Seen a lot of tweets from people saying Post Office investigator Stephen Bradshaw is coming across as incompetent and/or blindly defending the PO.
For those of us who have covered the inquiry before, let me tell you this is nothing new.
#PostOfficeScandal
Breaking: Former Post Office boss Paula Vennells has handed back her CBE.
Comes after Sunak called for her to be stripped of the honour and a petition reached 1.2 million signatures.
How is it determined whether a witness is warned about self incrimination? I don't understand why Mr Scott is any more or less at risk than other witnesses #PostOfficeInquiry#PostOfficeScandal
@aviationcomment Also, whatever integrity issues it was *designed* to cope with, even where the outcome is deterministic, it doesn’t follow that it actually did cope with those. Fujitsu’s evidence always says what it should do, not what it did do.
@aviationcomment I cannot argue against specific recovery scenarios because I don’t know how it was coded. But my point is that if you have two message stores and both can master transactions, you cannot know for sure which one is correct and which is is incorrect from just two sources.
@MaaleeshHabibi@CastletonLee@PostOffice@fujitsu_uk You have to know which message store is correct, and you cannot know that if you don’t know what transactions the SPM entered or expected to be entered. Where replication can happen in multiple directions, you may find that no individual message store is correct and complete.
@sjmurdoch @RichardMoorhead @FloraClairePage @teem50@stugoo17@brianwhelton@Jusmasel2015 So whichever counter you, the system, or a Fujitsu employee chooses as the master, they will lose transactions. And unless you *know* that a transaction you entered has been lost, you won’t re-enter it. You have to reconcile with the real world.
@sjmurdoch @RichardMoorhead @FloraClairePage @teem50@stugoo17@brianwhelton@Jusmasel2015 SPM is transacting on counter 1, and this activity is being replicated to counter 2. At some point there is a failure and counter 2 does not receive some of the transactions or they are corrupted in some way. SPM doesn’t know this and enters a transactions on counter 2.
@sjmurdoch @RichardMoorhead @FloraClairePage @teem50@stugoo17@brianwhelton@Jusmasel2015 The replication from counter 2 to counter 1 is also broken, so counter 1 doesn’t receive this activity. The postmaster notices that everything is out of sync and stops trading. Now you have transactions in counter 2 that don’t exist in 1 and vice versa.
@FloraClairePage @teem50@stugoo17 @RichardMoorhead @brianwhelton@Jusmasel2015 Some of this will depend on the details of Horizon (specifically Riposte), and as far as I know, this has never been published. However, the general problem it tried to solve is notorious as one of the most difficult in computer science. It was barely even understood at the time.
I have to respect Mr Jones for apologising, one of the first I have heard. I am not sure how Mr Castleton will feel about that. #PostOfficeInquiry#PostOfficeScandal