Thread by @nickwallis

We are at the High Court’s Rolls Building for Day 6 of the Horizon trial, part of the Bates v Post Office group litigation. @PostOffice director Angela van den Bogerd continuing to give evidence. In the last trial she was found to have misled the court.
#postofficetrial

@PostOffice Follow this thread for live tweets from court. Please note EVERYTHING I write is a paraphrase or summary of what happens in court. It is not a direct quote unless it is in “direct quotes”.

Short preamble before AB’s evidence continues re redaction of documents. @PostOffice are trying to withhold from the judge on grounds of confidentiality privilege and relevance. Judge implying he will decide what is relevant and if PO holds out on that factor he will take a view.

@PostOffice PO QC suggests it is an unusual approach for his Lordship to take.

“But not improper” intones the judge.

Patrick Green QC (PG) for the claimants has started xe-ing AB.

He’s taken AB to a document which shows a key driver of putting the Branch Support Programme is cost-saving.
AB partly
PG the measures you are adopting in the BSP will also reduced the number of SPM suspensions - yes - reduction of errors
AB yes

We are noting the lack of information gathering by the post office pre-2015 on Horizon helpline data - ie volume of calls. And discussing the data that has been gathered since - broken down into categories according to how it was progressed.

J asks this is relevant
PG says it is of small relevance and says he has pretty much finished. Asks about helpline budget increases - were there any?
AB we restructured to put more people onto Tier 2 support to help
PG so previously Tier 2 couldn’t see Credence data
AB correct.

[we move on to par 30 of AB’s witness statement (WS)]
PG you explain that for your evidence you have relied on former trainers and auditors who have honed their skills on Horizon investigations - a pretty impressive group of people, it sounds like?
AB yes

PG when there were documents referred to by your team - did you look at the documents they referred to - the ones in your WS
AB yes I did
PG and this bit is about ARQ data and helpline logs
AB yes
PG helpline logs are not available to SPMs are they?
AB no

PG in fact Angela Burke had to make a subject access request and pay a fee and provide ID to get the recording of her own call. Is that acceptable?
AB no. no. we should provide it if there is an issue.

PG let’s go to Mr Latif’s case. You say in your WS that Mr Latif has got things wrong. [lists how AB says how in her WS]. Do you understand the difference between transaction data and event data and session data.
AB tries to explain - PG pushes her. She accepts she doesn’t...

… have a high level understanding of it.
PG in your WS you refer to event data, when it is transaction data that you refer to in your WS - the three specific docs you call transaction data are event data docs
AB I understood them to be transaction data as well

PG well let’s have a look then.
PG [shows her the event.] It’s event data. It’s transaction data and we can see here that not every transfer out has a transfer in.
AB accepts
PG so you can’t actually see that
AB not from this one, no
PG all 3 documents...

… are references to event data. Where you aware this was taken up with your solicitors in February?
AB no. I didn’t put the numbers on the documents
PG But it doesn’t support what you said. I asked you at the beginning if you read the docs to see if supported what you...

… were saying in your WS.
AB I am mistaken.

PG [now talking about correspondence between the solicitors to try to get to the bottom of this.] We specifically asked about this. Were you aware of it?
AB no
PG it was actually disclosed on 21 Feb and Mr Latif went to Kashmire on 19 Feb.

PG [lists them again] But you didn’t know anything about that.
AB no I’m sorry
PG to get to the bottom of Mr Latif’s issue we need to see the transaction data, don’t we?
AB yes

PG when you made your statement were you aware there were 2 major bugs causing problems with the transfer between stock units.
AB no
PG mentions the Dalmellington and Callender Square bug
AB I knew about that.
PG did you have them in your mind when you looked at Mr Latif’s...

… situation?
AB not specifically. we looked at his data alone.
PG Let’s have a look at his transaction data, shall we
PG when you investigated did you filter this info [there are 28K rows of data] by date?
AB by date, yes
PG on paper or electronically
AB electronically

PG [takes her to row 3478]. Have a look at column L… sale value… so that’s on the 4 July - £2K with minus £2K underneath it. Were you looking at that when you looked at Mr Latif’s data
AB what? that specific row
PG you were looking by date
AB we looked at all of July

PG the short point is… you can’t tell what that is, can you?
AB we looked at June July August - all the transaction data, filtered by stock unit and data…
PG do you know what the code is - 4190 in there
AB no - I’d need the list. My team would know.

PG takes her to another disclosed document. Those are the product ID lists?
AB yes
PG and one of the things we know is that product ID numbers is that they are different length. Some are 5 digits, some are 2 some are 4…
AB agrees

PG and some product codes are provided by 3rd parties?
AB yeah I don’t know where they come from
PG let’s go back to your WS par 91.2 - that was your analysis in your WS when you did it in November and you say there there were 2 x transfers from 1 stock unit to another...

… both were successful. [PG reads out another]
PG and then you’ve had to amend your account of what transactions took place, haven’t you?
AB yes
PG so it should now be 2 x transfers of £2K “between" not "from" two stock units. PG says the list of 4 transfers noted in 91.3

… was actually 5. So your evidence is not correct.
AB it wasn’t as good as I liked
PG but it’s wrong - it’s not a question of full or not full - it’s wrong
AB I expanded 91.2 and changed 91.3
PG no - 91.2 was wrong

PG you said it was 2 x transfers between. It wasn’t. It was 2 x transfers between. It was wrong.
AB yes
PG Dr Worden suggests Transaction Corrections are part of Horizons robustness. This trial is not part of Horizon’s robustness. If we correct an eror in court it would...

… not prove Horizon was working correctly. It would be absurd.
AB yes.
PG well let’s look at Mr Latif’s transaction data again. With reference to the product code list which you did not refer to in your WS. We had to get it under disclosure...

PG so when Mr Latif was being questioned on this it was suggested to him that the two entries either side of it were back office entries he didn’t need to worry about.
Ab yes
PG but if we look at the product codes, one is cash. That’s important, right?

AB yes
PG and this other code - that’s a transfer
AB yes
PG are you just being agreeable or do you understand this?
Ab well - looking at it now yes
PG did you understand this when you signed your witness statement?

[AB says how in prep for the trial she looked more closely at the data again and when she saw what she saw that’s when she made the corrections.]
PG goes back into the data. Event data at F1354.

[sorry I need to reboot - this is forensic stuff - fascinating watch]

PG - AP branch receipt/transfer out slip office copy/and AP branch receipt - transfer in copy.
PG how many transfer in slips do you see printed out there?
AB 2
PG and they’re four seconds apart. Did you notice two transfer in slips were printed out when you made your WS
Ab no

PG two separate receipts being printed out at once was a feature of the Dalmellington bug, wasn’t it?
AB I can’t recall
PG let’s go to it. This [doc] is in the table annexed to Mr Parker’s doc. “Unexplained sicrepancies, duplicate rem in” and if we go to Mr Coyne’s comments...

… [Jason Coyne is the IT expert for the claimants]
PG did anyone in your team flag up to you that 2 receipts were printed out?
AB no
PG did anyone flag up to you that this was a feature of the Dalmellington bug
AB that’s because Dalmellington only affected outreach

PG no one pointed this out to you, did they
AB that’s because Dalmellington ONLY affected outreach
PG does anything about what you’ve said about Mr Latif’s case worry you
AB well i’d like to look at again
PG I understand

[Ab suggest there was nothing to suggest this was a bug from what she saw]

[we have moved on to transaction data again]

PG did you look at THIS data when you did your WS
AB at a high level. I’ve done some more work since, hence the changes to my WS

PG let’s go to your WS when you say what your theory is here. You haven’t referenced the data. It makes it difficult to see where and how you backed up your theory in your WS.
AB agrees
[But PG has it now - so we are going to it]

[we are going through a series of transaction logs which show sums of money which go up and down in the branch - explains that this was not searchable with the raw data - it had to be done by eye]
PG did anyone point this out to you when you wrote your WS
AB no
PG since?

AB yes we looked at it again

PG it may not be something that you know about or understand, you’re not required to understand it

AB my team are closer to it but I interrogated it - 2 lots of scratchcards came in the wrong way - they weren’t effectively remmed in and you can...

… trace that through as it wen through the branch.

PG okay well i’ll put this to you. we have two £25s which are positives. And if one should have been a minus he will be £50 short? Yes?

AB but he’s got two negatives here

PG yes but…

J let me try to speed this up

J PG is putting this to you in steps. His first step is a premise. You might not agree with the premise but follow it through

PG okay if my hypothesis is right it would have showed some £50 difference rather than zero. So on this premise - as he didn’t have the stock...

… he would think he was due a transaction correction for 50 cards rather than zero.

And when that TC comes through he would want his TC to increase his stock and reduce his cash…

AB yep on that premise

AB he should have had 2 x 25 = £500. You had to get them remmed in so they balanced. What came in took his £500 to zero - it took it down to £1000.
PG but he’s complaining about £500
AS Mr Latif would remm in scratchcards and then sell them to his retail

AB a lot of epopel do that. That would show as a sale. But here because he had negative stock figures, he couldn’t balance. So he then reversed the sale of the £500 but never returned the cash [?] so the scratchcards in retail hadn’t been paid for.

PG let’s look at the call logs…. 10 August 2015… “we’ve received the wrong Transaction Acknowldgment for £5 cards”
13 Aug 2015 TA’s received, been charged for free games
27 Aug 2015 rollover with discrepancy caused by lottery TAs being wrong

PG it’s happened quite a lot to him
And on 25 Jan 2016 - he’s got a duplicate lottery TA there as well. So there’s a history of him having these duplicate lottery problems and raising them with the helpline
PG now lets’ go to 25 Jan 2018 “TC received for £10 x 1000...

… scratchcards. If we accept this we will create a discrepancy.”
AB he booked in -£500 instead of £500
PG but he’s complaining about accepting a TC for 100 x £10 scratchcards. That’s his discrepancy.
AB but the value of the TC was £1000
PG precisely and that’s why it...

… overshot by £500.
AB because he remmed it in, instead of increasing it decreased. Instead of going £500 to £0 it went to £1000. It is confusing to him, but because of the way he sells the stock to the retail that is why what happened happened.

J says he entirely understands all the arguments being presented by Mr Latif, Mr Green and Mrs van den Bogerd with relation to this issues and doesn’t think there is anything further to be gained from Mr Green’s continued questioning.

We have been given an 8 minute break.

AB was pretty clear she was sure that was a user error in branch and no incontrovertible evidence from PG it was a Horizon error…

Win for the Post Office, I think. And I’m not sure the point before about the transaction data completely landed.

As m’colleague @Karlfl has just pointed out - part of the problem that is being highlighted is the difficulty of getting to the bottom of a potential problem - when AB has a team of 10 people working for her she still has to make corrections to her WS on the data. SPMS don’t...

@Karlfl … even get the data.

Yebbutt this trial is about whether SPMs have been held liable for incontrovertible Horizon errors.

we are back in court

We are on to Mr Tank’s situation in par 77 of AB’s WS (which is on postofficetrial.com).

PG you are blaming what happened to Mr Tank on user error
AB not entirely
PG what I will suggest to you is that ANY error by an SPM is used as evidence for a Horizon error...

… even if it is not causative. And yet in Mr Tank’s situation it was not causative and you state if he had not made his mistake it would not have happened. And that’s not correct, is it?
AB not in this case no

PG have you heard of UEB - user error bias?
AB no
PG it’s when people in IT routinely blame users first and I say that is the Post Office’s method with SPMs.
AB 99 times out of a hundred it will be user error, but I accept in Mr Tank’s case it wasn’t.

… yet you in your WS say it was.
AB yes
PG he specifically did what the Horizon system told him to. and you blamed him for it.
AB yes
PG goes through the error and explains the process. Is this Horizon working as it should?
AB in that it was caught in the back end, yes.

PG did you know there were more than 400 failed transactions at the specific time of Mr Tanks’ error?
AB no
PG and a KEL (known error log) was raised about this. Did anyone in your team tell you about this KEL?
AB no
PG reads from design of Horizon that a discrepancy...

… which occurs during certain recovery processes will not show up in branch transaction data, because the priority is to get the branch up and running. The SPM will get a discrepancy and must raise it with helpline to sort.
PG That’s Horizon working as it should?
AB yes

PG gives helpline log of a SPM who had a Horizon session collapse and lose a load of transaction data. It is Mr Tank’s story of trying to get to the bottom of a discrepancy: postofficetrial.com/2019/03/horizo… (search for “couldn’t promise anything”)

Mr Green asks if they way Mr Tank was treated was acceptable
AB no
PG typical?
AB well some people have bad experiences
PG have you heard of the helpline being called the hell-line before
AB yes

[first time POL management have publically agreed that helpline is called hell-line by SPMS - they all denied it on oath the witness box when asked in the last trial]

PG why was Mr Tank’s call afforded low priority
AB it doesn’t mean it wasn’t taken seriously

@threadreaderapp unroll pls