is a Fujitsu
       presentation on 10 December 2015 on the branch outreach
       issue.  "Initial findings", the first page confirms it
       is for Post Office's internal purposes only as
       confidential.  If we go over the page {F/1415/2} it
       confirms there are potentially two separate issues at
       play.  It:
           "Doesn't correctly close down on the post log on
       script.  This leaves the script on the 'stack' of
       incomplete processes."
           And also:
           "The pouch delivery script thinks it has completed
       doesn't explicitly finish".
           And there's some consideration of that.
           If we look at page 3 {F/1415/3}, as
       at December 2015, the date of this presentation,
       Post Office knew from Fujitsu that there had been 112
       occurrences that Fujitsu had identified of duplicate
       pouch IDs over the past five years, three issues with
       duplicate pouches, 47 outreach cases, 19 using the
       "Previous" button after a pouch scan, which was said to
       have been fixed March 2010, 46 remitted at multiple
       counters, fixed January 2011, 108 items corrected at the
       time either by transaction correction or subpostmaster
       referral, four items still to be confirmed, no items
       related to the period where the mediation scheme --
   MR JUSTICE FRASER:  What does BLE stand for at the top?
   MR GREEN:  I'm not --
   MR JUSTICE FRASER:  If you don't know it doesn't matter.
   MR GREEN:  I'm not sure I know, my Lord.  I was going to
       hazard a guess but I'm not sure I'm right.  Maybe we can
       check that.
           And if we go forward to page 7 {F/1415/7} of that
       document please, in 2010/2011 there were 65 incidents of
       a slightly different iteration where the "Previous"
       button -- if you used the "Previous" key during or just
       after the pouch barcode scans you get a multiple remming
       in problem and:
           "No more occurrences found post fix."
           And then 46 remittances at two counters, that's the
       same pouch being remitted at the same branch at more
       than one counter.  46TCs reviewed and applied by
       Post Office and then a fix applied in January 2011, "No
       more occurrences found post fix", fully aware of both at
       the time.
   MR JUSTICE FRASER:  And are these -- I think you said
       previous iterations?
   MR GREEN:  Previous iterations of related but not identical
       code issues that led to multiple remming in.
   MR JUSTICE FRASER:  Because they are not all under the same
       PEAK number, are they?
   MR GREEN:  They're not.  So tracing exactly where it is
       found is not absolutely straightforward.
   MR JUSTICE FRASER:  Because part of the document you showed
       me before was the people involved trying to find out if
       it there was an existing PEAK number to which it
   MR GREEN:  Precisely.
           Then if we go over to page 8 {F/1415/8}, this goes
       to the flow of information from Post Office to Fujitsu
       in order to ensure timely corrections of the software.
       One transaction correction completed by Post Office,
       five remittance transactions completed by PMs, zero
       calls raised with Fujitsu.  Nine incidents in 2012, zero
       calls with Fujitsu.  Seven incidents in 2013, zero calls
       with Fujitsu.  Two unknown outcomes noted in 2013,
       £25,000 and 2,500.  And those, until I think it was
       yesterday or the day before, we didn't have clarity on
       what had happened on those.  There is now greater
       clarity as a result of some additional documents that
       we've got from Post Office.
           The further "Detailed preliminary findings" on
       page 9 {F/1415/9}, identify 2014, nine incidents, no
       calls and then 2015 to 2016, one call raised with
       Fujitsu, which is effectively the call which I think is
       Ann Ireland's call which I think ultimately goes
       through.  Then "January 2016 fix to be applied".
           So that gives your Lordship a feel for how in this
       particular example the information flow appears to have
       worked.  There had been a five-year period over which
       double remming system errors had been detected, there
       had been some fixes of aspects of that, but it's
       uncertain -- it's difficult to analyse precisely the
       genesis of all of those incidents.
           If we now go to {F/1399}, we can see that
       Computer Weekly was running an article about the problem
       that the CWU was looking into and if we look on page 3
       of that document {F/1399/3}, we can see at the top of
       the page:
           "The alleged problem investigated by the CWU
       involves the process where subpostmasters transfer money
       from a core Post Office branch to a remote branch
       created to serve rural areas, known as an outreach,
       which is basically a branch on a laptop.  These
       processes are known as remittances."
           And so forth.  So "Post Office department recognises
       duplication problem" at the bottom:
           "Following a post on a web forum, another postmaster
       recognised the problem as something she had experienced
       in the past, and the financial department of the
       Post Office in Chesterfield was contacted on her advice.
           "According to the postmasters' branch of the CWU,
       the financial department recognised the issue and told
       the postmaster they would need to send a positive
       monetary change to the Horizon branch accounts ..."
           So that was following a post on a web forum.  So
       although it is clear that there are cases that have been
       identified and have been corrected, precisely how that
       has happened may vary from case to case.
           At the bottom:
           "The alleged fault, if proved, would question the
       Post Office's continued claims that there are no
       systemic faults in Horizon."
           So that was what was happening in November.  There
       is a follow-up article at {F/1405} which is
       a Computer Weekly article, on 18 November 2015.
   MR JUSTICE FRASER:  I'm not sure we are there yet.  Where do
       you want to go now?
   MR GREEN:  I'm not going to take you to the detail of it.
       Just so your Lordship knows, there is a follow-up to the
       point --
   MR JUSTICE FRASER:  Which is at F/1405?
   MR GREEN:  This is at F/1405.
           So just stepping back from this, this bug is
       significant in the sense that not only was it one that
       was at the top end of the visibility for proof for the
       subpostmaster, who actually got both ends of the
       receipts, but there is also public pressure being
       applied in the media specifically on the issue
       in November and then we have actually got the Fujitsu
       presentation about it, so there's an unusually rich
       source of information available to the claimants and
       the court in respect of that.
           There are then internal Post Office emails at
       {F/1495.1/5} and the first four pages of this are
       redacted but it is an email from Paula Vennells, the
       chief executive, citing a blog post by Mr McCormack, who
       was a fairly energetic blogger on this topic and
       I haven't taken your Lordship to his blogs but they were
       happening in parallel, and she says:
           "Dear both,
           "This needs looking into please ... can I have
       a report that takes the points in order and explains
           "Tim McCormack is campaigning against PO and
       Horizon.  I had another note from him this am which Tom
       will forward, so you are both in the loop.
           "We must take him seriously and professionally.
           "This particular blog is independent of Sparrow but
       clearly related in that it appears to present similar
       challenges that were raised in the course of the scheme.
           "I'm most concerned that we/our suppliers appear to
       be very lax at handling 24k."
           Pausing there, that appears to be a reference to the
       triple sum rather than to any of the outstanding cases
       they haven't traced because they were 2,500 and 25,000.
           And response on the same day at {F/1495.2/1} from
       a Rob Houghton who was head of IT at Post Office, so we
       there get:
           "I need an urgent review and mini <taskforce> on
       this one."
   MR JUSTICE FRASER:  Where are you reading?
   MR GREEN:  Sorry, immediately under Rob Houghton's response
       at the top, third of the way down, copied in to Angela
       van den Bogerd, "The Dalmellington error in
       Horizon/problems with POL" which was the news article.
           "I need an urgent review and mini <taskforce> on
       this one.  It probably needs to link up heavily with
       Angela's work as FSC are mentioned extensively ..."
           This is 1 July 2016, my Lord:
           "... - Angela cfi."
           Copied for information presumably:
           "I don't know how we respond to this but can we
       section a few inside people to get all over it and give
       me/Al/Paula evidence and understanding."
           And then at the top of the page the response at
       11.59, which is difficult to fit into the chronology,
       appears to be on the same day, Friday 1 July:
           "Can you stand down on this please?"
           "Any specific actions and I will revert.
           "My apologies."
           Pausing there, the claimants' letter of claim was
       sent on 28 April 2016 and the Post Office sent its
       letter of response on 28 July 2016 so nearly four weeks
       after the decision to stand down on the investigation of
       that bug.  So as at the date that the Post Office was
       responding in its letter of response, it had front and
       centre in its mind the existence of this bug.
           Now, that bug was not disclosed in the letter of
       response.  The bugs that were disclosed in the letter of
       response were the three that Second Sight had already
       found and Post Office letter of response, if we can just
       turn to {H/2/95}, stated --
   MR JUSTICE FRASER:  That's the letter of response, is that
   MR GREEN:  Yes.
   MR JUSTICE FRASER:  This is the schedule 6 rebuttal, is that
   MR GREEN:  Precisely.
           This is where the Post Office obviously rebuts the
       allegations that have been made in the letter of claim
       and at 1.5:
           "Second Sight only reported on a number of already
       resolved defects in Horizon (which they called 'bugs').
       Second Sight did not discover these defects through its
       investigations.  These were issues already known to and
       remedied by Post Office.  It was Post Office that
       disclosed them to Second Sight."
           So those are the three.  Then 1.6 deals with the
       lack of attribution of any shortfall in the scheme to
       Horizon.  Then 1.7:
           "The letter of claim also presents no evidence that
       a defect in Horizon has caused a postmaster to be held
       wrongfully liable for any shortfall in their branch.
           "Nevertheless, you make repeated references to the
       existence of historic defects in Horizon in order to
       give a false impression that Horizon deeply suffers from
       major defects, that Post Office does nothing about them
       and that these errors have caused postmasters losses
       which have gone unremedied.  In order to dispel any
       myths around the defects reported on by Second Sight and
       cited by other sources, we have set out below in detail
       what happened in these instances.  To be clear -
       Post Office does not claim that these have been the only
       defects in Horizon."
           So what they do do is they respond to the three that
       were known about, but they don't give any account of the
       Dalmellington bug or of its true extent which they well
       knew about from the Fujitsu presentation -- sorry,
       Dalmellington related issue on multiple remmings,
       because there are a number of different PEAKs and
       different software code problems that relate to multiple
           So if we just look over the page, just
       parenthetically, my Lord, if we may, at {H/2/96}.  In
       relation to the Callendar Square, Falkirk issue, that is
       said to be an issue that only affects one branch and it
       is ruled out in the -- at the bottom:
           "... also raised as part of a defence in a civil
       action by Post Office against a former postmaster,
       Lee Castleton ... The Court accepted the evidence from
       Fujitsu’s witness, Anne Chambers, and found 'no
       evidence' of the Falkirk bug in Mr Castleton’s branch."
           We now know from Mr Godeseth's witness statement of
       16 November that this bug affected 30 branches and
       resulted in mismatches at 20 branches.
           So it's not just in the letter of response that they
       don't deal with Dalmellington, but it's not a full
       account perhaps of what was being said.
   MR JUSTICE FRASER:  The Callendar Square one affected
       30 branches.
   MR GREEN:  Indeed.
   MR JUSTICE FRASER:  That's different to the Dalmellington
   MR GREEN:  Completely separate.
   MR JUSTICE FRASER:  Is there a list of those 30 branches
   MR GREEN:  I think we've got -- I will trace where it is.
   MR JUSTICE FRASER:  Just give me the reference in due
   MR GREEN:  Of course.
           Coming back to Dalmellington, so what in fact
       happens is it is not disclosed in the letter of response
       and then the experts are then provided essentially with
       8,000 KELs and about 220,000 PEAKs.  My Lord, just for
       your Lordship's note, in the appendix we have dealt with
       how we obtained the KELs and the genesis of that
       disclosure at {A/1/12-14} and then we have set it out in
       greater detail in the appendix.
   MR JUSTICE FRASER:  Which is at 106 I think.
   MR GREEN:  Indeed.
           So what in fact happens is there is the --
       eventually we get the KELs.
   MR JUSTICE FRASER:  I made an order I think, didn't I?
   MR GREEN:  Your Lordship made an order, exactly.  And they
       had first been -- the KELs had been requested in the
       letter of claim and your Lordship made observations
       about the response to that earlier.
           What in fact happens is Mr Coyne is the person who
       discovers the existence of the Dalmellington bug in his
       first report.  If we look at that, it's {D2/1/58}.  And
       he actually locates it -- it is paragraph 5.16 -- and
       your Lordship will see at paragraph 5.16 he explains
       what's happened and footnote 46 is the email thread and
       then he also finds the duplicated branch receipts and
       then if we can just go to page 60 of that document --
   MR JUSTICE FRASER:  Page 60 of which document?
   MR GREEN:  Of Mr Coyne's first report {D2/1/60}.
       Your Lordship will see that although he does refer to
       a KEL at 5.23:
           "... evidence of cash declaration discrepancies
       arising from clerks duplicating remittance in
       transactions ... because of wrong messages being
       presented on the Horizon counter screen ..."
   MR GREEN:  He has not at that stage worked out that this is
       the Dalmellington bug.  So it just gives your Lordship
       an insight into how we had to discover what was done.
           Then Mr Parker and Mr Godeseth's evidence is
       significant because the bug was not dealt with in the
       first round of the Post Office's evidence at all and
       only after it was referenced by Mr Coyne was it dealt
       with in the first statement of Mr Parker and the second
       of Mr Godeseth and Mr Parker confirmed the link between
       the KEL and Dalmellington and says temporary financial
       impact which was rectified by a transaction correction
       being issued.  Mr Godeseth then discloses the 112
       instances of duplicate bar codes and then appends the
       analysis that I have already mentioned to the court.
           Then Dr Worden's initial analysis, which is
       7 December 2018, does not reference Dalmellington, so
       Post Office appear not to have told him about it, or he
       has not dealt with it, we don't know which it is, and
       despite the fact that it had actually been referenced by
       Mr Coyne in Coyne 1 and he is responding to it, and
       referenced in Godeseth 2 and Parker 1, which are
       statements which Dr Worden -- I'm sorry, I was corrected
       by my learned friend earlier, I have forgotten to say
       that: my Lord, Dr Worden pronounces his name Dr Worden,
       I have been pronouncing it wrongly.
           Despite the fact that Dr Worden has referred to the
       defendant's witness statements on other matters.  He did
       mention the KEL and he regarded it as having --
   MR JUSTICE FRASER:  Are you talking about Worden 2 now or
       Worden 1?
   MR GREEN:  This is Worden 1.  Then Mr Coyne, in Coyne 2,
       identifies further PEAKs on his own without any
       assistance from Post Office and sets those out -- no
       need to turn it up, I will just give the references
       {D2/4/105} and the PEAKs concerned are the 207 PEAK
       which we have looked at which is {F/1393}, and 949 PEAK,
   MR JUSTICE FRASER:  Sorry, can you give me the first one
   MR GREEN:  Sorry, the 207 PEAK is {F/1393} and the 949 PEAK
       is {F/1389}.  Then Mr Coyne then attempts to look for
       PEAKs with FAD codes of those branches whose situations
       are unresolved and can't find anything.  There are four
       unresolved of which two are a significant sum and he
       mentions that at Coyne 2, paragraph 4.53, we don't need
       to go there but it is {D2/4/107}.
           Then in Dr Worden's second report he does refer to
       Dalmellington, explaining he didn't realise that the KEL
       he was looking at related to Dalmellington and he
       concludes that there was no permanent error due to the
       branch account because it was resolved by a TC, a PEAK
       and KEL were created, so the system worked well and it
       was fixed, and his opinion is re-enforced by the small
       maximum financial impact that he has estimated of
       incorrect TCs.  He doesn't actually expressly mention
       the 112 occurrences, the 87 other branches affected or
       there that it lay undetected for five years, at least
       for some parts of that aspects of it appear to have been
       corrected by various fixes, the lack of calls put into
       Fujitsu or actually directly address the four branches
       where there is no outcome confirmed.
           So stepping back, your Lordship will now have
       a picture of the task and the information available, why
       the experts are agreed that the KELs and PEAKs have
       relevant information in them but not a complete picture,
       and how this particular bug was addressed.
           Now, the postscript to this is that four hours and
       45 minutes after our written opening had been filed with
       the court, so at 4.45 on 4 March, PO disclosed for the
       first time 36 additional documents.
           If we look at {H/232} --
   MR JUSTICE FRASER:  Are these documents other than the ones
       you mention in one of your paragraphs as being
       outstanding that Mr Coyne had asked for?
   MR GREEN:  My Lord, they are -- well, I think they overlap
       because they relate to the Dalmellington bug, some of
   MR JUSTICE FRASER:  So 36 new documents.  Where are we going
   MR GREEN:  This is {H/232}, so this is the letter on
       4 March, 4.45 in the afternoon, so after written
       openings had been filed with the court and it says:
           "During preparation for trial a number of further
       documents which relate to the Horizon issues trial have
       come to Post Office's attention.
           "Disclosure of these documents is being provided by
       way of enclosed disclosure list.  The enclosed list also
       contains those documents which have been provided to the
       claimants but not yet formally disclosed."
           My Lord, I'm not going to deal with the contents of
       those documents now, but I am going to have to take
       Dr Worden to them and some of the witnesses to them, but
       in fact what seems to have happened, there's some sort
       of investigation that has been done in relation to the
       two outstanding branches that we said were unremedied in
       our opening and a conclusion is reached that it is
       probably all fine, if I can put it in fairly neutral
       terms at the moment.
           There are other bugs which we are obviously
       concerned about whether we got the full picture on but
       that will emerge in the evidence.
           The background, my Lord, if I can just very quickly
       take your Lordship through it.  There is correspondence
       which covers the disclosed documents here and the
       Dalmellington bug, which starts on 2 October 2018 at
       {H/122} in which we refer to the bug or error, we refer
       to bugs which are in addition to those acknowledged in
       the letter of response and ask for all documents
       relating to these bugs and errors.
   MR GREEN:  So that's when it starts.  My Lord, I'm not going
       to take your Lordship through all the correspondence.
       Can I just give you the references.
   MR GREEN:  There's a chaser on 13 December 2018 {H/147},
       there's the 11 January 2018 letter from Wombles at
       {H/165} offering voluntary disclosure of documents,
       providing 162 documents for Dalmellington; there's the
       22 January 2019 letter at {H/173/6} referring to the
       four unknown branches that I have just been talking
   MR JUSTICE FRASER:  That's a letter from your solicitors?
   MR GREEN:  From my solicitors.
           1 February, we've got Mr Coyne's second witness
       statement at 4.51 to 4.53 {D2/106}, 11 February, some
       disclosure, but 4 March these actual documents relating
       to those outstanding branches.
   MR JUSTICE FRASER:  Thank you very much.
   MR GREEN:  My Lord, that's hopefully a helpful introduction
       to what the micro texture of the facts looks like,
       trying to build up and answer the question.
           Can I just very briefly take your Lordship to
       a couple of documents which give an overview from the
       internal perspective of Post Office and your Lordship
       will have in mind that if we just quickly turn to
       {F/1325/145}, Post Office's public position in its
       response to Second Sight of 11 March 2015, paragraphs 18
       and 19:
           "Questions were raised about Post Office's plans to
       change to a new system when the Post Office's current
       contract with Fujitsu in respect of Horizon comes to an
       end in March 2017.
           "Post Office's intention to move to a new system
       does not reflect any dissatisfaction or lack of
       confidence in Horizon.  It is simply that the current
       contractual arrangements are due to expire."
           So that's the public position.  And there are
       a number of documents -- given the time I will just take
       your Lordship to a couple of them -- which suggest that
       there were concerns with both the overall IT system and
       also the back office aspects of that and the ability
       properly to control and understand the sorts of issues
       that would be required in reconciling data.
           If we can go very briefly to {F/1557}, this is
       22 October 2016, paragraph 3:
           "Our back office also struggles with the
       complications of dealing differently with each of our
       many clients, heavily manual processes, reconciling
       disparate sources of data, retrospective financial
       controls and a lack of flexibility.  This backlog of
       challenges, poor support contracts and a lack of skills
       have led to a prohibitive cost of change, preventing the
       about improvements that should occur as part of
       a business as usual."
           And if we go then forward please to {F/1587/1}.
       This is a draft document dated 28 November 2016, so
       I qualify it with the fact that it is a draft, and
       "Context" it says:
           "This document forms an update to the IT strategy
       approved in July by the POL board.  In July we outlined
       that IT was not fit for purpose, expensive and hard to
           If we go to page 3 of that document {F/1587/3},
       paragraph 1:
           "The IT strategy outlined a view of the current
       state of technology within POL as failing to meet POL
       aspirations on any assessment lens (cost, risk, delivery
       or service)."
           And paragraph 5:
           "Our greatest 'run the business' risk areas where we
       are outside of appetite are age and state of the legacy
       environment (eg Horizon availability, back office
       estate, lack of resiliance and [disaster recovery] of
       core platforms ...) ..."
           Then Mick 24.11.16 appears to be repeatedly
       referenced throughout this document.  One of the main
       risk drivers over the page at {F/1587/4} is lack of
       a robust IT and controls environment in the table at
       paragraph 7, that's one of the main risk drivers
       identified in this document in November 2017.  Then if
       we can just briefly go to the final version of that
       document at {F/1610}, that's dated 30 January 2017 and
       your Lordship will note from that page that in context
       again the line that was in the draft has survived the
       considered review that doubtless took place before it
       was finalised:
           "In July we outlined that IT was not fit for
       purpose, expensive and difficult to change."
           And the conclusion refers to reporting on the IT
       strategy in July.  And at page 2 {F/1610/2}, the first
       indented bullet point:
           "We need to quickly rationalise and resolve
       misaligned contracts enacted to support legacy IT,
       obsolescence, a lack of PO technical competence,
       particular focus on Fujitsu and Accenture."
           And on page 3 of that document {F/1610/3} at bullet
       point 1:
           "The IT strategy outlined a view of the current
       state of technology within PO as failing to meet PO
       aspirations on any assessment lens ..."
           Which is the quote I have already identified to
       your Lordship.  And quite a lot more of the same.  The
       only last page I would refer your Lordship to just very
       briefly -- we will have to go to these in more detail
       with other witnesses, but at page 8 of that document
       {F/1610/8} there's a diagram on page 8 and it is
       introduced by the bullet point just above it:
           "The following highlights the current operational
       risk areas referred to earlier in the document and the
       initiatives underway or proposed ... to migrate these
       into risk appetite - purple - severe risk, red -
       high risk, amber - within appetite but attention
       required ..."
           And your Lordship will note that the Horizon branch
       systems box is in red in the Post Office's own internal
   MR JUSTICE FRASER:  Which means it is high risk.
   MR GREEN:  It means it is high risk.
   MR GREEN:  My Lord, in overview then, your Lordship will be
       hearing some important factual evidence which we
       respectfully say is going to be an absolutely essential
       bedrock for the experts then to give their considered
       opinions on, after the factual evidence that Post Office
       has put forward has been properly tested, because I'm
       afraid we've got some aspirational "would have" evidence
       and conclusions feeding through into the expert reports.
           The three key points are the parties radically
       differ on their approach to working out the answers.  We
       are going to commend the approach of what actually
       happened rather than the hypotheses.  The second is that
       we are confident that the factual enquiry that will be
       undertaken will enable your Lordship to have a more
       robust basis for drawing any inferences as to the
       answers than without it.  And the final point is that
       the question of robustness we respectfully say, while
       relevant, is not going to be the determinative prism
       through which those issues fall to be answered.
   MR JUSTICE FRASER:  Final question.  Trial management, we
       are starting with Mr Latif tomorrow.
   MR GREEN:  My Lord, yes.
   MR JUSTICE FRASER:  He is on the video, is that right?
   MR GREEN:  He is.  He is travelling specially to Islamabad
       to get a better connection.
   MR JUSTICE FRASER:  In view of time difference do you want
       to start at 10 rather than 10.30?
   MR GREEN:  My Lord I think it is all teed up for 10.30.
   MR JUSTICE FRASER:  That's fine.  I need not interfere any
           Okay, so 2 o'clock.
   (1.00 pm)
                    (The luncheon adjournment)
   (2.00 pm)
            Opening submissions by MR DE GARR ROBINSON
   MR DE GARR ROBINSON:  My Lord, good afternoon.  I will just
       say something briefly first of all about the claimants'
       history, Mr Green's history, of how we got here.  He
       devotes 25 pages of his opening submissions to those
       questions.  My submission is that those questions are
       mostly a distraction from the important questions in
       this case, which is the Horizon issues.  Many of the
       points that he makes are unfair and some of them we say
       are downright wrong, but I will not take up time I don't
       have in addressing them in this opening.  Instead,
       subject to your Lordship, I propose to deal with them in
       writing in our written closings.  If I have time at the
       end of these submissions I may say a few brief things
       about some of them, but I suspect I won't have time.
           Laying those points to one side, the Horizon issue
       consists of 15 questions and, as your Lordship will have
       seen from our submissions, we say they can be
       conveniently divided into three groups.  In descending
       order of importance, the first question is is Horizon
       robust or does it cause branch shortfalls?  The second
       group of questions can be summarised as was Fujitsu --
       because we are talking about Fujitsu here -- secretly
       manipulating Horizon data?  And then thirdly there's
       a miscellany of factual questions about the function and
       operation of the system.
           Your Lordship has already heard from Mr Green
       indicating that the operational/functional issues are
       largely agreed.  Despite the apparent differences
       between the experts on proper analysis we say there is
       no material disagreement between them.  They are agreed,
       for example, that subpostmasters could run over 100
       different types of report covering transactions
       conducted at the branch.  In the ordinary course these
       reports would show enough information for
       a subpostmaster to be able to balance his or her
           My Lord, these reports can be used to identify the
       causes of some types of discrepancies but not others
       because to do more would require knowledge of complex
       back-end systems which subpostmasters cannot be expected
       to have and for your Lordship's note that's in the
       second joint statement at paragraph 9.3 {D1/2/39}.
           It is important to note that it is not suggested by
       the claimants that the -- or at least by Mr Coyne at any
       rate -- that subpostmasters should be given direct
       access to information of a back-end kind and indeed in
       his second report your Lordship will note that Mr Coyne
       positively disclaims any suggestion of that sort.  He
       deals with that at paragraph 5.380 which is at
           So why are those issues here?  They are here because
       it is the claimants' pleaded case that there was
       something wrong with the way in which Horizon was
       designed and operated for subpostmasters.  For example,
       they say there was an asymmetry of information which was
       inappropriate.  Horizon did not give them the
       information that they should have done.  Amongst other
       things, the suggestion seems to be that it should have
       notified them of bugs, it should have given them far
       more information than it did to enable them to dispute
           Your Lordship will see that from paragraph 19.3 of
       the generic particulars of claim.  I won't take
       your Lordship to it but for your note the reference is
       {C3/1/7}, where there is a reference to limitations on
       subpostmasters' ability to access, identify and
       reconcile transactions recorded on Horizon and the lack
       of any or any adequate report writing features as
       repeatedly raised by Mr Bates.
           Now, that is what is noted by Dr Worden in his first
       report and he discerns an assumption, he says, in the
       claimants' case that there is something wrong with the
       level of information that's provided to subpostmasters
       and your Lordship will see that, for example, in
       paragraph 954 of his first report, the reference to
       which is {D3/1/213}.
           My Lord, in my submission he correctly discerned
       that allegation in the claimants' case, but in
       Mr Coyne's second report, paragraph 5.380, that's
       {D2/4/226}, Mr Coyne has now firmly disassociated
       himself from that suggestion.  So that's good news, it
       means we can put that aspect of the claimants' case to
       one side.
           But the claimants' opening does go further.  It
       spends a great deal of time addressing the effectiveness
       of the information that was provided to subpostmasters,
       the reports that are covered in the functional and
       operational issues.  In my respectful submission, that
       is unnecessary and inappropriate for the purposes of
       this trial.  It is not a Horizon issue -- the
       effectiveness of those reports, what can be done with
       them is not a Horizon issue.  Whether in any given case
       a report is sufficient to enable a subpostmaster to
       identify the root cause of a discrepancy depends on the
       nature of the discrepancy that he or she is faced with.
       It is essentially a breach issue and it is not something
       that should take any time up in this trial.  There is
       already quite enough to be arguing about.
           So we can put the operational issues largely to one
       side.  We say, as your Lordship will be well aware, that
       the critical issue is one of robustness.  The claimants
       seem to think that this is the wrong question.  Indeed,
       they criticise Post Office for seeking to include it in
       the Horizon issues, even though all that Post Office did
       was seek to formulate issues that reflected the
           Now, why is robustness critical?  It is explained in
       our opening at paragraphs 23 to 26.  When the court
       decides individual claims an important question will be
       whether Post Office can generally rely on accounting
       data from Horizon as evidence of what the postmaster or
       his staff keyed in at the relevant time, or whether the
       relevant claimant can say the court should infer or
       presume that the shortfall shown by the accounting data
       was caused by a bug.  The question of robustness is not
       dispositive of that question.  What the court decides in
       any breach trial will depend on the evidence of that
       particular case.  For example, it could be shown that
       the shortfall was caused by a particular bug operating
       in a particular way at a particular time, or it could be
       shown that the shortfall was caused by a particular
       transaction correction and that that transaction
       correction was erroneous, but these are individual
       breach issues which will turn on the facts of the
       individual breach trials.  This trial is to decide the
       generic issues.  Hence the wording of the Horizon issues
       themselves.  I don't know whether your Lordship has
       a copy of --
   MR JUSTICE FRASER:  Well, the ones I'm using are at your
   MR DE GARR ROBINSON:  Well, I'm grateful.
   MR JUSTICE FRASER:  They are in a few places but it seemed
       the most useful.
   MR DE GARR ROBINSON:  Yes, we hoped that would be helpful to
       your Lordship.
           Issue 1:
           "To what extent was it possible or likely for bugs,
       errors or defects of the nature alleged to the
       effects ..."
           Set out there.  Your Lordship will note that the
       issue did not stop at the word "possible".  It is common
       ground that it's possible.  No one is claiming that
       Horizon is perfect, no one has ever claimed that Horizon
       is perfect.  The interesting question is not whether
       it's possible; we know it has happened.  The interesting
       question is whether it is likely.
           Similarly with issue 3:
           "To what extent and in what respects is the Horizon
       system robust and extremely unlikely to be the cause of
       shortfalls in branches?"
           That asks an extent question.  It asks a likelihood
       question.  Why?  Because as I have sought to explain to
       your Lordship, that's a useful issue which will then
       have a bearing on the conduct of breach trials to be
       conducted in the future.
           Similarly with issue 4:
           "To what extent has there been potential for errors
       in data recorded within Horizon to arise in data entry
       and so on?"
           Again, it is another extent question.  And, my Lord,
       in paragraph 6, Horizon issue 6:
           "To what extent did measures and/or controls that
       existed in Horizon prevent, detect, identify, report, or
       reduce to an extremely low level the risk of the
       following ..."
           That's a risk question, an extent of risk question.
           So, as I say, my Lord, the critical question that's
       raised by the Horizon issues, the relevant Horizon
       issues here is not whether it's possible that bugs have
       occurred in Horizon which have caused branch shortfalls;
       we know it's possible, it's actually happened.  That's
       not even an issue on the pleadings, my Lord.  The
       critical issue is the issue that's raised in
       paragraph 16 of the generic defence and perhaps I could
       ask your Lordship to look at that.  It's at bundle
       {C3/3/5}.  Paragraph 16 reads:
           "Highly generalised and speculative allegations are
       made that Horizon ... is unreliable or vulnerable to
       manipulation and thus may have been the root cause of
       some of the losses in branches.  Like any IT system,
       Horizon is not perfect, but Post Office maintains that
       it is robust and that it is extremely unlikely to be the
       cause of losses in branches."
           My Lord, that means it is extremely unlikely -- in
       any given case, when you see a loss in a branch, it is
       extremely unlikely that that loss is going to be caused
       by a bug.
           So, as I say, my Lord, the question is not whether
       it's possible for Horizon to create shortfalls, the
       question is whether in relation to any given shortfall
       Horizon is likely or unlikely, or extremely unlikely, to
       have caused those shortfalls.  In my submission that's
       the essential question at the heart of this trial.  It
       will allow the parties to know whether at any breach
       trial the accounting information on Horizon for the
       relevant branch and the relevant month is a valid
       generally reliable starting point.
   MR JUSTICE FRASER:  I'm not in any way being difficult,
       I think we may as well just deal with it upfront at the
       beginning.  Am I to read "robust" as meaning "extremely
       unlikely to be the cause", or is there more meaning to
       "robust" than that?  Because I think whatever it is, we
       all have to make sure we are using the word the correct
       way, or the same way.
   MR DE GARR ROBINSON:  The concept of robustness is a concept
       which involves reducing to an appropriate low level of
       risk, the risk of problems in Horizon causing shortfalls
       which have a more than transient effect on branches.  So
       it involves both measures to prevent bugs arising in the
       first place but those measures are never going to be
       perfect and it includes measures which operate once
       a bug has actually occurred and triggered a result.  It
       is both aspects of the equation.
           I don't say that the word "robust" necessarily means
       "extremely low level of risk", but what we say is that
       if you have a robust system it produces a result in
       which the system works well in the overwhelming majority
       of cases and when it doesn't work well there are
       measures and controls in place to reduce to a very small
       level the risk of bugs causing non-transient lasting
       shortfalls in any given set of branch accounts.
   MR JUSTICE FRASER:  All right, thank you very much.
   MR DE GARR ROBINSON:  So it is not necessary or indeed in my
       submission appropriate to investigate all the bugs or
       errors that have ever been found in Horizon if those
       bugs have no real impact on branch accounts.  There
       simply isn't the time for that.
           The important thing is to focus on bugs which have
       a financial impact on the branch accounts.  And, as
       I have already sought to explain to your Lordship, one
       needs to bear in mind that robustness does not mean bugs
       can't occur, or will almost never occur; it includes the
       controls and measures which have the effect that when
       they do occur they are likely, very likely, to be picked
       up and remedied.
           One consequence of that is that some bugs may have
       an effect on branch accounts which is only going to be
       transient because the system supporting Horizon will
       pick up the impact and correct it as night follows day
       in the ordinary course of the business.
           Let me give your Lordship an example.  The example
       of remming in and remming out.  My learned friend spent
       some time this morning explaining the Dalmellington bug
       and what he described as linked bugs, some other linked
       bugs that were also dealt with in the report that
       your Lordship saw.  Actually those other bugs were
       separate bugs -- they had a similar symptom but they
       were separate bugs.
           In remming it is an easy case.  It is particularly
       easy when it is an internal rem, as it were, from the
       branch to an outreach branch.  There are a very small
       number of outreach branches.  But also more generally
       when you are remming in cash that has been provided by
       Post Office and is sent to the branch, the Post Office
       will have its own record of how much money was sent out
       and the branch will have its own record of how much
       money was received.  If those two numbers don't
       reconcile that is going to be picked up as night follows
       day and that is going to be the result of some process
       which will result in the problem being resolved.
       Your Lordship will have seen from the Dalmellington bug
       report itself that you saw this morning that a large
       number of the impacts that were discussed in the course
       of that report were sorted out by transaction
       corrections without the need for any intervention by
       Fujitsu or anybody else.
           In fact, Fujitsu still fixed those bugs,
       your Lordship will note, so Fujitsu picked up on the
       bugs even though they weren't the result of reports from
       subpostmasters to Fujitsu.  So Fujitsu itself -- it's
       quite a good example of how problems in the system, in
       reconciliation, will result in Fujitsu becoming aware of
       things because of its own automatic systems and sorting
       them out.  But the important point is that it will be
       extraordinarily rare for a remming in or a remming out
       problem ever to produce a result that would have
       a lasting effect that will cause a branch to suffer
       a shortfall which leaves it powerless to resist.  That's
       simply not in the nature of the remming process and the
       reconciliation processes that are associated with it.
           While dealing with reconciliation, my learned friend
       I noticed tried to kick me in the heels by inviting me
       to state what my case is on reconciliation generally and
       transaction corrections --
   MR JUSTICE FRASER:  I always ignore those sorts of comments.
   MR DE GARR ROBINSON:  My Lord, it is a tempting thing for
       him to do, I understand why he did it, perhaps I would
       have done it if I was in his position, but the important
       thing to note is that the reconciliation process, we all
       know how it works, it is essentially automated and
       discrepancies are picked up between the system and when
       discrepancies are picked up they are then investigated.
           There isn't enough time in this trial to investigate
       the processes -- and there are many of them -- by which
       discrepancies are investigated.  It simply wouldn't be
       possible and that's why those issues were positively
       excluded from the Horizon issues.  There was a debate
       about it and that part of the process was excluded and
       that's why your Lordship will see from Horizon issues 5
       and 15 which deal with reconciliation and TCs,
       your Lordship will see that those questions are
       essentially factual.
           So to the extent that my learned friend is
       suggesting that one can only have a Horizon trial if one
       had disclosure of every single reconciliation process
       that has been operating in Post Office or Horizon for
       the last 20 years and every single transaction
       correction that's ever been sent in the last 20 years,
       my learned friend is in my respectful submission
       obviously wrong.
           Now, does that mean, as my learned friend would like
       to pretend, that it is therefore not possible to have
       a trial of the robustness question?  And the answer
       is: obviously not.  We all know that the process by
       which reconciliation happens means that where there are
       irreconcilable figures, where there are discrepancies
       between figures that should be the same, they will be
       looked at and in the vast majority of cases they will
       result in conversations, they will result in enquiries,
       which will cause the problem to be resolved in
       a satisfactory way.
           My learned friend wants to say "Well, sometimes" --
       it being a human process -- "sometimes the corrective
       process will come to the wrong conclusion."  Well,
       my Lord, that doesn't alter the fact that the process
       itself is a corrective process, it's a process that will
       tend to improve the reliability of the ultimate figures
       rather than take away from the reliability of the
       ultimate figures.
           So, my Lord, that's the very short answer to
       my learned friend's question.  If he wants a longer one,
       or if your Lordship wants a longer one, we can of course
       discuss it later.
           My Lord, another example of measures which prevent
       problems from arising is in relation to recoverable
       transactions.  Your Lordship will have seen some
       discussion of recoverable transaction in the expert
       reports and also in some of the witness statements.  So,
       for example, when the system crashes, there's a power
       failure or something similar, it is always possible that
       a transaction being undertaken at the branch will not at
       that point have reached the Horizon system, for example
       because the outage occurs before the basket for the
       relevant transaction is closed.  It is when the basket
       is closed that the transaction gets added to the
       branch's Horizon accounts.
           Now, in that situation the accounts will not show
       the transaction that has been done and it might be that
       that transaction has already resulted in cash having
       changed hands.  My Lord, does that mean there's
       a problem in Horizon?  Is that the result of a bug which
       is a matter of some criticism?  My Lord, the answer is
       no because what Horizon does have is a system which
       enables the transaction that hasn't been recovered to be
       identified and your Lordship will see from the witnesses
       who are addressing this issue, Mrs Burke for example,
       that the unrecovered transaction was actually
       specifically identified to her and your Lordship will
       also see that Post Office, because of its own
       reconciliation processes, was aware of it and was able
       to deal with it and Mrs Burke's -- or rather Mr Burke's
       branch did not suffer a lasting loss as a result of what
       happened on that day.
           So, my Lord, those are just two very broad examples
       of mechanisms in place that have a tendency to reduce to
       an even smaller level the risk of lasting discrepancies,
       lasting shortfalls being inflicted on branch accounts as
       a result of bugs.
           But to take a step back, the critical point I would
       like to address your Lordship on is that it is not
       helpful simply to ask the question whether it is
       possible that there are bugs out there which affect
       branch accounts.  We know it is possible.  The question
       is, the interesting question for your Lordship is
       whether bugs are likely or unlikely to affect a given
       set of branch accounts in a given month in a given
       breach trial and, my Lord, on that question possibility
       is not the same as probability.  You can't simply say
       "Well, it's possible it happened", that's not enough.
           Now, in their submissions the claimants at times
       appear to recognise this, but neither they nor their
       expert Mr Coyne acts on it.  My learned friend took
       your Lordship to paragraph 17 of the claimants' opening
       submissions this morning.  Perhaps we could have another
       look at that.  It is at {A/1/9}.  The claimants say:
           "As to the precise wording of the issues, the
       claimants previously made clear that they consider the
       wording of some of the issues insisted upon by
       Post Office to be unhelpful, but their inclusion was
       ultimately agreed by the claimants in order to reach
       agreement with Post Office.  Post Office had very clear
       views on two particular points.
           "First, on issue 1 (Robustness): Post Office was
       insistent on including this issue, formulated by the
       defendant is whether Horizon is 'robust' and 'extremely
       unlikely' to cause shortfalls.  This reflects language
       pleaded in [the defence], and indeed 'robustness' has
       been one of Post Office's 'narrative boxes' and
       a favoured term in Post Office's public relations
       pronouncements ... Coincidentally or otherwise, it has
       also featured in the NFSP's defence of Post Office
       relied upon by Mrs Van den Bogerd.  However, as the
       claimants made clear in their [reply], whereas the
       claimants' is that it is relatively robust and has
       become more robust over time - but not so as to be an
       answer to the Claim (and in so far as 'robustness' has,
       in this case, a sufficiently clear meaning – addressed
       further herein).  The combination of Horizon's admitted
       imperfections (and discovered bugs and remote access)
       and the volume of many millions of transactions, 14 is
       entirely consistent with the levels of errors reflected
       in the Claimants' case."
           My Lord, two points are being made in that
       paragraph.  The first one I can deal with quickly.  It's
       that the claimants' own case is that Horizon is
       relatively robust.
           The second point is that the combination of admitted
       imperfections -- and I apprehend my learned friend would
       now say the impressions that have been shown as a result
       of Mr Coyne's analysis -- and the volume of the many
       millions of transactions that are done through Horizon,
       is consistent with the level of shortfalls in relation
       to which the claimants make their claims in this
           My Lord, as to the first point, it is simply wrong.
       Mr Green has shown you paragraph 37 of his reply, that's
       at bundle {C3/4/21}, where it is specifically denied
       that Horizon is robust.  So on the pleadings at least it
       is not the claimants' case that Horizon is relatively
       robust, the claimants obviously don't like that, but
       after seeing the expert evidence the claimants, as
       I apprehend Mr Green, are now admitting the central
       issue which was previously denied.  In my submission
       that's a major achievement from having this trial.  In
       my further submission it is the opposite of unhelpful to
       have raised it as an issue in this trial.
           My Lord, the second point -- perhaps we could look
       at paragraph 37 -- well, we can pick it up in
       paragraph 17.2.  What the claimants are saying is that
       the relatively small chance of bugs in Horizon, because
       of the volume of transactions undertaken in Horizon, is
       likely they say to produce the very picture that's
       reflected in the claimants' case.
           Now, what's important to understand about that is
       that the claimants are relying on the concept of
       likelihood to support their case, so, as in my
       submission they have to, they are groping towards
       a numbers argument, the very kind of numbers argument
       that Dr Worden makes in his reports and about which they
       have so many critical things to say in their
       submissions.  They say that the level of bugs in Horizon
       is likely to reflect the fact of the claimants' case.
       But what's entirely unstated in their evidence is how
       the level of bugs encountered in Horizon is likely to
       produce anything like the £18.7 million of shortfalls
       that the claimants say should not have been included in
       their accounts.  My Lord, and it is unsupported by their
       expert who refuses point blank to engage in any
       numerical analysis at all and yet at the same time they
       criticise Dr Worden for trying.
           So in my submission, my Lord, the claimants are in
       an awkward position.  In making the case which I have
       just referred to in paragraph 17.1 they are engaging in
       an intellectual process which their own expert refuses
       to engage in and which when Dr Worden tries to engage in
       it they then criticise him.
           The best that they can say is that having found 29
       bugs it is possible that other bugs have arisen, but
       they won't say anything about how likely it is other
       bugs have arisen, how big the effects of those bugs are
       likely to be, they won't engage in any kind of questions
       of that sort at all.
           Now, in their submissions the claimants say that
       they will challenge Dr Worden's numerical analyses.
       That is to be welcomed.  It will assist your Lordship to
       assist the soundness of his calculations.  At the moment
       there is no engagement really by Mr Coyne with any
       questions of likelihood or extent, there are just some
       criticisms made of some of the assumptions that
       Dr Worden makes in his report.
           Now, it is worth noting that Dr Worden has a number
       of different calculations, some of which are more
       complicated and some of which involve more assumptions
       than others.  Let me just deal with one very simple
       calculation.  This requires no understanding of
       statistics or mathematics.  It is set out in section 8.5
       of Dr Worden's first report which starts at {D3/1/148}
       and it has changed a little bit in Dr Worden's second
       report but we don't need to address that in any detail
       at this stage.
   MR JUSTICE FRASER:  I should just tell you for interest I do
       understand mathematics and statistics.  I'm not being
       funny, but I do.
   MR DE GARR ROBINSON:  No, that's very helpful, my Lord.
       I thought I did, my Lord, I have several maths A-levels,
       but I realised that my own sense of my own mathematical
       abilities was rather greater than it turned out to be.
   MR JUSTICE FRASER:  I mean this one is just a simple
       multiplication, isn't it?
   MR JUSTICE FRASER:  I think most school children would
       probably follow this one.
   MR DE GARR ROBINSON:  Exactly.  It is one I understand:
           Over the period 2000 to 2018 the Post Office has had
       on average 13,650 branches.  That means that over that
       period it has had more than 3 million sets of monthly
       branch accounts.  It is nearly 3.1 million but let's
       call it 3 million and let's ignore the fact for the
       first few years branch accounts were weekly.  That
       doesn't matter for the purposes of this analysis.
           Against that background let's take a substantial bug
       like the Suspense Account bug which affected 16 branches
       and had a mean financial impact per branch of £1,000.
       The chances of that bug affecting any branch is tiny.
       It is 16 in 3 million, or 1 in 190,000-odd.  The chances
       of affecting a claimant branch are even tinier because
       the claimant branches tended to be smaller than ordinary
       branches.  One could engage in all sorts of
       calculations, but your Lordship may recall from
       Dr Worden's second report that he ends up with
       a calculation of a chance of about 1 in 427,000-odd.  So
       for there to be a 1 in 10 chance for a bug of this scale
       to affect one set of monthly account for a claimant
       branch, one would need something like 42,000 such bugs.
           Of course there's a much simpler way of doing it
       which really is just a straight calculation.  There have
       been 3 million sets of monthly accounts so the chances
       of the Suspense Account bug affecting any given set of
       monthly accounts is 60 in 3 million or about 5 in
       a million, so to get a one in 10 chance of such a bug
       you would need to have 50,000 bugs like it.
           But, my Lord, all the roads lead to the same basic
       result which is that even for a significant bug of that
       sort, the number of bugs that would need to exist in
       order to have any chance of generating even a portion of
       the losses that are claimed by the claimants would be
       a wild number that's beyond the dreams of avarice.  It
       is untenable to suggest that there are 40,000 or 50,000
       bugs of that scale going undetected in Horizon for
       20 years.
           Dr Worden explains that in paragraphs 643 and 644 of
       his first report and the reference to that is
       {D3/1/152}.  And it is interesting, my Lord, that the
       claimants very sensibly do not suggest that there will
       have been bugs of that scale in that number operating --
       lurking secretly in Horizon for the last 20 years and
       they don't suggest it because they can't.  It's a matter
       of common sense.  And in my respectful submission just
       that calculation demonstrates that the claim made at the
       end of paragraph 17.1 of the claimants' submissions is
       untenable.  A combination of Horizon's impressions with
       the volume of transactions done in Horizon is not
       entirely consistent with the errors reflected in the
       claimants' case.  In my respectful submission it is
       obviously inconsistent with that.
           Just to be clear, that's not to say that a claimant
       could not have been hit by a bug.  As I hope I have made
       clear to your Lordship, Horizon is not perfect.  It
       remains a possibility, but the important point is how
       unlikely it is.  But of course the question of whether
       an individual claimant has suffered an impact as
       a result of a bug is not a point for this trial.  That
       is a breach issue to be dealt with in an individual
       case.  This trial is about setting a baseline for
       Horizon's reliability, not a final conclusion that will
       govern every single breach case that comes before
       your Lordship.
           Now, before addressing the expert reports on
       robustness it is worth noting the large measure of
       agreement that now exists between the experts.  There is
       no dispute about the architecture or capabilities of
       Horizon.  There's no suggestion that Horizon lacks
       important capabilities or that it doesn't generally
       perform satisfactorily.  There is no suggestion of any
       systemic problem lurking in Horizon.
           In short, it is accepted that Horizon works well for
       the overwhelming majority of cases and consistently with
       that it is now common ground between the experts that
       Horizon is robust and that its robustness has improved
       over time and your Lordship already has the reference,
       it is the joint statement, the third joint statement,
       page 2, {D1/4/2}.
           Now, what does relatively robust mean?  It means
       robust as compared with comparable systems -- big
       systems, systems that keep aircraft in the air, that run
       power stations and that run banks.
           My Lord, by the same token it is common ground that
       the Horizon is not infallible.  It has and will continue
       to suffer faults every now and then.  Sometimes, in
       a really small number of cases, those faults will have
       an effect on branch accounts, but it should be
       remembered that robustness is not just about preventing
       bugs from appearing in the first place, it is also about
       limiting the lasting detrimental effects when they do
           Your Lordship will hear evidence that bugs affecting
       branch accounts are given a high priority when they are
       addressed by Fujitsu.  They are not ignored.  And,
       my Lord, the evidence also shows that bugs which have an
       effect on branch accounts occur only very rarely indeed.
       There is a dispute between the experts as to precisely
       how rarely, but in the context of a huge system that's
       been in continuous operation for 20 years, that dispute
       in my submission does not have a material bearing on the
       outcome of this trial.  In the overwhelming majority of
       cases, branch accounts will not contain a shortfall
       caused by a bug and the scale of bugs that would be
       needed to undermine that simple fact would be enormous.
           Putting the point another way, the difference now
       being played out between the experts is at the margins.
       They accept that there are imperfections in the Horizon
       system with the result that in some rare cases bugs
       affecting branch accounts occur and will not be
       immediately fixed.  The issue between them is how slight
       are the relevant imperfections.
           The scale of this difference is magnified by the
       adversarial process but in the scheme of things, in my
       submission, it is in fact tiny and to plagarise
       Lord Justice Lewison in section 1 of his first chapter
       in the interpretation of contracts, the lazy reader can
       stop here.
           My Lord, we say that what is already common ground
       between the parties means that the claimants must fail
       in their primary endeavour to persuade the court to draw
       the inference or make the presumption that they want
       the court to make or to draw or make, to the effect that
       when faced with a shortfall in a set of branch accounts
       the shortfall was caused by a bug in Horizon.
           Now, against that background let me say a few words
       about the experts.  The claimants submissions are most
       unfair to Dr Worden.  Your Lordship will see those at
       {A/1/33}.  First of all, it is quite wrong to accuse him
       of bias.  He is an independent expert whose views are
       his own.  He is not a mouthpiece for Post Office's case
       and the claimants should not be suggesting that he is,
       as I see that they appear to do in their submissions.
           Secondly, the submissions give the impression that
       Dr Worden's analysis is limited to the three admitted
       bugs that -- the three bugs that were discussed with
       Second Sight and that were then the subject of the
       pre-action correspondence and your Lordship gets that
       from paragraph 156 of their submissions {A/1/54} where
       they say:
           "Dr Worden's analysis extrapolates from only three
       bugs which happen to be those previously acknowledged by
       Post Office.  It appears that Post Office had not
       disclosed to him the existence of other bugs, which he
       could not have taken into account."
           My Lord, that is an unfair and incorrect account of
       the evidence that's given by -- or will be given by
       Dr Worden.
           If your Lordship would go to Dr Worden's first
       report, annex D -- and, my Lord, it starts at
       {D3/2/95} -- your Lordship will see a very lengthy
       appendix which describes itself as containing:
           "... some tables of KELs which are referred to in
       the report.  The tables are:
           "First a sample of 30 randomly selected KELs
       (selection of every 100th KEL in alphabetically sorted
       list), with commentary on the robustness countermeasures
       which acted in the case of each KEL, as well as its
       potential financial impact."
           Then secondly:
           "62 KELs mentioned in Mr Coyne's report, for which
       I have also analysed which robustness countermeasures
       applied and analysed the possible impact on branch
           "A further sample of 50 randomly selected KELs (also
       every 100th KEL in an alphabetically sorted list) which
       I have analysed for possible financial impact, but
       I have not tabulated my analysis for robustness
           And fourth:
           "A sample of 50 KELs which each include the symbol
       '£' in their text - because in my opinion, this makes
       them more likely to be concerned with possible financial
       impact on branch accounts."
           So your Lordship will see that Dr Worden was not
       simply looking at the three admitted bugs and then
       performing all sorts of mind games on the basis of those
       bugs, nor did he overlook and ignore the Dalmellington
       bug which Mr Green addressed your Lordship on this
       morning; he looked at all of them.  He looked at more
       than just the bugs that were identified by Mr Coyne, he
       took random selections and he tried to use a form of
       searching which would disclose bugs that were more
       likely to have a financial impact on branch accounts.
           So, my Lord, that's a first point I should make to
       your Lordship.  It would be quite wrong to proceed, as
       I apprehend my learned friend would have you proceed, on
       the basis that Mr Coyne has done an in-depth analysis of
       the problems in Horizon, whereas Dr Worden has just
       looked at some pretty pictures about three admitted
       bugs.  That isn't what Dr Worden did at all.
           My Lord, the second point I should note is that in
       his second report Dr Worden increased the number of KELs
       and PEAKs that -- I should say he looked at the
       associated PEAKs as well, he didn't just look at KELs --
       he increased the number of samples that he reviewed to
           And thirdly I need to make it clear to your Lordship
       that the review that he conducted was thorough, we say
       much more thorough than Mr Coyne's review.  What
       Mr Coyne intended to do -- this will be obviously
       investigated with the witnesses -- is Mr Coyne tended to
       find phrases in particular documents that indicated
       a problem and he would stop there.  He didn't,
       for example, seek to ascertain in JC1, in his first
       report, he didn't seek to ascertain whether any
       particular bugs would actually have a branch effect at
       all.  That's not the exercise that he did.  What he did
       in his first report was just find as many problems as he
       could.  But what Dr Worden did was he considered both
       potential branch impact and he considered the operation
       of the countermeasures in practice.
           Your Lordship will apprehend that Dr Worden is
       criticised for engaging in some kind of armchair
       analysis of countermeasures on the basis of design
       aspirations.  My Lord, that characterisation is quite
       unfair, we would say.  But more importantly what
       Dr Worden did is that he looked at the operation of the
       system, the PEAKs and the KELs relating to particular
       problems, and he looked to see how the countermeasures
       he had identified he had seen built into the system, how
       those countermeasures worked in any particular case.
       My Lord, that's an important function to perform if one
       is engaged in the process of seeking to obtain
       a balanced view of the robustness or otherwise of an IT
       system and it is important to note that it is not
       something that Mr Coyne did at all, certainly not in his
       first report.  Even in his second report he goes no
       further than saying "Well, there are some bugs which
       were missed by the countermeasures."  Well, my Lord, so
       there are, of course there are going to be bugs that get
       missed by all countermeasures.  It is not suggested that
       it is impossible for bugs to arise which have a lasting
       impact on branch accounts.  The critical question is how
       likely it is that such bugs will lurk in the system and
       be undetected in the way that the claimants would have
       you find.
           It is in that respect worth noting that Dr Worden
       found more bugs with non-transient branch impacts than
       Mr Coyne did in his report and your Lordship will see
       that from our opening submissions at paragraph 219.3 and
       for your Lordship's note the reference is {A/2/76}.
           Your Lordship will be aware that of the 29 bugs that
       are now listed in JS2, Dr Worden thinks there is
       evidence to suggest there may be up to 12 bugs that had
       a non-transient financial effect on branch accounts.  Of
       those 12 it is worth noting that Dr Worden identified
       five of them.
           So this is not a case, my Lord, where some kind of
       blindfold -- the picture that my learned friend seeks to
       paint of the process that Post Office went through and
       the process that Dr Worden undertook in arriving at his
       report was one in which he was blindfolded by
       Post Office and somehow stumbled into providing
       a positive report which just considered the three bugs
       that Post Office had already identified.  My Lord,
       that's simply a travesty of the true facts.
           On the basis of the documents he reviewed, Dr Worden
       took a balanced view of the design of Horizon and its
       countermeasures and of the operation in practice of its
       countermeasures as evidenced by PEAKs and KELs and
       associated documents.  My Lord, and he also took
       a balanced view of the service history of Horizon, the
       support function provided by Fujitsu and its efficiency.
           Now, that will be a battleground in the rest of this
       trial, I know, but even the documents that your Lordship
       saw this morning with the Dalmellington bug,
       your Lordship will see the rigour that's applied.
       There's concern that postmasters aren't given advice
       that might be incorrect.  The rigour associated with
       that process and the determination of Fujitsu and the
       other people involved, the other stakeholders, to get to
       the bottom of what happened is quite striking in my
           In relation to the Dalmellington bug they did get to
       the bottom of what had happened.  They identified 112
       potential branches with financial impact, 108 of which
       had been fixed or made good.  Of the other four only two
       had a significant problem and, my Lord, of those two
       further research showed that those two branches were not
       actually affected by the Dalmellington bug at all, they
       just had similar symptoms that were the result of an
       entirely separate cause and that were fixed entirely
       separately and my learned friend gave you I think the
       reference to the document which shows that.
           So by January 2016 the Dalmellington bug had been
       fully investigated and it was quite clear what the
       results of that bug were and my learned friend seeks to
       suggest that there was more investigation to be done and
       that Post Office somehow stopped that further
       investigation being done.  That's not right at all.
       By January 2016 Dalmellington had been fully analysed
       and its consequences bottomed out and my learned friends
       took your Lordship to a number of emails from
       Post Office, they were from July 2016, six months later.
           So, my Lord, Dr Worden's conclusion is that Horizon
       is well designed and well supported by a team of people
       who have been working on it from the start and who are
       very thorough when investigating possible bugs and
       your Lordship will I'm sure bear in mind Mr Parker's
       witness statement where he says that they would keep
       going until they spotted the problem.  My Lord, that's
       Mr Parker's second witness statement at paragraphs 20 to
       23 and for the transcript it is at {E2/12/7}.
           Now, contrast the work that Dr Worden did with the
       work that Mr Coyne did.  In the first joint report
       before the experts even did their work on -- produced
       their reports, he said that he was going to look for
       bugs.  And the reference is {D1/1/3-4}.  Now, of course
       both parties were looking for bugs, but Dr Worden, as
       I say, found more branch-affecting bugs than Mr Coyne
       did with lasting effects.  But, my Lord, Dr Worden tried
       to do more than that.  He tried to take a balanced view
       of how Horizon operated in practice and how well its
       countermeasures operated in fact and that is something
       that Mr Coyne didn't do and as a result it shows a want
       of quality of his analysis, we would submit.
           Now, let's look at the bugs in issue.  Could I ask
       your Lordship to go to bundle {D1/2/3} which is the
       joint statement, the second joint statement.  If we
       could pick it up at page 3, my Lord.  Now, it is a long
       table.  It looks as if the experts are moving away from
       each other, but in fact it's a very useful list.  This
       represents the first opportunity the experts have had to
       identify where they agree and where they disagree on
       these bugs and to explain what they say about them.  I'm
       not sure how we would have managed if a document like
       this hadn't been produced.
           It is worth bearing in mind how we arrived at this
       list of 29.  Your Lordship will see the heading "Table
       of bugs/errors/defects with acknowledged or disagreed
       evidence of financial impact".  Mr Coyne's first report
       didn't specifically look at branch-affecting bugs.  His
       report was more a catalogue of bugs of all sorts and all
       shapes and sizes.  It was a scattergun analysis.  A few
       of the bugs that he identified at great length were
           Now, it was Dr Worden who looked for
       branch-affecting bugs and in his first report he noted
       what Mr Coyne had done and what he had not done.  The
       result was that Mr Coyne responded in his second report,
       which starts at {D2/4/1} and in his second report -- and
       my learned friend took you to it this morning -- at
       paragraph 3.21 of that report he set out a table
       consisting of 22 bugs, of which 21 were said to be
       branch-affecting and, my Lord, the reference is
           Several of those bugs had not been seen before, had
       not been mentioned before.  So in JR2, this was the
       first opportunity that Dr Worden had to comment on and
       address those bugs and through discussion between the
       experts, seven additional bugs were added to the list.
       Bugs 23 to 26 are bugs that are referred to in
       paragraph 742 of Dr Worden's first report, the reference
       to which is {D3/1/170} and these are identified as
       potentially branch-affecting bugs and those were added,
       they were not already part of Mr Coyne's 22 bugs.
           Bugs 27 to 29 were added as a result of discussions
       between the experts.
           So if we look at the heading of the table, bugs with
       acknowledged or disagreed evidence of lasting financial
       impact {D1/2/3}.  So I think we are intended to take it
       that this is a table consisting of 29 bugs which
       Mr Coyne says all have financial impact.
           It is acknowledged that there is evidence in 12
       cases for lasting financial impact and the ones where
       they are acknowledged, my Lord, just for your note, it
       is bug 1, which is a receipts and payments mismatch and
       your Lordship will see from Dr Worden's comments that
       a subpostmaster report wasn't necessary for that bug.
       Bug number 2, that's the Callendar Square bug.  Bug
       number 3, that's the Suspense Account bug.  And,
       my Lord, not Dalmellington because that would have been
       picked up as a normal remming reconciliation as night
       follows day.
           By the way, contrary to the suggestion that the
       claimants at points make in their submissions, Dr Worden
       did analyse that bug in his first report.
           Next is bug number 10 which is a Data Tree build bug
       which occurred very early in 1999, right at the
       beginning of Horizon.  It was very noticeable, says
       Dr Worden, and very quickly fixed.
           Bug number 13, "Withdrawn stock discrepancies" and
       it may be worth noting that Dr Worden's opinion that it
       wouldn't have lasted very long.
           Bug number 14, bureau discrepancies.  Bug number 18,
       concurrent log-ins.  Again that's a very early one in
       1999, 1999/2000.  Bug number 23, bureau de change.  Bug
       number 24, wrong branch customer change.  Bug number 25,
       like a top-up.  Bug number 27, TPS.  And bug number 28,
       drop and go.
           My Lord, it is worth noting that there is a link to
       JR2.  I would invite your Lordship to look at that link
       in due course.  It is a link in which Dr Worden includes
       an assessment of the financial impact of those 12 bugs
       with his own base estimate and then a more generous,
       a more conservative estimate -- he describes it as
       conservative anyway -- favouring the claimants.
           In fact could we look at that.  It is the link to
       JR2 which is {D1/2/1}.  It might not be possible,
       perhaps not.  Your Lordship can do it on Magnum anyway,
       your Lordship will see there is a link --
   MR JUSTICE FRASER:  Is that the one-page document?
   MR DE GARR ROBINSON:  It is two pages.
   MR JUSTICE FRASER:  Is it?  I thought it was one page.
   MR DE GARR ROBINSON:  My Magnum seems to be disappeared.  In
       any event --
   MR JUSTICE FRASER:  The mean impact spreadsheet?
   MR DE GARR ROBINSON:  Yes and then there is a second page
       which is his comments on it.  So if your Lordship looks
       at the comments he says:
           "This is an estimate of the mean financial impact on
       all branches across the Post Office network ..."
   MR JUSTICE FRASER:  The link is just coming up with the
   MR DE GARR ROBINSON:  I see.  There should be a second tab
       to the spreadsheet.  Is that not there?
   MR JUSTICE FRASER:  I'm not sure that's loaded on, because
       I did have a look for this.  I have seen this and
       I haven't seen any text that goes with it.
   MR DE GARR ROBINSON:  I downloaded it from Magnum yesterday
       but it may be that it hadn't quite got --
   MR GREEN:  There is an "Explanation" tab, if you mean the
       one on there.
   MR JUSTICE FRASER:  I see.  You can't see it at the bottom
       of the screen.
   MR DE GARR ROBINSON:  I see.  Does it go on forever?
   MR GREEN:  No, the "Explanation" tab.
   MR JUSTICE FRASER:  I've got it now.  Is that the one -- it
       has 13 cells with text in and the first says "This is an
   MR DE GARR ROBINSON:  Yes.  Can one click on the explanation
       of calculations on the second sheet?
   MR JUSTICE FRASER:  I think on Magnum the tab has dropped
       off the bottom of the common screen but there is
       a second page -- well, I've got mine up anyway,
       Mr De Garr Robinson.
   MR DE GARR ROBINSON:  I'm very grateful to your Lordship.
           Your Lordship will see the first paragraph Dr Worden
       explains that he has estimated the mean financial impact
       and he has given his central estimate in relation to 12
       bugs.  Second paragraph he says:
           "For each of these bugs the table contains a very
       approximate estimate of the financial impact of the bug
       if that impact had not been corrected by Post Office in
       some way ..."
           And your Lordship will appreciate I say that's a big
           "... and my estimate of the probability that any
       financial impact would have been corrected.  My central
       estimate of impact on the SPM reflects that probability
       whereas my conservative estimate does not."
           Then he says:
           "As can be seen from the table, my opinion is that
       the larger financial impacts would have been corrected
       with high probability.  Smaller financial impacts might
       not have been corrected."
           If your Lordship goes back to the table
       your Lordship will see there is a column -- the 12 bugs
       are identified that I have listed to your Lordship with
       the associated KELs where there is one.  There's an
       estimate of financial impact column.  Then there's
       a probability that the postmaster was compensated column
       and then there's Dr Worden's estimate of loss to
       postmasters and then on the final column there's a loss
       to postmasters assuming that none of them were made good
       during reconciliation or other processes.  So
       your Lordship sees --
   MR JUSTICE FRASER:  When you say "made good" though I think
       you need to be clear what you mean because "made good"
       has a technical meaning in terms of the accounting.
       I think you mean remedied.
   MR DE GARR ROBINSON:  I mean where there is a financial
       discrepancy caused in the accounts that discrepancy is
       corrected by some kind of financial transaction.
   MR JUSTICE FRASER:  Yes.  Because making good under the
       accounting system meant the SPM paying it in.
   MR DE GARR ROBINSON:  I'm so sorry.  You are ahead of me.
   MR JUSTICE FRASER:  That's why I'm just checking.
   MR DE GARR ROBINSON:  It is the common issues trial
       nomenclature that I have to be careful of.
   MR JUSTICE FRASER:  But that's what you meant when you
       explained how you did just then.  I understand.
           Your Lordship will see that Dr Worden's previous
       estimate had been £1,000.  He increased that to £2,029.
       That's per SPM.  But in the other column your Lordship
       will see "Total impact" of all these bugs was £165,000,
       mean impact per SPM who was affected, £13,800 and that
       compared with his previous estimate of £6,000.  So the
       figures have gone up a bit and as he explains in his
       commentary it is mainly because of the Data Tree build
       bug, the 1999 bug, of £105,000.
           Your Lordship will see it is those sort of figures
       that then lead one to the calculation I explained to
       your Lordship before which leads one to a requirement of
       tens of thousands of these bugs in order to have any
       hope of being responsible for even a relatively small
       portion of the £18.7 million claims made by the
           So that's the agreed bugs.
           As regards the disagreed bugs, a number of short
       observations that I would like to advance to
       your Lordship.
           First of all, Mr Coyne refers to bugs where there
       was no bug at all in the PEAK he has cited.  This is
       something we will explore with him.  For example, bug
       number 17 relates to branch customer discrepancies in
       Legacy Horizon.  This was not caused by a bug in
       Horizon.  The system crashed part way through
       a transaction for an unknown reason and then it was
       successfully detected and the missing transaction was
       recovered.  I explained to your Lordship before how the
       recovery system works to identify transactions that may
       have been missed.  That's not an example of a bug,
       that's an example of Horizon working.  And it is worth
       noting, my Lord, that this issue was spotted not as
       a result of any human intervention by a subpostmaster,
       it was spotted as a result of an automated report that
       was available to Fujitsu.
           My Lord, secondly, Mr Coyne includes bugs, we say,
       which were only found in testing and didn't make it into
       the live system.  An example of that is bug number 21,
       dealing with transaction corrections issues.  He makes
       a reference to one particular bug referring to a PEAK at
           The important point to note about this bug is that
       it was eradicated before the relevant software even
       reached the live system.  It was spotted in testing.
       That's what testing is for.
           My Lord, a third point observation to make is that
       Mr Coyne includes bugs which have automatic fixes which
       Mr Coyne himself acknowledges.  I have referred to
       remming in my opening already.  Just one example would
       be bug 5.  It was a remming in issue.  Mr Coyne's own
       analysis correctly records that this could cause
       a duplicate pouch of cash to be remmed in by the user
       causing a shortfall, but what he doesn't say is that
       Post Office's reconciliation processes check the pouches
       sent out by Post Office and pouches received by the
       relevant branches, an automatic report is run which
       defects all sort of remittance discrepancies whether
       caused by user error, bugs or otherwise and then
       transaction corrections are generated to correct the
       problem.  So it may be a bug, but, my Lord, it's a bug
       which in the ordinary course of things is almost never,
       if ever, going to result in a lasting loss to any
           Finally, there are some bugs which both experts
       agree had no financial impact, or at least one bug.
       My Lord, that's bug 21.  So perhaps the heading to the
       table in the joint statement should be corrected.  It's
       not 29 bugs that are said to have financial impact, it
       is 28.
           But whether it is 12 or 28, my Lord, does not
       matter.  In a system handling the volumes of transaction
       that Horizon handled, which was used by the number of
       branches that used it, over a period of 20 years, bugs
       of this number and scale can have no material impact on
       the overall robustness of Horizon, or putting the point
       more pertinently, the chances of one of these bugs
       affecting a given set of accounts is vanishingly small.
       Even if you sum them all up, the chances of any one of
       these bugs affecting a given set of accounts is
       vanishingly small.
           So what are we left with?  We are left with
       a suggestion that it is possible that there are other
       undetected bugs in the system which may have an effect
       on branch accounts but, my Lord, as I have said
       possibility is not the same as probability.  Mr Coyne
       does not say anything about probability.  He makes no --
       in fact he disclaims any ability to make any inferences
       about how many bugs there are, what their scale is
       likely to be and what their impact is likely to be.
           Now, in my respectful submission that is the
       approach which is not helpful.  Mr Green and Mr Coyne
       describe it, in my submission rather curiously, as
       a bottom-up approach.  It might be more accurate to
       refer to it as a bottom-down approach.  Mr Coyne stays
       at the bottom, sticks just with the bugs that he has
       found and says nothing more, nothing about the
       implications of those bugs on the overall robustness of
       the Horizon system.  He refuses to say anything on that
       question, so we are left with his agreement that the
       Horizon system is robust.
           So he doesn't grapple with the critical question
       whether the court should infer or presume that the
       shortfall is shown in any given branch accounts was
       caused by a bug, my Lord, and it is not a realistic
       approach either.  My Lord, it is common ground that if
       you have a bug it may well have a range of impacts, some
       may be smaller, some may be larger, some may affect more
       than one branch, but in the scheme of things where you
       have a bug which has different effects on different
       people, one has to ask oneself what are the chances of
       that bug evading any detection by anyone as a result of
       any of its impacts at any time?  That's always going to
       be relatively low.  And if one were to postulate a tiny
       bug which had an impact of a few pence, my Lord, there
       would need to be millions of those bugs in order to
       begin to be significant, to have a significant bearing
       on the robustness of Horizon.
           In the real world we are concerned with bigger bugs,
       with a wider range of impacts, such as the
       Suspense Account bug and I ask forensically or
       rhetorically: how likely is it that there are tens and
       tens of thousands of such bugs lurking in the system,
       escaping detection for year after year?  In my
       submission the answer is obvious.
           My Lord, before -- I perhaps have a few minutes just
       to address one point that has a bearing on robustness
       before moving on.  Your Lordship will see both in the
       claimants' submissions and in Mr Coyne's report
       frequently repeated assertions that Post Office operated
       in accordance with the cost-benefit analysis.  I think
       the impression is sought to be achieved that because
       Post Office addressed some problems by adopting
       a cost-benefit analysis, this means that the procedures
       in place for dealing with or identifying bugs and fixing
       them and so on were not robust, that there was some kind
       of threat to the system because Post Office was cutting
       corners or something like that.  But, my Lord, one needs
       to be careful with the evidence that is cited in support
       of these claims.  Many of the claims of this sort rely
       on two or three pieces of paper, no more than that.
           It is worth looking at one example of the evidence
       that's relied on in this case.  My Lord, it is the claim
       in paragraph 144.1 of the claimants' submissions
       {A/1/50}.  I don't know if your Lordship has a hard copy
       of it there.
   MR DE GARR ROBINSON:  Your Lordship will see the claim made
       in 144.1:
           "It has been identified that known issues/bugs were
       often deferred and dealt with on a cost-benefit basis."
           So your Lordship will see the point and
       your Lordship will see exactly what they are trying to
       make of the point.
           Footnote 105 refers to a paragraph in Mr Coyne's
       first report, paragraph 5.161.  That paragraph refers to
       a document which is at {F/11/40}.  This is minutes of
       a risk and compliance committee on 16 September 2013 and
       your Lordship will see the attendees.  If I could ask
       your Lordship to go forward --
   MR JUSTICE FRASER:  Were any of those attendees legally
   MR DE GARR ROBINSON:  I'm so sorry?
   MR JUSTICE FRASER:  Are any of those attendees legally
   MR DE GARR ROBINSON:  My Lord, I'm not in a position to
       answer that question.
   MR JUSTICE FRASER:  Maybe you could just find out because
       that might affect what I do in terms of ordering
       a review of the redactions, that's all.
   MR DE GARR ROBINSON:  My Lord, I should tell your Lordship
       that questions have been raised about redactions and
       a further review has been undertaken by my instructing
       solicitors and also by junior counsel --
   MR JUSTICE FRASER:  It has been done already, has it?
   MR JUSTICE FRASER:  All right.  I assume the result of that
       review was that they were properly redacted?
   MR DE GARR ROBINSON:  The result of the review is that there
       are a relatively small number of cases where a different
       judgment call has been made, usually where one can see
       why a different decision was made, but in the main the
       redactions have remained.
           My Lord, I can tell your Lordship that
       Susan Crichton was general counsel.
   MR JUSTICE FRASER:  Susan Crichton.
   MR DE GARR ROBINSON:  Who is the first attendee.  In fact
       the chair.
   MR JUSTICE FRASER:  Can you just -- not now, because I don't
       want to know now, but can you just identify for me on
       how many separate occasions the review led to disclosure
       being made where they had previously been redacted,
       following the review you have just told me about.  Not
   MR DE GARR ROBINSON:  I will find out.
           If we could go forward to page 3 {F/1140/3} the
       document says:
           "It was reported that following the recent Ernst &
       Young external audit four risks [have] been identified.
       Three of the risks raised had been addressed, however
       the final risk, relating to the communication by Fujitsu
       of changes made to the Horizon system, was still
           "It was identified that it would cost over
       £1 million to implement the mitigation being suggested
       by the audit and that this was not proportionate to the
       risk being managed."
           And the decision made was that:
           "The committee agreed that the risk be accepted with
       Dave Hulbert as the owner and Lesley Sewell being
       ultimately responsible."
           My Lord, this is evidence that is asserted by
       Mr Coyne and adopted by the claimants as evidence that
       known issues/bugs were often deferred and dealt with on
       a cost-benefit basis.  So you will see how high the
       claim is put, but it is worth looking at the Ernst &
       Young report that's referred to.  My Lord, that's at
       bundle {F/1127}, or rather a committee meeting which
       considers the Ernst & Young recommendations.  This is
       a risk and compliance committee meeting relating to the
       acceptance of risk following Ernst & Young audit of
       2012/2013.  "Purpose":
           "The purpose of this paper is to:
           "Update the risk and compliance committee to a risk
       that IT&C have 'accepted' following the 2012/13 Ernst &
       Young IT audit.
           "2.1.  The 2012/13 Ernst & Young IT audit found no
       significant exceptions but did identify a small number
       of improvement opportunities.  Four high level
       improvement opportunities were recorded.  Three have
       progressed and are either complete or in the process of
       completing.  For one, IT&C believe we have sufficient
       process and mitigation in place to accept this risk.
       This paper is to highlight this decision to the Risk &
       Compliance Committee."
           And paragraph 2.2 reads:
           "The specific observation was with regard to change
       management monitoring control.  The actual observation
       read 'management should make use of a system-generated
       list of changes in performing the monitoring control to
       further enhance its effectiveness'."
           "The risk being that changes may be made to the
       system that are not approved and not found through
           So it's a process suggestion, my Lord, it's not
       something that has any bearing on the quality of any
       changes to the system.
           "The Post Office service management team currently
       monitor IT system changes on a monthly basis by
       cross-referencing known and approved changes against
       a list produced by Fujitsu.  E&Y observed that this
       could be enhanced if the list was generated by the IT
       system rather than by change records.
           "IT service management engaged with Fujitsu to
       understand how this could be achieved and it was
       concluded a very difficult and potentially expensive
       approach to adopt this as all changes are recorded as
       'events' within the IT system of which there are
       multiple thousands per day with changes only being
       a small percentage.  The cost and difficulty in
       extracting these specific change events on a regular
       basis would outweigh the value in monitoring the
           So the options considered were:
           "1.  Continue with the existing approach of using
       a list generated by the change records."
           "2.  Develop an approach with Fujitsu to generate
       a list from the IT system of events."
           And then over the page, my Lord {F/1127/2} the
       proposal at paragraph 5:
           "To continue with the existing process of monitoring
       but to additionally raise this as a risk within IT&C and
       to monitor any exceptions found through the existing
       process.  If exceptions are found then reconsider the
       proposal from E&Y and assess the impact of the change
       versus the benefit."
           And, my Lord, if we then go down to paragraph 7 "Key
           "There is a risk that some changes may not be
       monitored using the existing process.  In mitigation of
       this, the current process has never uncovered any
       changes that have not been monitored and that is
       supported by the annual IT audits by Ernst & Young."
           So, my Lord, that's the decision.  That explains
       what the proposal was, how carefully it was considered
       and what the conclusion was.  That is not evidence for
       the proposition that when a bug arises, quite often what
       Post Office and Fujitsu would do was do nothing because
       it costs them money.  Your Lordship will bear in mind
       Mr Parker's witness evidence in which he said that bugs
       with a branch impact were always treated as high
           So, my Lord, obviously I have not made my case just
       simply by referring to one document, but I would invite
       your Lordship to be cautious when dealing with
       submissions of that sort from the claimants and indeed
       from Mr Coyne.
   MR JUSTICE FRASER:  We are probably going to have to have
       a break.
   MR JUSTICE FRASER:  Is now a good time?
   MR DE GARR ROBINSON:  It is a perfect time, my Lord.
   MR JUSTICE FRASER:  We will come back in at 25 past and
       don't forget at the very end we just have to deal with
       that timetable point.
   (3.15 pm)
                          (Short Break)
   (3.27 pm)
   MR DE GARR ROBINSON:  My Lord, having dealt with the
       operational issues and robustness I was now going to
       move as quickly as I could to remote access.  To save
       time I'm proposing to take your Lordship to some
       relevant paragraphs in our opening submissions and for
       that purpose I would invite your Lordship to go please
       to paragraph 40 of those submissions, which is at
       {A/2/17}.  This is a context point.  It's a really
       really important context point.  It's a point that
       applies to transaction corrections as well, but for
       present purposes I'm making it in relation to remote
       access.  It is the second order issue that your Lordship
       I'm sure will be familiar with now.  At paragraph 40 we
           "On a number of issues on which Mr Coyne places
       considerable emphasis in questioning Horizon's
       robustness, Mr Coyne overlooks that they are second
       order issues that can only have minimal effect.  At best
       they represent a very small fraction of a very small
       fraction of any cases.  Remote access is one such issue.
       Transaction corrections are another."
           Paragraph 41 reads:
           "Taking remote access as an example, the need for
       remote intervention affecting branch accounts will
       obviously be rare.  On any view, the occasions on which
       privileged users at Fujitsu have exercised their ability
       to remotely inject, edit or delete branch transactions
       or accounting entries will represent a tiny percentage
       of the relevant transactions/accounting entries.  And
       the occasions on which they have done so negligently or
       dishonestly will, in turn, represent a very small
       percentage of those occasions.  So, compared with the
       volume of business recorded in branch accounts, the
       number of cases in which false data will have been
       remotely introduced will be extremely small (multiplying
       a small chance by a small chance).  This is a 'second
       order effect' ... which is, by definition, extremely
       unlikely to have any significant impact on the
       robustness of Horizon."
           My Lord, if we take a step back and think again
       about the 3 million sets of monthly branch accounts that
       have occurred over the last 20 years, I ask rhetorically
       how many interventions, insertions, deletions, editing
       would one need into branch accounts -- and I emphasise
       into branch accounts, not other kinds of remote access,
       we're talking about interventions which have an impact
       on branch accounts, transactions done in the branch or
       stock or cash held at the branch, those kind of things,
       how many of those interventions would one need in order
       to generate a material risk that any given branch's
       accounts are likely to be wrong?  You would need
       hundreds of thousands because the chances of any
       intervention actually being wrong -- all human
       intervention is subject to error so there will be an
       error rate, or there might be an error rate, but it's
       going to be a relatively low error rate on any view, so
       one would need hundreds of thousands of those even to
       produce the number of interventions that could have even
       a small impact on the general body of branch accounts.
       The numbers would have to be unimaginably huge for this
       to have any material impact on the issues with which
       your Lordship will be wrestling in the Horizon issues
           It is hard to overstate the importance of this
       point, my Lord.  These remote access issues are here
       because of concern expressed that Fujitsu -- and we are
       only talking about Fujitsu, who are professionals -- was
       using its administrator rights in such a way and on such
       a scale as to undermine the reliability of the
       claimants' branch accounts.  And it is vital to
       understand, in my submission, that in the real world
       this is not going to be an answer to that question, this
       is not going to have a bearing on the robustness of the
       system and on the reliability of any given branch
       accounts in any given month.
           So, my Lord, that is a first and in my respectful
       submission critical point of context that should always
       be borne in mind.  There's a sense of unreality.  It is
       quite understandable, the issues here are quite
       eye-catching, my learned friend quite understandably has
       developed his submissions as to how the history of this
       issue has arisen and how matters have been revealed in
       the course of time, one quite understands why he has
       done that, but when looking at the history of the case
       and when looking at the witnesses and what they say
       about remote access it's really important to bear in
       mind in my submission that even if the claimants' case
       succeeds at its highest, it's never actually going to
       have a material impact on the reliability of branch
       accounts.  That's just not what this activity is about.
   MR JUSTICE FRASER:  Now, this is the same point I think, is
       it not, that where the dispute used to be this couldn't
       be done, now the dispute is in theory it can be done but
       in practice it never would have been; is that a fair
   MR DE GARR ROBINSON:  No, my Lord, I would put it as
       follows.  If your Lordship would like me to address on
       what --
   MR JUSTICE FRASER:  No, it is just so that I understand
       firstly that it's the same point that Mr Green developed
       along those lines and secondly if it is the point which
       I accurately understand the way you are putting it, it
       is that Fujitsu did have the ability to do that but in
       practice never would.
   MR DE GARR ROBINSON:  No, my Lord, I'm going to take you to
       this in a few minutes, but let me -- first of all one
       has to distinguish between Legacy Horizon and
       Horizon Online.
   MR DE GARR ROBINSON:  Horizon Online, there has been one
       occasion when a transaction was inserted into one
       branch's accounts at the BRDB, one occasion, and
       my learned friend wants it to be more, so does Mr Coyne
       and there will be evidence given on that question.
           In relation to Legacy Horizon it is different.
       There's a complication because whereas with
       Horizon Online access was made to a central database, or
       server I should say; with Legacy Horizon the data in the
       first instance was maintained on server's held at the
       branch, or what were called counters.  So there there
       was an element -- there was a greater complication
       because counters held data and counters could go wrong
       and they would need backups.  Transactions could
       sometimes be inserted into branch accounts when there
       had been problems.  It's a very different process from
       the position in relation to Horizon Online and I'm not
       seeking to give you a full account of Legacy Horizon,
       that's in the evidence and I don't want to be taken to
       be summarising it.
           There were transactions insertions made into
       counters on occasion.  There were also occasions when
       counters broke and their data had to be basically backed
       up from another source.  It was called replicating from
       another source.  And there were variants of those two
       processes.  My Lord, those were occasions therefore when
       it could be said that what we mean by remote access took
           So I don't say remote access never took place but
       what I say is, take the claimants' case at its highest,
       how many occasions of remote access of that sort in
       Legacy Horizon would there have to be in order to
       generate a material risk to the reliability of any given
       set of branch accounts?
   MR JUSTICE FRASER:  I understand that point and I understand
       that's your case and in a way that is central to the
       different way in which the two parties are approaching
       it and I'm not in any way suggesting that one of those
       ways is preferred at the moment.  All I was doing was
       seeking to establish -- all it comes down to really,
       Mr De Garr Robinson, is are you and Mr Green using the
       term, or using your broad categorisation in the same
       way, that's all.
   MR DE GARR ROBINSON:  No we're not and that's one of the
   MR JUSTICE FRASER:  That then brings me on to my next point.
       You don't appear to be using the terms in the same way.
   MR DE GARR ROBINSON:  That is one of the problems.
       My learned friend -- and I think Mr Coyne also -- adopts
       a construction of the phrase "remote access" which is
       unimaginably wide.  It includes, for example,
       transaction corrections, because transaction corrections
       are a process by which Post Office presses a button in
       some office somewhere and something pops up on the
       screen and then a postmaster has to decide whether or
       not to allow that change, which is specifically notified
       to the postmaster to allow that change into his or her
       accounts.  My learned friend treats that as an example
       of remote access.  That's not how I understand it at all
       and indeed it is not how I understand the pleadings.
       But it shows your Lordship the width of -- I would say
       the lack of utility of the definition that he is seeking
       to propose.
           I on the other hand have a more practical approach
       which is what's this question about?  It's about whether
       or not people remotely could make changes to branch data
       such that their branch accounts would be wrong.  There
       are all sorts of things that could be done remotely that
       would have no impact on branch accounts.  You might have
       a counter for example in Legacy Horizon that was locked,
       there was a binary bit that was on a zero instead of a 1
       and it needed to be turned to a 1.  There was a process
       by which Fujitsu could unlock the locked counters,
       locked items of that sort.  That had no conceivable
       impact on branch accounts.  Now, one could call that
       remote access if one wanted to, but it's not a relevant
       form of remote access.
           When I talk about remote access I'm talking about
       action taken remotely to either inject new transactions
       or to edit existing transactions or to delete existing
       transactions in a way that could change the accounting
       position of the relevant branch.
   MR JUSTICE FRASER:  And by Fujitsu only?  Because I think
       you started this passage by saying only Fujitsu.
   MR DE GARR ROBINSON:  Yes, only Fujitsu.
   MR JUSTICE FRASER:  That's very helpful because that defines
       for me the way you are using the term.
   MR DE GARR ROBINSON:  I'm grateful.
   MR JUSTICE FRASER:  All right, thank you very much.
   MR DE GARR ROBINSON:  So your Lordship has the second order
   MR DE GARR ROBINSON:  As I say, in my submission it
       overwhelms everything else.  It makes the next
       40 minutes really rather -- I don't want to say
       irrelevant, that's disrespectful, but of marginal
           My Lord, further points of context I would ask
       your Lordship to note.  If I could move on in the
       submissions to page 83, paragraph 243 {A/2/83},
       paragraph 243, headline point, my Lord, irrelevance of
       or immateriality -- is that a word, I'm not sure -- of
       the suggestion of some master criminal sitting in
       Fujitsu's offices manipulating branch data for his
       personal benefit:
           "Although claimants will probably draw attention to
       a dramatic, but only theoretical, possibility of the
       malicious alteration of transaction data by a rogue
       employee of Fujitsu, there's no evidence that this ever
       took place, nor any allegation to that effect.  It is
       unlikely to assist the court for the parties to engage
       in speculation as to what might have happened, in
       theory, if there had been some rogue employee who had
       abused his or her access rights.  If there is any
       reason, in a specific case, to think that it might have
       happened, that can of course be explored ... but there
       is no generic issue as to any malicious alteration of
       transaction data."
           So again this is a real world point, my Lord.
       Your Lordship will see speculation -- well, a discussion
       in Mr Roll's witness statements about speculation as to
       whether it might be possible to pay a gas bill or make
       a bank transfer using the transaction insertion facility
       that was available in Legacy Horizon, but that is all it
       is, it's just a theoretical possibility that need not
       detain the court for five minutes.
           And in paragraph 244 we continue:
           "The real thrust of claimants' case is that, in the
       course of using their abilities to insert, edit or
       delete data to correct problems arising in the operation
       of Horizon, Fujitsu personnel might accidently have
       introduced a further or different error into transaction
       data.  This is a clear example of what Dr Worden calls
       a second order issue ..."
           And I have addressed your Lordship on that.  As we
       say at the end of that paragraph:
           "Such second order issues are, for this reason
       alone, not viable candidates to explain any significant
       proportion of disputed shortfalls."
           So having got rid of the master criminal theory and
       having made it clear of how peripheral this issue really
       is, let's then move on to the position and let's move on
       to paragraph 246 {A/2/83}:
           "... the picture is complex.  There are different
       Fujitsu methods for altering different kinds of data
       remotely, and Fujitsu's current methods are themselves
       different from the tools and methods that are available
       under Legacy Horizon.  The Fujitsu witnesses have done
       their best to recollect, investigate and to some extent
       hypothesise as to what may have been possible in the
       fairly distant past.
           "The picture is also complicated by the fact that
       the parties do not agree on what counts as remote
       alteration of data."
           My Lord, this addresses the question your Lordship
       has just asked.
           For example, Mr Coyne includes transaction
       corrections.  Curiously, my Lord, it is worth noting
       that although Mr Coyne includes transaction corrections
       as a form of remote access, he excludes the predecessor
       that applied until 2005, namely error notices.  So error
       notices weren't, TCs were and the reason he gives for
       that is because TCs are electronic, that they were
       communicated to the branch through the system rather
       than being sent by post.  This to him is a major
       difference.  I say it is a distinction without
       a difference.
           Then, my Lord, paragraph 248, another red herring to
       do with global users.  I say red herring, that's perhaps
       unfair to my learned friend.  There is an issue between
       the parties as to whether global user rights could be
       exercised outside the branch.  It is quite clear that
       there were some people who did have the privileges
       needed to go into a branch and make changes to the
       branch, to use the branch machines if that's what they
       wanted to do.  That's common ground.  There's an issue
       as to whether it was possible to use those rights
       remotely, go to somewhere else from some kind of central
       office at Fujitsu and use your global user rights to
       make changes to the branch.  My Lord, Fujitsu's evidence
       is that that is not possible, but it is challenged by
       Mr Coyne and the claimants, so that's an issue that
       your Lordship will have to decide.
           Then, my Lord, another important contextual point is
       in paragraph 249 {A/2/84}.  There are complications of
       terminology and this is to do with different kinds of
           "Mr Coyne discusses the deletion of session data and
       the deletion of transaction data as though these were in
       essence the same thing.  That is misleading.  Horizon
       holds more than just transaction data; it stores and
       tracks a lot of other forms of data that are needed to
       make the system operate properly, including data that
       does not relate to customer transaction and which does
       not affect a branch's accounting position.  Session data
       is a good example of this: it may well not include
       transaction data at all; it can consist entirely of
       other types of data associated with a counter session
       (such as the data 'flag' that determines whether a stock
       unit is locked or unlocked and available for use).  The
       deletion of data of that kind does not affect branch
       accounts and cannot create a discrepancy.  It merely
       affects the availability of the system to a [postmaster]
       or an assistant wishing to perform transactions or
       conduct other branch business."
           So again it is very important to be clear about what
       kind of data we are talking about, when we are talking
       about editing, deleting or inserting data, what kind of
       data are we talking about, because there are all sorts
       of irrelevant data the ability to change which is
       utterly benign and indeed is important, an important
       part of the system and the system wouldn't work without
       it.  So that's an important distinction as well.
           Having made those contextual points clear, my Lord,
       let me just say a very few things.  Most of these points
       will be clear to you already.
           First of all, we are not really concerned about the
       reading of data, we are concerned about the addition,
       editing or deletion of data, obviously.
           Secondly, the position is very different now from
       what it was when Legacy Horizon was in operation about
       nine or ten years ago.  Horizon Online's data, all the
       data is written directly to the BRDB, the central
       databases maintained I think at Fujitsu's offices, but
       with Legacy Horizon data was held on counters before
       being downloaded to the central data services, I think
       as I recall, at the end of the day.  So where remote
       access is concerned it was possible that access could be
       obtained to counters until 2010, but it's different now.
       It's important to understand that.
   MR JUSTICE FRASER:  Well, I can understand why you make that
       submission and it is important, but I have to consider
       Legacy Horizon and Horizon Online.
   MR DE GARR ROBINSON:  Oh, of course.  I'm not suggesting any
           Thirdly, I should mention the way in which the
       evidence has developed in relation to Legacy Horizon.
       I will be absolutely frank with your Lordship, I wish it
       hadn't developed in quite the way that it has, but we
       are talking about the state of affairs that persisted
       during 2000 and 2010, it's a long time ago.  The
       system -- there is no Legacy Horizon system still in
       operation that people can go and check, that people are
       still operating.  People are working off their
       recollections and off design documents that are now very
       elderly and that problem has been compounded by the fact
       that when Mr Roll made his witness statement in 2016 --
       it was provided in September last year -- the things
       that he said in that witness statement were quite hard
       to follow and we will see how hard they were to follow
       when he gives evidence later on this week.  The result
       is that the responsive evidence by Fujitsu witnesses
       wasn't as focused as it should have been and wasn't as
       clear as it should have been and wasn't as accurate in
       certain respects as one would like it to have been and
       I must face up to that completely, but what I do suggest
       to your Lordship is that it wouldn't be right to draw
       any great inferences as a result of the way in which
       this the evidential position has developed.
           I make no criticism of Mr Roll for giving his
       evidence in the way that he did, he is not being blamed
       for this process, it is simply a fact of the matter,
       it's simply what happened and it is only right that
       your Lordship should be aware of that.
           Against that background, my Lord, it may be helpful
       to have a summary of what I would include as remote
       access in this case and, my Lord, that's in
       Mr Godeseth's third witness statement, which is {E2/14}.
       I would like to pick it up at page 4 of that witness
       statement {E2/14/4}.  He says at paragraph 14:
           "Having further explained that global users and the
       TIP repair tool cannot insert, inject, edit or delete
       transaction data remotely, to the best of my knowledge,
       the following types of remote access as defined in
       paragraph 3 above, are or have been possible:
           "14.1.  Privileged users could, theoretically,
       inject, edit or delete transaction data in
       Legacy Horizon ..."
           "As far as I am aware", he says, "this has never
           "Members of the SSC could inject transaction data
       into a branch's accounts in Legacy Horizon."
           And then, my Lord, when we are still talking about
       Legacy Horizon it may help to jump forward to
       paragraph 14.5:
           "In Legacy Horizon ..."
           And Legacy Horizon alone:
           "... Fujitsu could cause data to be rebuilt from
       copies of the same data as described ..."
           In Mr Parker's second witness statement.
           So this is a simplified summary.  There are
       complications particularly in relation to the rebuilding
       of data referred to in 14.5, but your Lordship has
       a sense of the three basic categories of remote access
       that we say was possible in Legacy Horizon and, my Lord,
       as regards remote access in Horizon Online, at
       paragraph 14.3:
           "Privileged users can, theoretically, inject, edit
       or delete transaction data in Horizon Online.  As far as
       I am aware, this has never happened."
           And 14.4:
           "Members of the SSC can inject additional
       transactions into a branch's accounts in Horizon Online
       using a designed piece of functionality called
       a balancing transaction (BT).  Audit records show that
       this has happened once."
           So your Lordship will see the scope of the dispute.
       It is disputed by the claimants that there's only ever
       been one balancing transaction in Horizon Online.
       I believe it is not accepted by the claimants that
       privileged users didn't use their privilege user rights
       to muck about with branch accounts, if I can put it that
       way, but I'm not sure about that.  I'm not sure what
       their positive case is.  I think their case may be just
       that it was possible, not that it actually happened.
           And then in Legacy Horizon it was possible to inject
       transactions and that happened more frequently than has
       happened with balancing transactions and it is possible
       to rebuild data from machines which had problems.
           So is that a helpful summary of what I say --
   MR DE GARR ROBINSON:  I'm grateful.
   MR JUSTICE FRASER:  Thank you.
   MR DE GARR ROBINSON:  Now, Mr Roll deals with the
       Legacy Horizon position.  He worked in third line
       support in the software support centre at Fujitsu.  In
       that capacity he could make transaction insertions and
       he could rebuild data.  But his evidence is quite
       unclear about this, or at least his evidence in his
       first witness statement is not very clear and it will
       need to be unpicked as to what data he is talking about.
           Your Lordship will have seen in paragraph 14 there's
       a reference to the TIP repair tool, the transaction
       information processing repair tool.  My learned friend
       addressed your Lordship on that this morning.  That's
       addressed by Mr Godeseth at paragraph 10.  I say it is
       a red herring.  Paragraph 10 {E2/14/3} he explains how
       Mr Coyne is confusing balancing transactions with
       exercises of the TIP repair tool and then he explains
       why the TIP repair tool doesn't have an impact on branch
           ""The TIP repair tool (which was available in
       Legacy Horizon and is available in Horizon Online) is
       used on data that has failed validation on the transfer
       between the BRDB and the TPS system on Horizon Online
       and is therefore quarantined within TPS.  I understand
       from speaking with colleagues that it served a similar
       role in Legacy Horizon in relation to data moving
       between the Riposte Message store and the TPS system."
           And, my Lord, this is important:
           "The TPS system is used to transfer data out of
       Horizon and on to other external systems.  The TPS
       system (in either Horizon Online or Legacy Horizon) does
       not hold or generate data that is used to produce
       a branch's accounts from a subpostmaster's perspective.
       Accordingly, an error or change in TPS data will not
       affect a branch's accounting position.
           "The TIP repair tool is used where the format or
       content of the data output from Horizon is incompatible
       with the systems to which it is being delivered.  For
       example, a system may require that certain data fields
       are populated.  If these criteria are not met, the
       receiving system may reject the data.  The TIP repair
       tool is used to correct these incorrect or missing
       attributes.  The correction does not change the core
       information about the transaction."
           And then there is a reference to a PEAK which shows
       mandatory fields were omitted from four messages and the
       TIP repair tool was used to insert suitable values:
           "These are all time stamps so have no impact on
       accounts, but the receiving system expects the fields to
       be there.  In practice, there are multiple time stamps
       in messages, so other, appropriate time stamps would
       have been used (which may differ by a few seconds from
       the missing one)."
           My Lord, here is an important point:
           "The changes are made to the data in the TPS system
       not in the BRDB (or the Riposte Message store in
       Legacy Horizon)."
           Your Lordship has already heard submissions about
       front of the system and the back-end of the system.
       Branch accounts are in the front end of the system, they
       are within Horizon, Horizon generates accounts for the
       branch.  The data that is in the Horizon system then has
       to be extracted from the Horizon system into the
       back-end systems, so Post Office and Fujitsu -- mainly
       Post Office -- so that it can be used by Post Office for
       its accounting functions.  The most important accounting
       system that Post Office currently uses is called POLSAP.
       It deals with all its internal financial matters from
       paying rent on buildings to salaries and also movements
       resulting from Horizon transactions.  POLSAP draws data
       down from the front end.  The important thing is that
       the TIP repair tool changes the data that is taken from
       Horizon and put into POLSAP, but changing data in POLSAP
       would not change anything in the branch accounts.  It's
       just -- they're just different things.  My Lord, the
       accounting systems are separate.  The only way in which
       POLSAP could be used in such a way as to result in
       a change to branch accounts would be if Post Office used
       POLSAP to decide to issue a transaction correction,
       Post Office then issued a transaction correction to the
       branch and the postmaster accepted that transaction
       correction.  That's the only way it could happen.
           So it is really important to understand the
       fundamental difference between facilities which enable
       changes to be made to back-end systems and facilities
       which enable changes to be made to branch accounts.
       And, my Lord, that is overlooked in the claimants'
       submissions.  A good example of that is at paragraph 273
       and perhaps we could look at that.  For the record it is
       {A/1/94}.  Paragraph 273, my Lord.
   MR JUSTICE FRASER:  Yes, I've got it on the common screen.
   MR DE GARR ROBINSON:  "Mr Coyne's view is that the above
       tools have the potential to affect transaction data and
       potentially branch account data by way of incorrectly
       altering the transactions prior to entering the
       recipient systems such as POLSAP and external clients
       (after processing by the counter).  The end result may
       be the issuing of a flawed TC by Post Office who may not
       be aware of the error."
           My Lord, actually the point is well made there and
       I apologise because I was suggesting this was
       a paragraph which showed that the point wasn't
       understood.  The point is that when we're talking about
       changes to back-end systems like this we're not talking
       about second order issues, we're talking about third
       order issues.  We're talking about a change to the
       POLSAP figure as a result of some intervention which is
       wrong, which results in a mistake being made by
       Post Office in deciding to issue a transaction
       correction, which then results in a mistake being made
       because the postmaster accepts the transaction and it is
       loaded into his accounts.
           One is not going to find an answer to the question
       whether Horizon is robust by referring to tools which
       can affect the data that goes into POLSAP.
           My Lord, I note that the claimants and Mr Coyne
       place emphasis on what they say is a failure to follow
       proper permission controls by Fujitsu and there is
       a reliance on an audit recommendation by Ernst & Young
       in 2011.  That is referred to again and again and again
       in Mr Coyne's report and again and again and again in
       the claimants' submissions.  My learned friend took
       your Lordship to it.  It's at {F/869} and it is referred
       to -- well, it is referred to in many many paragraphs,
       but including paragraphs 5.161 and 5.196 in Mr Coyne's
       first report, the references to which are {D2/1/97} and
       {D2/1/107}.  And it is also referred to in paragraph 282
       of the claimants' opening submissions, for the
       proposition that Fujitsu had weak user management
           My Lord, could we go to that document please.  It is
       {F/869}.  You have already seen it once.  My Lord, but
       your Lordship hasn't seen the first page, or rather --
       if it your Lordship goes to --
   MR JUSTICE FRASER:  I think you mean the second page.
   MR DE GARR ROBINSON:  It is the second page {F/869/3}.
       Paragraph 1 "Executive summary":
           "The finance leadership team at Post Office Limited
       has implemented and process improvements throughout the
       organisation during the past financial year.
           "In particular, focussed management action has
       addressed many of the issues raised in our prior year
       management letter and led to significant improvement in
       the overall payroll control environment.  The
       recommendations we have made in this report should be
       seen as refinements rather than fundamental control
       deficiencies in comparison."
           My Lord, they are put forward in the evidence as
       fundamental control deficiencies.  Again this doesn't
       win me the case but I do invite your Lordship -- when
       your Lordship sees frequent references to documents, it
       can sometimes be very beneficial to read the whole
       document carefully.  My Lord, that's my first submission
       about that document.
           The second submission I would make about the
       document is that although Mr Coyne refers to it several
       times and indeed there's a suggestion that Ernst &
       Young's recommendations aren't being carried into
       effect, he doesn't refer to a later document, Ernst &
       Young control themes and observations 2013, which is at
       {F/1138}.  If I could ask your Lordship to go to page 2
       of that document {F/1138/2} and again it is the letter
       which introduces this document, second paragraph down:
           "As part of our audit of the financial statements,
       we obtained an understanding of internal controls
       sufficient to plan our audit and determine the nature,
       timing and extent of testing performed.  Although the
       purpose of our audit was to express an opinion on the
       financial and not to express an opinion on the
       effectiveness of internal control, discover weaknesses,
       detect fraud or other irregularities (other than those
       which would influence us in forming that opinion) and
       should not, therefore, be relied upon to show that no
       other weaknesses exist ..."
           I'm so sorry, my Lord, I'm actually reading the
       wrong text.
   MR JUSTICE FRASER:  "... to show that no other weaknesses
       exist or areas require attention."
           In other words we have identified some of them but
       don't reply on that as saying there aren't any others.
   MR DE GARR ROBINSON:  Exactly.  If your Lordship goes on to
       page 4 {F/1138/4}
   MR JUSTICE FRASER:  Page 4?  "Overview".  Is that the one?
   MR DE GARR ROBINSON:  Third paragraph down:
           "Focused management action in the past few years has
       addressed many of the issues raised in prior year
       management letters.  Whilst there continue to be
       challenges in areas including POL's IT environment,
       management have taken steps to ensure these challenges
       are and continue to be addressed."
           So, my Lord, what your Lordship gets from a reading
       of the documents as a whole that the audit processes
       which were regularly undertaken by Ernst & Young of
       Fujitsu and of its permission controls and operating
       controls, regulations, the picture actually to be
       derived from those documents is not that Fujitsu was not
       doing the right thing, the picture to be derived was
       that Fujitsu was being monitored for proper conduct and
       although there could on occasion be improvements, they
       were essentially acting in the right direction.
           My Lord, unless I can assist your Lordship further,
       that summarises my overview of remote access.
   MR JUSTICE FRASER:  That's very helpful.  I have read
       everything that I have been asked to -- I haven't read
       everything obviously in terms of the database, but
       I have read all of the documents that both of the
       parties asked me to read, and that's very, very useful,
       thank you very much.
   MR DE GARR ROBINSON:  My Lord, I was going to address
       your Lordship -- your Lordship may not want to be
       addressed on this -- on two points which I apprehend are
       included by way of prejudice.  One is on the process of
       disclosure that's been followed in this case and the
       other is the so-called shadow expert allegation.
       I don't know whether your Lordship would feel the need
       to hear any submissions about any of those things.
   MR JUSTICE FRASER:  Well, I would have thought the best
       place to do that is probably in closing because actually
       at the moment whatever the situation is suggested to be
       vis-√†-vis shadow experts, or so-called shadow experts,
       it will become a lot clearer after the
       cross-examination, I would have thought, but I'm not
       shutting you out from addressing me on it if you want
   MR DE GARR ROBINSON:  No, my anxiety is to help
       your Lordship not to set hares running.
   MR JUSTICE FRASER:  I would have thought closing.
   MR DE GARR ROBINSON:  Very well.  My Lord, all I would say
       about shadow experts is two things.  First of all, dark
       inferences are drawn as if having an advisor advising
       you before you have actually appointed an expert is
       something that is somehow untoward or inappropriate.
       My Lord, in cases of this scale it is commonplace in my
   MR JUSTICE FRASER:  Well, it depends on what they have been
       doing, which is why it is probably best left until after
       the cross-examination I would have thought.
   MR DE GARR ROBINSON:  My Lord, that brings me to my second
       point.  There appears to be if not a suggestion then an
       implication sought to be achieved that the technical
       advisor has in some way been used in order to shield the
       experts and the parties from adverse documents.
       My Lord, if that is the suggestion being made, it is
       wrong and it should not be made.  I can tell
       your Lordship that all the documents reviewed by the
       advisor were reviewed with a view to ascertaining
       whether any of them constituted adverse documents and
       any adverse documents that were revealed as a result of
       that review have been disclosed and it is important that
       your Lordship should not allow yourself to be -- it's so
       easy in cases of this kind to be drawn into the sort of
       painting of pictures.  My learned friend is actually
       creating something out of nothing and it is --
   MR JUSTICE FRASER:  I don't think you need concern yourself
       that I am easily drawn into painting pictures.
   MR DE GARR ROBINSON:  I'm very glad to hear it, my Lord.
           So, my Lord, assuming your Lordship doesn't wish to
       hear me on disclosure and the procedure that was adopted
       and your Lordship has submissions in the annex to our
       opening -- written opening submissions, all that remains
       is for me to talk about timings, the trial timetable.
   MR DE GARR ROBINSON:  Since the PTR I have been giving
       anxious thought as to the extent to which I can properly
       cross-examine Mr Coyne within the compass of two and
       a half days and were I required to do that then I would
       do that.
   MR JUSTICE FRASER:  Of course.
   MR DE GARR ROBINSON:  It is my view, having thought about it
       quite carefully, that it would be very, very difficult
       to get it under three days and three and a half days
       would be doable but under three would be very hard
       indeed, but if your Lordship were to say it should be
       two and a half days then it can be done in two and
       a half days.
   MR JUSTICE FRASER:  Mr De Garr Robinson, I will -- and just
       for the purpose of everyone in court because obviously
       counsel know this but not everybody will.  I separately
       raised this point of my own volition --
   MR JUSTICE FRASER:  -- about three weeks ago after I read
       the first joint statement which emerged from the meeting
       which I think was taking place the very day of the PTR
       or possibly a couple of working days later, so the trial
       timetable that was set down for the PTR was not chiseled
       into granite and I expressly invited this debate on
       Day 1.
           If you are going to have longer, and on current
       understanding of the different expert issues it seems to
       me that that's sensible, we just have to grasp the
       nettle now about how that's going to impact the second
       half of the trial.  I mentioned this morning
       difficulties in respect of another case on Friday the
       5th anyway.  It seems to me that the best way forward is
       to keep the shape of the first half the same, that the
       week commencing 1 April you should be given your four
       days to cross-examine Mr Coyne.  We then won't sit on
       the 5th and on Monday the 8th we will start with
       Dr Worden, which if Mr Green wants three days he can
       have them, if he wants four days he can have them but
       I know he said at the pre-trial review he would only
       want three but what's sauce for you is sauce for him,
       which means we are going to have to address closings.
           Now, I know from time to time I have come across as
       somewhat more robust about counsels' availability but
       obviously in this case that's slightly different now
       because we are starting the trial and this is an
       alteration and I know at the pre-trial review Mr Green
       had some observations about going into the week after
       that and you said this morning that the beginning of the
       next term might be a better time for them.
           So I'm open to suggestions and debate about that,
       but the primary decision is when do we do the evidence
       and it seems to me you want your four days and I'm
       prepared to give you your four days.
   MR DE GARR ROBINSON:  Well, my Lord, I certainly want three
       to three and a half days.
   MR JUSTICE FRASER:  All right, if you go three and a half
       days what we will do is do a hard start with Mr Coyne on
       the Monday morning, because apart from anything else
       people need to plan or organise their diaries and it's
       bad practice to start an expert just before the weekend
       and then he is in purdah and Mr Green can't speak to
       him, or vice versa.
   MR JUSTICE FRASER:  So that's my current thinking.
   MR DE GARR ROBINSON:  I'm grateful.  Shall I discuss it with
       Mr Green and then we can come back to your Lordship with
       hopefully an agreed approach.
   MR JUSTICE FRASER:  Mr Green, two things.  Longer for
       cross-examination, I have made that clear.
   MR GREEN:  My Lord, I'm anticipating within three days.
   MR JUSTICE FRASER:  All right, that's fine.  You will have
       to discuss it with Mr De Garr Robinson about when you
       might do your closings.
   MR GREEN:  Indeed.
   MR JUSTICE FRASER:  The beginning of the term afterwards at
       the moment I'm supposed to be on TCC business but that
       can and will be rearranged to fit around you.  I think
       term starts on 1 May which is a Wednesday, so you could
       have 1 and 2 May, or you could have two days the week
       after.  I think any longer than that it's beginning to
       get difficult in terms of the timetable for the whole
       year and actually the difficulties in the timetabling
       are probably more my difficulties than yours because I'm
       the one that has to write the judgment.
           So do you want to have a chat about that, about
       closings, but for the moment as far as cross-examination
       of the experts, Mr Coyne will be called on Monday,
       1 April and you, Mr De Garr Robinson, can have as much
       of the four days that week up to and including Thursday
       the 4th as you want and the same the other way round:
       Dr Worden will be called on Monday the 8th and you can
       have as much of the three into three and a half days
       that week to cross-examine him as you want.
   MR GREEN:  My Lord, I think the Thursday is Maundy Thursday,
       I don't know whether that --
   MR JUSTICE FRASER:  No, I think that's a week out, isn't it?
       I think Maundy Thursday is the following week because
       Easter is so late.
   MR GREEN:  Yes exactly.  It is so late.
   MR JUSTICE FRASER:  So that whole week we don't have any
       Easter issues at all that week.
   MR GREEN:  That's excellent.
   MR JUSTICE FRASER:  All right?  Is that helpful?
   MR DE GARR ROBINSON:  My Lord, it is extremely helpful,
       thank you.
   MR JUSTICE FRASER:  Couple of other admin points.  I have
       obviously been working on hard -- well, I have actually
       been working on the system but I've got three screens
       already and if I bring my laptop down I'm going to end
       up with four screens so I would like a file please from
       you of your witness statements that actually have the
       trial bundle pages printed on them.  I don't need the
       exhibits.  And the same for you, Mr De Garr Robinson,
       just a file of your evidence of fact at the moment with
       the trial bundle pages printed on them.
           Obviously I would like yours for tomorrow morning.
   MR GREEN:  My Lord, yes.
   MR JUSTICE FRASER:  Yours can come at some stage during the
       week and at some point I would like a hard copy file of
       your two experts' reports in the same file, no
       appendices, again with the trial bundle pages on them
       and the same for your expert as well.
   MR DE GARR ROBINSON:  Your Lordship may prefer to have the
       appendices as well because there is quite a lot of work
       done in the appendices of the --
   MR JUSTICE FRASER:  I've got the appendices -- actually if
       you think -- you are right, it is probably sensible to
       have those with the trial bundle references on.  The
       reason I said no appendices was simply to try and reduce
       the amount of work you have to do, that's all.  Is that
       quite clear?
   MR GREEN:  It is easy to do.
   MR JUSTICE FRASER:  And obviously the witnesses will need to
       have, as they did in the common issues trial, a bundle.
   MR GREEN:  Indeed.
   MR JUSTICE FRASER:  And I think there have been some recent
       statements, haven't there, just making corrections
       et cetera.
   MR GREEN:  My Lord, yes, there are some small changes.
   MR JUSTICE FRASER:  So I imagine examination-in-chief will
       be quite straightforward.
   MR GREEN:  Yes.
   MR JUSTICE FRASER:  Is that everything?  Thank you both very
       much and tomorrow morning.
   MR DE GARR ROBINSON:  My Lord, actually having nodded when
       your Lordship said "Is that everything" I should mention
       that Mr Henderson and Mr Draper will be doing some
       cross-examining --
   MR JUSTICE FRASER:  I had guessed that.
   MR DE GARR ROBINSON:  So in proper reflection of my real
       role in this case I will be allowing them to take pole
   MR JUSTICE FRASER:  Just so that you know, because I think I
       did say this at the beginning of the common issues trial
       but I will just say it to you, as far as counsel are
       concerned I don't mind where you sit, where they sit.
       They can sit up front -- there's actually a practice
       direction about this for the Rolls Building
       specifically.  So please don't think that if you would
       rather have Mr Henderson next to you or Mr Draper next
       to you all the time, that as junior counsel they have to
       sit behind you, because as far as we are concerned we
       don't do that.  But I anticipated they would be doing
       some of the cross-examination and I'm sure that might be
       the case for the other side as well.
           Is that everything?
   MR JUSTICE FRASER:  Thank you all very much.  Tomorrow
   (4.15 pm)
         (The court adjourned until 10.30 am on Tuesday,
                          12 March 2019)


   Housekeeping .........................................1
   Opening submissions by MR GREEN ......................8
   Opening submissions by MR DE GARR ...................79

    Housekeeping .......................................159

Choose an amount