PO Scandal

[Tedx]

Ian Hislop v some Tory MP

https://youtu.be/K3Y-YwTQ-jo?si=2WT7NWV9wr29TB4a

[Nimrod103]

Ian Hislop v some Tory MP

https://youtu.be/K3Y-YwTQ-jo?si=2WT7NWV9wr29TB4a

The MP is Sir Jake Berry.
The Govt is reluctantly enacting legislation to exonerate the subpostmasters because the courts are moving so slowly and are completely overwhelmed by the scale of the miscarriage of justice. However, this is controversial new territory because it cuts right across principle of the separation of legislature and justice systems. If Parliament can declare that somebody is innocent, they can equally declare that somebody else should go to prison, or somebody else should be let off punishment.

Hislop I am sure is equally outspoken in condemning any idea that the Govt should solve the illegal immigrant issue by withdrawing from the ECHR or declaring that Rwanda is a safe country. IMHO Ian Hislop has to be careful with the what he is arguing for. Consistency is not his strong suit.

[UncleEbenezer]

My experience and expectation is that financial systems will have a reconciliation process, and if funds aren't where they should be, then they are somewhere else that they shouldn't be. No matter how bad the Fujitsu system is/was, I'm sure that it was designed to reconcile.

But as I said above, Horizon was prone to creating phantom transactions. This would result in a real deficit, with the postmaster forced to pay in real money.

The books now reconcile, but the postmaster is out of pocket.

Scott.

Exactly.

Being fiat money and totally electronic, it can be created or destroyed. In a sufficiently buggy system, that can happen without any trail, or anyone knowing.

It seems to me a real likelihood that money was simply destroyed. The electronic equivalent of banknotes thrown onto the fire 'cos they were in an envelope thought to be rubbish!

[UncleEbenezer]

Ian Hislop v some Tory MP

https://youtu.be/K3Y-YwTQ-jo?si=2WT7NWV9wr29TB4a

The MP is Sir Jake Berry.
The Govt is reluctantly enacting legislation to exonerate the subpostmasters because the courts are moving so slowly and are completely overwhelmed by the scale of the miscarriage of justice. However, this is controversial new territory because it cuts right across principle of the separation of legislature and justice systems. If Parliament can declare that somebody is innocent, they can equally declare that somebody else should go to prison, or somebody else should be let off punishment.

Nothing new in that. The big precedent was set when the Magnitsky Act enabled a minister to punish someone without any pretence of Judicial process. Obviously intended primarily as a reminder for users of London's money-laundering services not to neglect their Party donations, but ripples far wider than that.

April 26th 2021. The very day the trial of Serco executives collapses for (as I understand it) technical reasons of the kind known informally as innocent-until-proven-broke, 24 people punished by a ministerial pen without any kind of process. Two standards, one heck of a contrast!

Hislop I am sure is equally outspoken in condemning any idea that the Govt should solve the illegal immigrant issue by withdrawing from the ECHR or declaring that Rwanda is a safe country. IMHO Ian Hislop has to be careful with the what he is arguing for. Consistency is not his strong suit.

Hislop is good at what he does: holding the powerful to account (though in the clip he seems to get rather carried away). Which doesn't mean he'd be good at wielding power.

[Alaric]

Everyone knows the price of a stamp, and there are effective barriers at the counters, so unless there was an armed hold up, I fail to see how anything could disappear.

In its simplest form, the allegation was that a stamp was sold for 30p, the coins were pocketed by the postmaster and thus the end of day cash reconciliation would show up 30p down which the postmaster had to return. But suppose the sale was a phantom invented by a flaw in the Horizon system?

[Nimrod103]

Everyone knows the price of a stamp, and there are effective barriers at the counters, so unless there was an armed hold up, I fail to see how anything could disappear.

In its simplest form, the allegation was that a stamp was sold for 30p, the coins were pocketed by the postmaster and thus the end of day cash reconciliation would show up 30p down which the postmaster had to return. But suppose the sale was a phantom invented by a flaw in the Horizon system?

I understand that and agree. What I was getting at is that it is so simple and blatant that, if they were indeed stealing, nobody could have imagined they would have got away with it. I start with 2000 stamps, but at the end of the week I only have the money for 1950 stamps. Nobody would have dared to think they could have persuaded the management that that was OK. So if there was a shortfall on the screen, there had to be a good chance it was generated by the software. Likewise for all the other stuff a PO sells.
So how come the PO management suspected that widespread fraud was taking place, and hence were pleased when Horzon appeared to confirm it. It doesn't really make much sense. all IMHO.

[Alaric]

So how come the PO management suspected that widespread fraud was taking place, and hence were pleased when Horzon appeared to confirm it. It doesn't really make much sense. all IMHO.

They wrote the Postmaster contract to make Postmasters responsible for shartfalls. Did they use Horizon on their in house main Post Offices? Increasingly Post Offices moved to branches of WH Smith. Did they ever accuse any of these?

One of the earlier Computer Weekly reports suggested that the underlying hardware and operating software was Windows NT. It wasn't DOS, Windows 3.11 or Windows 95 but even so weren't systems requiring high security based around Unix or Linux?

It might have seemed a long shot but wasn't there the possibility for the PO to be defrauded by hardware and software hacks finding a backdoor to the Branch machines? The PO Investigator being grilled by the enquiry recently claimed no computer knowledge.

[elkay]

Everyone knows the price of a stamp, and there are effective barriers at the counters, so unless there was an armed hold up, I fail to see how anything could disappear.

In its simplest form, the allegation was that a stamp was sold for 30p, the coins were pocketed by the postmaster and thus the end of day cash reconciliation would show up 30p down which the postmaster had to return. But suppose the sale was a phantom invented by a flaw in the Horizon system?

This doesn't hold, because reconciliation would include accounting for the physical stamps held. If a "phantom" stamp was sold, then there would be more physical stock then the system would have in its accounts.

The passport validation exmaple is a different propostion though.

[elkay]

My experience and expectation is that financial systems will have a reconciliation process, and if funds aren't where they should be, then they are somewhere else that they shouldn't be. No matter how bad the Fujitsu system is/was, I'm sure that it was designed to reconcile.

But as I said above, Horizon was prone to creating phantom transactions. This would result in a real deficit, with the postmaster forced to pay in real money.

The books now reconcile, but the postmaster is out of pocket.

Scott.

Without knowing how the system is designed, that may be a possibility...but would be facilitated by very poor design.

But it doesn't address the point that if these aren't occurring now, someone must have addressed the problem, and therefore it was known about.

[ReformedCharacter]

One of the earlier Computer Weekly reports suggested that the underlying hardware and operating software was Windows NT. It wasn't DOS, Windows 3.11 or Windows 95 but even so weren't systems requiring high security based around Unix or Linux?

I don't think Linux was in widespread use in that era and Unix probably would not have run on a PC then. Windows NT was a sensible choice and simple to interface with a modem which I assume it did.

RC

[UncleEbenezer]

One of the earlier Computer Weekly reports suggested that the underlying hardware and operating software was Windows NT. It wasn't DOS, Windows 3.11 or Windows 95 but even so weren't systems requiring high security based around Unix or Linux?

I don't think Linux was in widespread use in that era and Unix probably would not have run on a PC then. Windows NT was a sensible choice and simple to interface with a modem which I assume it did.

RC

Windows NT was never a sensible choice at a technical level. It was chosen by companies like ICL/Fujitsu because for them it was the obvious choice to sell to the Suits (not the techies) at their target client companies, and it ticked boxes.

A generation earlier (in computing terms), the equivalent was "noone ever got fired for buying IBM".

[swill453]

Windows NT was never a sensible choice at a technical level. It was chosen by companies like ICL/Fujitsu because for them it was the obvious choice to sell to the Suits (not the techies) at their target client companies, and it ticked boxes.

A generation earlier (in computing terms), the equivalent was "noone ever got fired for buying IBM".

Possibly just as well ICL weren't still trying to flog the OPD, based on the Sinclair QL.

Scott.

[UncleEbenezer]

Windows NT was never a sensible choice at a technical level. It was chosen by companies like ICL/Fujitsu because for them it was the obvious choice to sell to the Suits (not the techies) at their target client companies, and it ticked boxes.

A generation earlier (in computing terms), the equivalent was "noone ever got fired for buying IBM".

Possibly just as well ICL weren't still trying to flog the OPD, based on the Sinclair QL.

Scott.

At the risk of thread derailment, ICL itself was of course a me-too project of the UK government. Other countries - above all the US - had big computer companies, and Blighty had to have one. The private sector wasn't going to create one[1], so we got ICL. A more modern (though thankfully shorter-lived) equivalent would be OneWeb.

[1] Of course in real life the private sector created many. But it was a long process before ARM became, through its own merits (one of whom was our late, lamented Gengulphus) a genuine world-class company.

[ReformedCharacter]

I don't think Linux was in widespread use in that era and Unix probably would not have run on a PC then. Windows NT was a sensible choice and simple to interface with a modem which I assume it did.

RC

Windows NT was never a sensible choice at a technical level. It was chosen by companies like ICL/Fujitsu because for them it was the obvious choice to sell to the Suits (not the techies) at their target client companies, and it ticked boxes.

What would the techies have chosen c. 1999? At a reasonable price? AFAIK, none of the problems found were due to the OS but poor programming (by the 'techies') and other levels of dishonesty.

RC

[swill453]

What would the techies have chosen c. 1999? At a reasonable price? AFAIK, none of the problems found were due to the OS but poor programming (by the 'techies') and other levels of dishonesty.

To reiterate a post from earlier in the thread quoting a Computer Weekly article:

“A lot of the messages in there were nonsense, because there was no data dictionary, there was no API that enforced message integrity. The contents of the message were freehand, you could write whatever you wanted in the code, and everybody did it differently. And then, when you came back three weeks later, you could write it differently again.”

He gave an example of a message stored previously when a customer bought a stamp. It was feasible that a new message for buying a stamp weeks later could be slightly different.

“When the cash count came along, it found a message it was not expecting and either ignored it, tripped up, or added something it shouldn’t be adding,” he said.

Scott.

[88V8]

What would the techies have chosen c. 1999? At a reasonable price? AFAIK, none of the problems found were due to the OS but poor programming (by the 'techies')...

To reiterate a post from earlier in the thread quoting a Computer Weekly article:

“A lot of the messages in there were nonsense, because there was no data dictionary, there was no API that enforced message integrity. The contents of the message were freehand, you could write whatever you wanted in the code, and everybody did it differently. And then, when you came back three weeks later, you could write it differently again.”
He gave an example of a message stored previously when a customer bought a stamp. It was feasible that a new message for buying a stamp weeks later could be slightly different.
“When the cash count came along, it found a message it was not expecting and either ignored it, tripped up, or added something it shouldn’t be adding,” he said.

I have no difficulty relating to that. Years ago I was involved in departmental implementation of a new system, and speaking to the programmers about our particular requirements, I found that there were multiple ways of telling the system to do something, that not all the programmers recognised all the ways, and that the programme was poorly remmed if at all, so there were multiple possibilities that new programmers arriving later would fail to understand the programme.
The remming got worse as the job end neared and contract programmers scampered for the exit.

I imagine this is a pretty common situation, and in part at least accounts for the difficulties of adding new arms n legs to legacy systems.

V8

[mc2fool]

Windows NT was never a sensible choice at a technical level. It was chosen by companies like ICL/Fujitsu because for them it was the obvious choice to sell to the Suits (not the techies) at their target client companies, and it ticked boxes.

What would the techies have chosen c. 1999? At a reasonable price? AFAIK, none of the problems found were due to the OS but poor programming (by the 'techies') and other levels of dishonesty.

Beh. All software has bugs, and in one sense you can say all bugs are down to "poor" programming, or at least errors in design and/or implementation, but we don't know if that's particularly true in this case, or whether there were just a "normal" level of errors.

It's because some level or another of bugs is inevitable that there are (should be) multiple levels of testing. Any software house larger than a garage outfit will (should) have a quality assurance department - separate to the development team - who's job it is to develop test suites to knock the heck out of software to try and discover any bugs the developers themselves didn't catch. And then there's field (beta) testing where the software is given to first a limited set and then larger sets of real users to use for real, which inevitably finds bugs that neither the developers nor the QA dept found. And it is an absolute truism that even after such multiple levels of testing, after the software is declared ready and is released and used by it's target audience, more bugs will surface.

The question then is how those are handled, and that's where the real failing in this scandal lies. It does seem clear that the PO were prosecuting people even while knowing that there were bugs in Horizon that accounted for the apparent discrepancies. Truly atrocious.

[ReformedCharacter]

Beh. All software has bugs, and in one sense you can say all bugs are down to "poor" programming, or at least errors in design and/or implementation, but we don't know if that's particularly true in this case, or whether there were just a "normal" level of errors.

It's because some level or another of bugs is inevitable that there are (should be) multiple levels of testing. Any software house larger than a garage outfit will (should) have a quality assurance department - separate to the development team - who's job it is to develop test suites to knock the heck out of software to try and discover any bugs the developers themselves didn't catch. And then there's field (beta) testing where the software is given to first a limited set and then larger sets of real users to use for real, which inevitably finds bugs that neither the developers nor the QA dept found. And it is an absolute truism that even after such multiple levels of testing, after the software is declared ready and is released and used by it's target audience, more bugs will surface.

The question then is how those are handled, and that's where the real failing in this scandal lies. It does seem clear that the PO were prosecuting people even while knowing that there were bugs in Horizon that accounted for the apparent discrepancies. Truly atrocious.

I'd absolutely agree with that, but, as mentioned upthread:

“A lot of the messages in there were nonsense, because there was no data dictionary, there was no API that enforced message integrity. The contents of the message were freehand, you could write whatever you wanted in the code, and everybody did it differently. And then, when you came back three weeks later, you could write it differently again.”
He gave an example of a message stored previously when a customer bought a stamp. It was feasible that a new message for buying a stamp weeks later could be slightly different.
“When the cash count came along, it found a message it was not expecting and either ignored it, tripped up, or added something it shouldn’t be adding,” he said.

Does not seem like good design\programming.

RC

[servodude]

I don't think Linux was in widespread use in that era and Unix probably would not have run on a PC then. Windows NT was a sensible choice and simple to interface with a modem which I assume it did.

RC

Windows NT was never a sensible choice at a technical level. It was chosen by companies like ICL/Fujitsu because for them it was the obvious choice to sell to the Suits (not the techies) at their target client companies, and it ticked boxes.

A generation earlier (in computing terms), the equivalent was "noone ever got fired for buying IBM".

Hear! hear!
And there would be behind that a backline somewhere running SCO, HP Vue, or someone similar trying badly to integrate to something running Vax because... ??? IT?
This was the era where I'd have a part time job where I'd write something in Foxpro, have it ported to dBase and post it live (for the Scottish Office) the day before you f**"Ed off to Yorkshire to start a master's degree.
Fun.... but f'all accountability!

[mc2fool]

:
Does not seem like good design\programming.

RC

And that seems like quite an understatement.

That Computer Weekly article can be read in full on the Wayback machine.

"The Post Office’s Horizon IT system should “never have seen the light of day” and bosses at supplier Fujitsu allowed it to be rolled out into the Post Office network despite being told it was not fit for purpose, according to a senior developer who worked on the project before it went live.
:
“Everybody in the building by the time I got there knew it was a bag of s**t”, he said. “It had gone through the test labs God knows how many times, and the testers were raising bugs by the thousand.”
:
Central to his allegation is that Horizon’s Epos system was initially built with “no design documents, no test documents, no peer reviews, no code reviews, no coding standards”."

https://web.archive.org/web/20210219095756/https://www.computerweekly.com/news/252496560/Fujitsu-bosses-knew-about-Post-Office-Horizon-IT-flaws-says-insider

The page has at the bottom a (very) long list of articles since they first reported on it in 2009, and their very first, with case studies of some now familiar names is at: https://web.archive.org/web/20130127131005/http://www.computerweekly.com/news/2240089230/Bankruptcy-prosecution-and-disrupted-livelihoods-Postmasters-tell-their-story