When the Second Sight interim review into the state of the Post Office's Horizon IT system was published in July 2013, I asked him straight up if it was possible that Horizon IT errors could be responsible for discrepancies in Subpostmaster accounts. His simple "Yes, of course" by way of response was unequivocal.
We stayed in touch. I read some of his work, including his clear view of where the court was wrong in the prosecution of Seema Misra, the West Byfleet Postmaster sent to prison whilst pregnant.
Mr Mason's central argument is that a recommendation given by the Law Commission in 1997 to the courts is helplesssly, hopelessly and dangerously wrong. The guidance states:
"in the absence of evidence to the contrary, the courts will presume that mechanical instruments were in order at the material time... The principle has been applied to such devices as speedometers and traffic lights and in the consultation paper we saw no reason why it should not apply to computers." [my italics]
You don't need to be a legal eagle or IT genius to see that the Law Commission's decision to leap from a speedometer to the vastly complex digital ecosystems underpinning modern computer networks is mutton-headed, to say the least.
Yet it is this presumption the courts have relied on throughout the 21st century in the prosecutions of Subpostmasters and unfortunate individuals in other spheres, who were, it turns out, entirely innocent.
Mr Mason has been attempting to get the legal position on computers changed for as long as I can remember. Now he has a document which might force the issue - the Horizon Issues judgment, handed down on 16 Dec 2019.
On the anniversary of the judgment, Mr Mason, along with the barrister Paul Marshall and eight other learned authors, published a paper positing alternatives to the way digital evidence is currently handled in the courts.
It is called: "Recommendations for the probity of computer evidence" and it states:
"All computers have a propensity to fail, possibly seriously. That is to say, they have a latent propensity to function incorrectly."
As a general rule of thumb:
"A program on a mobile telephone might hitherto have contained tens of thousands of lines of software code. A program such as Horizon will contain tens of millions of lines of code, and will be exceedingly complex. Programming is a human task and programmers make mistakes; an error rate in writing software code of 10 errors per thousand lines of code is considered good, 1 error per thousand lines is rarely if ever achieved."
By way of example the paper cites a 2006 University of York review of a study conducted by the Ministry of Defence into the safety of the software contained in its Hercules C130J air transporters. The MoD found that the Hercules software contained about 1.4 safety-critical faults per thousand lines of code (kLoC) with an overall flaw density of around 23 per kLoC
The review notes:
"whilst a fault density of 1 per kLoC may seem high it is worth noting that commercial software is around 30 faults per kLoC, with initial fault injection rates of over 100 per kLoC."
Mr Mason et al's clear recommendations for a protocol on the way electronic evidence is treated by the legal system can be read here. They seem eminently sensible.
If you are able to give £20 or more you will be added to the secret email list. This alerts you to upcoming developments on this scandal before they are made public, as well as links to articles and stories posted here on this blog or elsewhere.