Mr Bates vs the Post Office

[vacuum_tubes]

I'd be surprised if you could create a suitable test environment. Tens of thousands of sites and hundreds of thousands of miles of comms links with huge spikes in traffic at the end of a day is a significant contributor to a risk of intermittent failure.

It is possible and I have done it (numerous times)

Not on the same scale but I worked on a project years ago to migrate Boots from dial up/analogue to a Frame Relay network for their PDQ and tills etc.

We built a full test environment with ISDN2 and Kilostream circuits and used equipment that could emulate circuit noise, delays and corruption etc. It was all fully tested with failure scenarios etc.

 

[raysablade]

It is possible and I have done it (numerous times)

Not on the same scale but I worked on a project years ago to migrate Boots from dial up/analogue to a Frame Relay network for their PDQ and tills etc.

We built a full test environment with ISDN2 and Kilostream circuits and used equipment that could emulate circuit noise, delays and corruption etc. It was all fully tested with failure scenarios etc.

I well remember a HR system supplier complaining that their load testing had not anticipated thousands of managers on hundreds of sites logging on and using the system at the beginning and end of the day, when they were tired and just wanted to go home

 

[Seeker_UK]

The issue is in writing a comprehensive set of tests scripts for UAT to cover most scenarios and 'real world' scenarios, as it's then when the real faults come to the surface. However, even then things will get missed, and 'features' will come to light as V1.0 is deployed and gets used. That's when having a good fault reporting and resolution process comes to the fore.

 

[Sue Pertwee-Tyr]

And I’d expect the first assumption would be system fault / bug / user error, rather than fraud. You’d assume the user needed retraining on the system, before you even begun to think they might be a wrongun. I suspect the contractual duty to make up shortfalls made for a culture of ‘not my problem’ in Post Office admin.

 

[vacuum_tubes]

I well remember a HR system supplier complaining that their load testing had not anticipated thousands of managers on hundreds of sites logging on and using the system at the beginning and end of the day, when they were tired and just wanted to go home

But all of that is easy to test, when doing contact centres or hosted telephony we used load testing products like the call simulator from Hammer etc.

We had a licence that would simulate up to 100,000 concurrent voice calls hitting the media servers.

The issue is in writing a comprehensive set of tests scripts for UAT to cover most scenarios and 'real world' scenarios, as it's then when the real faults come to the surface.

Not really, the key is getting the requirements defined correctly

You then write the test scripts against those requirements, I know that Fujitsu use IBM DOORS internally and that is a good product.

 

[Seeker_UK]

Not really, the key is getting the requirements defined correctly

As someone who has spent a long time in requirements management and acceptance, I can tell you that is a rare thing. Something *will* be missed in the translation of user to system requirements. This is usually because stakeholders and 'end users' tend to get forgotten at that point; it's seen to be about the application development.

You then write the test scripts against those requirements, I know that Fujitsu use IBM DOORS internally and that is a good product.

IME, access to a tool like DOORS (or Siemens Polarion or Dimension RM etc) does not mean a complete or correct requirements set is guaranteed.

Even if requirements are complete and correct, system context can be lost and that can also lead to problems at the Demonstration phase. This is why a proper solutions architecture is essential, especially if a large and complex programme is being delivered by a number of suppliers through a prime integrator.

 

[vacuum_tubes]

The other issue is the whole govt procurement process.

I've written and presented enough public sector business cases to understand that presenting the more expensive option that better meets the requirements is often a difficult thing. With procurement it's often the preference to select the cheapest option for "best use of public money" and then hide the change request costs to make that solution actually meet the requirements.

We talk hypotheticals about how things "should be done" but the truth is that govt procurement is a bloody nightmare and often a blocker to doing things right.

I now work in safety and for example I'm doing an upgrade of a safety critical system at the moment. The cost of that project is around £22million but of that total cost, around £13 million is to upgrade the test system and do the testing/safety validation.

 

[vacuum_tubes]

As someone who has spent a long time in requirements management and acceptance, I can tell you that is a rare thing. Something *will* be missed in the translation of user to system requirements. This is usually because stakeholders and 'end users' tend to get forgotten at that point; it's seen to be about the application development.

Agreed but not fundamental things like testing how the solution reacts under failure scenarios.

From the TV show some of the errors were alluded to power cuts or connectivity issues, that is a very simple thing to define as a requirement and test for.

 

[paulfromcamden]

Regardless of initial testing they knew full well the system was shagged twenty years ago and continued with the prosecutions regardless.

 

[vacuum_tubes]

This is why a proper solutions architecture is essential, especially if a large and complex programme is being delivered by a number of suppliers through a prime integrator.

I'm technical and have done both design and solution architecture but business requirements still sends a shiver down my spine. I've met some very good business analysts but it's a highly skilled and very difficult job, I take my hat off to them.

Regardless of initial testing they knew full well the system was shagged twenty years ago and continued with the prosecutions regardless.

I'm thinking from a legal perspective that if they can prove the govt/PO/Fujitsu knowingly put a not fit for purpose system live there is no need to argue or prove they continued with the baseless prosecutions.

 

[Sue Pertwee-Tyr]

No, but it should be a massive aggravating factor for damages.

 

[Jim Audiomisc]

prosecutions of postmasters 1999 - 2015

1999 - 2007 : Blair
2007 - 2010 : Brown
2010 - 2015 : Cameron/Clegg

i don’t think it is fair not to include Labour and Liberals in this appalling charade - NOT a party political issue in my view, all about covering arses when things go wrong (something all parties are pretty adept at)

Yes and No. PE has reported on the Blair/Brown period as well as the later ones. Blair/Brown AIUI basically just told the PO + Fujisu to get the system working properly having been told about the 'development' problems. So far as I've been able to tell, they were led to assume problems had been fixed... when they hadn't. The reality only emerged from behind the PO/Fujitsu wall of darkness later on, and everyone outside those 'in the know' at PO/F weren't being told. i.e. *Government* was kept in the dark as Postmasters were crucified on the basis of lies.

But given the epic scale of this scandle, that could be wrong.

 

[Kirk]

Scandal is often bandied about, but this, without question, is one. Let's hope the victims receive the recognition and compensation they deserve. That said, it's another example of how quickly the State, left unchecked, can act in an authoritarian and exploitative manner.

 

[ff1d1l]

Just sitting in the Wairose cafe, reading todays times.
Although sporting the front page headline "Post Office fury intensifies" pages 1, 2 and double page spread on 7 and 8 seem to be prioritising sticking a not entirely undeserved knife into the Lib Dems, who supplied ministers including nominative determinism proponent Ed Balls in various positions of oversight who did little or nothing.

Still, nice to know the story is not so much about suffering and injustice, but about damaging tory opponents in an election year.

 

[vacuum_tubes]

Yes and No. PE has reported on the Blair/Brown period as well as the later ones. Blair/Brown AIUI basically just told the PO + Fujisu to get the system working properly having been told about the 'development' problems.

Similar to what I heard 'off the record' but it was more along the lines of 'this is a flagship govt project, we won't accept any delays, it is going live when we said it would'

 

[Jim Audiomisc]

Everyone knows this sort of shit never works first time, the job is to make it work. Behaving like a Boris Johnson, Matt Hancock or Grant Shapps in this environment just doesn’t work as one is dealing with logic and data. It just needed to be debugged and fixed. No easy way out by lying.

Yes. The key here is that we all should know that software *will* contain bugs, 'features', 'exploit paths', etc. Thus the correct reaction to Postmasters should have been acceptance and bug-squashing fixes done, along with corrections. Where needed via independent ye olde accounts methods on paper, etc.

The problem was that F+PO simply hid and denied the reality to *everyone* around them. And continued to maintain this by false prosecutions and continuing denials, concealment, etc.

 

[Jim Audiomisc]

I well remember a HR system supplier complaining that their load testing had not anticipated thousands of managers on hundreds of sites logging on and using the system at the beginning and end of the day, when they were tired and just wanted to go home

Must confess that I've tended to assume that it makes more sense for the central collecting to control pooling round the outstations overnight. i.e. let the central collection be done via the enter choosing which outstaions to read at a time, and collect the data spread over the night whilst the shops are closed. Means the system is synched and updated overnight with the transfer and computing spread out as postmasters sleep. But I've never written any software for a task like this, so am probably clueless.

 

[Tony L]

The problem was that F+PO simply hid and denied the reality to *everyone* around them. And continued to maintain this by false prosecutions and continuing denials, concealment, etc.

Yes, exactly, and that was clearly a series of very conscious management decisions to ignore bug reports, tamper with live data, blame end users, and lie to them via the call centre helpdesk. Those responsible need to be personally identified and prosecuted with everything that can be thrown at them right up to and including corporate manslaughter IMO. If this occurred within Fujitsu then it needs suing into oblivion.

 

[Enfield boy]

Pity govt stole up to £37bn to give to Dildo Harding for a MS Office based spreadsheet rather than sort this mess and build a few hospitals, pay doctors etc. It's almost like they don't gaf.

 

[Seeker_UK]

The other issue is the whole govt procurement process.

I've written and presented enough public sector business cases to understand that presenting the more expensive option that better meets the requirements is often a difficult thing. With procurement it's often the preference to select the cheapest option for "best use of public money" and then hide the change request costs to make that solution actually meet the requirements.

I agree with most of that except the issue is not with the process but with those in positions of authority to select the right option. IME, that's as you state, about being seen to be careful with public finance, but also not having any real appreciation (or care) for the risk the cheapest option often carries.