Largest ever miscarriage of justice?

I do think that Jenkins is probably right about one aspect in his testimony... and that is, the faults themselves in Horizon were never directly the cause of the prosecutions, and that it was the Post Office's behaviour (their failure to carry out full and proper investigations, their complicity hiding how widespread the problems were, by lying to the SPM's when they told each one that they were the only ones with problems). I can have sympathy for his position that faulty computers don't prosecute people.

However, what he seems to leave out is that Fujitsu and Post Office clearly conspired to keep the whole thing from coming to light... and his invocation of "The Nuremberg Defence" is just laughable.
 
Last edited:
Filippo Lippi said:
Auditing must have been ruled "out of scope," probably for cost reasons, at some point, there can be no other excuse for it?
Click to expand...
Any such imposition should have been fought, quite possibly on legal grounds. I would have told a client who attempted to impose such a block to go screw themselves, and have the data, paper and recording trail to show this when it erupted.
 
Filippo Lippi said:
Auditing must have been ruled "out of scope," probably for cost reasons, at some point, there can be no other excuse for it?
Click to expand...

I would imagine there was some sort of record keeping system logging what was happening on Horizon, if only to allow the honchos to be able to dob others into it. Even if it wasn't set to be accessible by anybody it was probably still there.
 
Over the course of three decades of poking around in systems I have encountered many systems without audit trail capabilities and even more where the system logs are utterly unusable. I'm currently in major conflict with most of my current employers' IT department because I want to know who has been doing what in one of our in-house systems and they're dreaming up every excuse imaginable why none of the data I want exists in any meaningful sense. I'm still baffled that implementations can get signed off without this kind of thing enabled, tested and verified.

ETA: I should add that pleading ignorance of remote access functionality is total bollocks. Trump-level lying to inquiries.
 
Last edited:
There was at least one room and many staff allocated to the shenanigans, not just a few screens used on the fly, it must have cost millions over the years.
 
Everybody thinks enterprise software is a fully integrated and smooth-running machine, with all the functions and features under the hood that anyone could reasonably ask for.

The truth is it's all spittle and baling wire and most of the stuff you'd expect is either not there or doesn't work the way you need it to work.
 
Fujitsu were fixing bugs in old code then adding it back to the system after the old code had been rewritten, thus creating more bugs.
 
That's common in projects that are way out of control, as this one clearly was.

For those not in IT development -

Let's imagine a problem is reported in the Pensions subsystem.

Support Team 3 identify a bug in PENSION4, a core part of that subsystem. They code and fix the bug and apply it to the live system, problem goes away.

However, Development Teams 1 and 4 are both working on enhancements to PENSION4, and unbeknown to anyone Support Team 2 are working on another bug that was reported. All three teams' version of PENSION4 contain the bug that Support Team 3 fixed. If any one of them release their version without applying (and testing, fingers crossed) Support Team 3's fix, then the fixed code will be lost and the bug will re-emerge in the live system.

Change management should have dealt with this. It is shocking, but not surprising that ICL/Fujitsu failed to have sufficient controls
 
Filippo Lippi said:
Jenkins resigns from British Computer Society over potential investigation
Click to expand...

“He is currently under investigation by the Metropolitan Police for potentially committing perjury during the prosecutions of subpostmasters. The police launched its investigation into Jenkins and former Fujitsu colleague Anne Chambers in November 2020, after High Court judge Peter Fraser referred them to the Director of Public Prosecutions.”

The police sure are taking their time about it.
 
Filippo Lippi said:
That's common in projects that are way out of control, as this one clearly was.

For those not in IT development -

Let's imagine a problem is reported in the Pensions subsystem.

Support Team 3 identify a bug in PENSION4, a core part of that subsystem. They code and fix the bug and apply it to the live system, problem goes away.

However, Development Teams 1 and 4 are both working on enhancements to PENSION4, and unbeknown to anyone Support Team 2 are working on another bug that was reported. All three teams' version of PENSION4 contain the bug that Support Team 3 fixed. If any one of them release their version without applying (and testing, fingers crossed) Support Team 3's fix, then the fixed code will be lost and the bug will re-emerge in the live system.

Change management should have dealt with this. It is shocking, but not surprising that ICL/Fujitsu failed to have sufficient controls
Click to expand...

Its like having more that one person editing a document they have downloaded to their computers from a network. If person 1 downloads the document to edit, and while he is doing that, person 2 downloads the same original document, whichever of them uploads their final edited document last, that document is not going to contain any of the edits made by the person who uploaded their edited document first.

There are supposed to be protocols to prevent that from happening, but clearly, Fujitsu didn't have anything like that in place.
 
smartcooky said:
Its like having more that one person editing a document they have downloaded to their computers from a network. If person 1 downloads the document to edit, and while he is doing that, person 2 downloads the same original document, whichever of them uploads their final edited document last, that document is not going to contain any of the edits made by the person who uploaded their edited document first.

There are supposed to be protocols to prevent that from happening, but clearly, Fujitsu didn't have anything like that in place.
Click to expand...

Old school
 
malbui said:
Over the course of three decades of poking around in systems I have encountered many systems without audit trail capabilities and even more where the system logs are utterly unusable. I'm currently in major conflict with most of my current employers' IT department because I want to know who has been doing what in one of our in-house systems and they're dreaming up every excuse imaginable why none of the data I want exists in any meaningful sense. I'm still baffled that implementations can get signed off without this kind of thing enabled, tested and verified.
ETA: I should add that pleading ignorance of remote access functionality is total bollocks. Trump-level lying to inquiries.
Click to expand...
Because they're signed-off by technically illiterate managers?
 
zooterkin said:
He did ask for immunity, at least twice, but it seems it wasn't granted, as far as I can find out.

https://www.gbnews.com/news/post-office-scandal-gareth-jenkins-immunity
Click to expand...
Ah yes, I have found the application in the inquiry documentation. It was to be made to the AG [an 'undertaking'] by the Inquiry as the latter cannot grant such immunity.

By letter dated 15 September 2022, Mr Jenkins’ legal representatives, Corker Binning, confirmed what I understand to be a matter of public knowledge, namely, that Mr Jenkins is subject to an ongoing criminal investigation related to the subject matter of the Inquiry. The letter went on to say that as a consequence, it would be open to Mr Jenkins to invoke his privilege against self-incrimination (“the Privilege”) and refuse to answer many of the questions within the Request.

On 3 October 2022 Corker Binning wrote to the Inquiry confirming that Mr Jenkins would seek to exercise the Privilege in relation to questions 16 to 49 (inclusive) of the Request. However, they also invited me to seek an undertaking from the Attorney General that would limit the use to which any written or oral evidence which Mr Jenkins gives to the Inquiry would be put. No specific terms were proposed, although it was suggested that it was “standard” for such undertakings to provide that any evidence given could not be used as evidence against the individual in question in criminal proceedings (or for the purpose of determining whether to prosecute him/her).

However, there is a strong feeling amongst the SPMs that the ongoing criminal investigation and any potential future criminal proceedings against Mr Jenkins should not be prejudiced or impeded in any way. There is also a substantial body of opinion amongst SPMs that the grant of immunity, whatever its precise terms, would have the effect of prejudicing or impeding criminal proceedings against Mr Jenkins notwithstanding that an Attorney General’s undertaking is not a grant of immunity against criminal proceedings. All the SPMs represented by Hudgell Solicitors oppose my making the request to the Attorney General on that basis, as do all those represented by HJA and a number of those represented by Howe + Co.

In all the circumstances currently known to me, I am not persuaded that it is necessary, to discharge the Inquiry’s Terms of Reference, that I should request an undertaking from the Attorney General. Rather, I am of the view that a ‘wait and see’ approach is the more appropriate course.
Click to expand...
[This has been somewhat edited by me for brevity]
 
Last edited:
A question.

Firstly, there are about 23,000 sub-postmasters in the UK. About 900 were prosecuted... that is less than 4%. I'm not sure how many more were having issues with money going missing, but it seems Horizon worked fine for about 95% of the SPMs

Secondly, fibre Optic cables didn't start being laid and used for internet in the UK until abput 2008, so all of these SPM systems must have been connected with copper landlines and dial-up or VDSL. If I recall my VDSL protocols correctly, you had to be within a certain distance of the cabinet/exchange (a half mile?) for VDSL to work properly.

So my question is, what was it that was peculiar about those who were having trouble? Did they have something in common, like, were they rural? Or using dial-up, or VDSL a long way from their nearest cabinet?
 
smartcooky said:
A question.

Firstly, there are about 23,000 sub-postmasters in the UK. About 900 were prosecuted... that is less than 4%. I'm not sure how many more were having issues with money going missing, but it seems Horizon worked fine for about 95% of the SPMs

Secondly, fibre Optic cables didn't start being laid and used for internet in the UK until abput 2008, so all of these SPM systems must have been connected with copper landlines and dial-up or VDSL. If I recall my VDSL protocols correctly, you had to be within a certain distance of the cabinet/exchange (a half mile?) for VDSL to work properly.

So my question is, what was it that was peculiar about those who were having trouble? Did they have something in common, like, were they rural? Or using dial-up, or VDSL a long way from their nearest cabinet?
Click to expand...

I vaguely remember there being something about having more than one terminal at the post office being a factor, though it may have been more than that.
 
Unstable internet connections were a big factor, and power cuts, even an unstable mains supply. Also the terminal being moved. One SPM had no trouble at all until the PO was moved into a portacabin for the main shop to be renovated, then huge problems. She was removed from her job and the trouble continued for the next incumbent.

When the terminal was moved back the engineer did a lot of detailed checks and re-setting, saying that these things were notorious for going wrong if they were moved. No such checks had been done at the time of the original move.

It's all in Nick Wallis's book.
 
smartcooky said:
A question.

Firstly, there are about 23,000 sub-postmasters in the UK. About 900 were prosecuted... that is less than 4%. I'm not sure how many more were having issues with money going missing, but it seems Horizon worked fine for about 95% of the SPMs

Secondly, fibre Optic cables didn't start being laid and used for internet in the UK until abput 2008, so all of these SPM systems must have been connected with copper landlines and dial-up or VDSL. If I recall my VDSL protocols correctly, you had to be within a certain distance of the cabinet/exchange (a half mile?) for VDSL to work properly.

So my question is, what was it that was peculiar about those who were having trouble? Did they have something in common, like, were they rural? Or using dial-up, or VDSL a long way from their nearest cabinet?
Click to expand...

There were multiple things that went wrong for SPMs, including bugs, software, hardware, and connectivity.

Plus there’s the whole illegalities of the SPM contracts, and the way the PO bullied them into “admitting” their “thefts” and withheld evidence from them.
 
This is the dumpster fire I can't stop watching burn. I've been watching a lot of the inquiry on YouTube and just started listening the "The Great Post Office Scandal" audio book. It's such a spectacular failure at every possible level.

Maybe it's just a difference between US and UK civil procedure but can't a UK solicitor compel a "yes" or "no" question? In the US, on cross, an attorney can force a "yes" or "no". It seems like you could bit into a lot of the equivocation by the POL and Fujitsu employees if you denied the opportunity to explain in some of the questions.

I'd have liked to have seen the inquiry solicitors phrase questions about their clients as accusations. There were a few where the questions were about "what happened to Mr/Ms whatever". The better question would have been about what, "you did to "Mr/Ms. whatever".
 
Rolfe said:
Unstable internet connections were a big factor, and power cuts, even an unstable mains supply. Also the terminal being moved. One SPM had no trouble at all until the PO was moved into a portacabin for the main shop to be renovated, then huge problems. She was removed from her job and the trouble continued for the next incumbent.

When the terminal was moved back the engineer did a lot of detailed checks and re-setting, saying that these things were notorious for going wrong if they were moved. No such checks had been done at the time of the original move.

It's all in Nick Wallis's book.
Click to expand...

Orphia Nay said:
There were multiple things that went wrong for SPMs, including bugs, software, hardware, and connectivity.
Plus there’s the whole illegalities of the SPM contracts, and the way the PO bullied them into “admitting” their “thefts” and withheld evidence from them.
Click to expand...


That sounds like being in a remote-ish rural area would increase your chances of having problems!
 
smartcooky said:
That sounds like being in a remote-ish rural area would increase your chances of having problems!
Click to expand...

Maybe. The book does talk about a couple of SPMs who lived in remote areas, I recall.

It also talks about a SPM who had electrical problems that affected the data transfer.

Maybe worth mentioning the footnote on page 102 as an aside:

“The arrival of Horizon Online in 2010 was being presented as a big shift. Data would be updated at Horizon’s back end in real time rather than being stored locally and updated overnight. Its implementation was a disaster.”

But yeah, I’m not sure an analysis has been done as to whether remoteness has made errors significantly more likely. I’d say it’s possible.
 
smartcooky said:
That sounds like being in a remote-ish rural area would increase your chances of having problems!
Click to expand...


Yes, it was widely recognised that remoter rural POs were very disproportionately represented among the branches with problems.
Nevertheless main POs in urban locations also experienced shortfalls.
 
Last edited:
Craig4 said:
This is the dumpster fire I can't stop watching burn. I've been watching a lot of the inquiry on YouTube and just started listening the "The Great Post Office Scandal" audio book. It's such a spectacular failure at every possible level.

Maybe it's just a difference between US and UK civil procedure but can't a UK solicitor compel a "yes" or "no" question? In the US, on cross, an attorney can force a "yes" or "no". It seems like you could bit into a lot of the equivocation by the POL and Fujitsu employees if you denied the opportunity to explain in some of the questions.

I'd have liked to have seen the inquiry solicitors phrase questions about their clients as accusations. There were a few where the questions were about "what happened to Mr/Ms whatever". The better question would have been about what, "you did to "Mr/Ms. whatever".
Click to expand...
It would be up to the judge to instruct a witness.

Giving evidence some years ago I had a QC, as was, attempt to force me to give "a simple yes or no answer" to her questions, to assist her client's case.
"Your Honour I do not believe that question can be adequately answered yes or no and I request to court's indulgence to answer it properly." Seven times.
She was not happy. Though her client may have been less happy.
 
Wudang said:
Updates to remote systems had multiple solutions even in those days over dial up systems. Accepting duplicate entries in a financial system in that day was not a dark art.
Click to expand...
True. I remember working on a database system that used 32kb satellite links to ships in the Atlantic. It was more reliable than Horizon.
 
catsmate said:
True. I remember working on a database system that used 32kb satellite links to ships in the Atlantic. It was more reliable than Horizon.
Click to expand...

In the 1980s and 1990s I was part of a team that designed and implemented a data processing system for country grain elevators across western Canada. The computers were an oddball IBM system that maxed out at 256K RAM and 64 MB hard drives. Communications to the mainframe at head office was initially done overnight using dial-up modems at 2400 bps.

Over time we increased the number of communications to five per day and transferred the preceding few hours worth of transactions each time. We also transferred the software from the original hardware to emulation on Intel hardware running a version of Unix. But we never got rid of the dial-up processing because it was baked into the system at each end.

A big difference was the system was designed to handle only 120 locations instead of the thousands of post offices the United Kingdom has. And I suspect the operation of a western Canadian country grain elevator is simpler than a British post office. However, our implementation was robust and had checking in it to ensure no individual transaction was ever lost.

In the early 2000s the system was phased out and replaced with a much more modern one. I don't know how well that went because I was laid off by the company even before the last of the system I worked on was retired.

I looked at the design of Horizon as documented in one of the court cases. It appears to have been robust. But the devil is in the details and by all accounts the good high-level design was undone by poor programming of the individual components, and the apparent lack of an audit trail that could be examined when unexpected things happened.
 
Memory popped up. I was on a IBM CICS application programming course in 1985 and the instructor started with "There are 3 important principles in programming. Data integrity, data integrity, and data integrity."
 
Wudang said:
Memory popped up. I was on a IBM CICS application programming course in 1985 and the instructor started with "There are 3 important principles in programming. Data integrity, data integrity, and data integrity."
Click to expand...
:thumbsup::thumbsup::thumbsup:
 
You must log in or register to reply here.

Back
Top Bottom