Altman Review III: Relying on the unreliable?

The Altman General Review advised PO in 2013 that the Cartwright King sift (a review of many prosecutions conducted for the Post Office) was fundamentally sound. Was it?

An important issue for Cartwright King’s sift was which cases to look at. That was partly a decision based on the timing of the cases. The CK Sift was limited to cases within a particular period (shortfalls identified by Horizon after 1 January 2010). The logic was the Horizon bugs Jenkins failed to disclose in Seema Misra’s case only applied to Horizon Online, a system in operation from 2010. Cases before that could be ignored.

Altman says, in his executive summary, that the time limits on scope are, “logical, proportionate and practicable in light of all the known circumstances.” (para 5(i)). I note, perhaps unfairly, he does not say the time limits are right.

His analysis later in the report is, however, in tension with his approval. Altman describes Simon Clarke's reasoning for the time limit,

was that any sub-postmasters prosecuted under the former Horizon data regime would have served any sentence of imprisonment, or performed any unpaid work requirement or paid a fine; and at all events the publicity from SS's report would put those defendants on notice.

This shows Clarke basing a critical decision on the scope of the review on something that is patently, one might fairly say crassly, irrelevant. Altman spots this saying people who have served their sentence nevertheless, “have an interest if their conviction was unsafe.” Altman also says, “Resourcing and POL's reputation are also beside the point,” suggesting that these two further factors have inappropriately influenced Simon Clarke’s decision-making.

He then says when he, “queried the rationale behind the cut-off date,” he was told, “prior to each branch rollout, a cash audit was done so that each branch balanced,” and that this justified the date being chosen. I am tempted to joke here that the Post Office said, you don’t like our reasons? We have other reasons!  By swhich I mean, the cash-audit point might suggest this is a post-facto rationalisation rather than a well-justified time limit. But who knows.

Interestingly, although approving the time-limit, it seems to me to give Altman concerns. He says, “ I advised in the conference and repeat here that although POL has no positive duty to seek out individuals before the I January 2010 start date for a review of their case, nonetheless if POL was approached it would need to make ad hoc case-specific decisions about the need for disclosure.” One such case would have been Seema Misra’s case. Jenkins actually gave evidence at her trial. Her case was a pre-Horizon Online case.

He seems to be warning them that the 2010 date might not hold, whilst also suggesting that they can let pre-2010 related cases lie for now unless they are raised by the defendants themselves.

He specifically mentions Seema Misra’s case as an example of the type of case where, “the issues raised in the case, which were made late by the defendant in one or more defence statements, were very similar to those generally being raised currently in relation to the Horizon Online system.”

It is a subtly made point if it is being made deliberately; Seema Misra was raising a concern about the system pre-2010 which suggested the bug discovered in 2010 might also have existed before Horizon Online came in. If others raise a similar defence, he seems to be saying, adding weight to Seema Misra’s claim, then this would suggest the 2010 date might be wrong, “if POL was approached it would need to make ad hoc case-specific decisions about the need for disclosure.” (para 64) He comforts his anxiety here somewhat with Second Sight’s review as, in his words, “[involving] Horizon issues predating the 2010 rollout of Horizon Online…” and yet not discovering (by that time) bugs with the Horizon software that predate 2010. And so he concludes:

In my judgment, the I January 2010 start date for CK's review is both a logical and practicable approach to take. That is not to say however that if a case pre-dating the rollout of Horizon Online presents itself POL and CK should exclude it from consideration.

So the scope test applies generally but not ad hoc. It’s a bit of a have your cake and eat it argument. The cut-off period is logical, but it should also be ignored in appropriate cases. Sometimes. Only if the defendant asks them to review cases. And if Cartwright King think that an ad hoc review is merited.

Again it suggests that Altman is not confident about the January 2010 cut-off which he nonetheless formally supports as logical, proportionate, and practicable. Nor do we see any evidence in the General Review that such ad hoc review was taking place. And we know Altman is anxious about inflexibility. So I think a reasonable assumption for now is that it was probably not taking place.

It’s also worth noting that the view that it should be January had been decided in, “a telephone conference held on 4 October 2013, in which representatives of POL, BD, CK and myself participated.” There is a risk, but again this is speculation on my part, that Altman has made himself comfortable with a collective view that he does not fully share.

Relying on evidence from Jenkins

There is also the rather startling problem that parts of the General Review rely on evidence emanating from a witness statement of Gareth Jenkins (dating from 2013) and appended documents (from 2009 and 2012 respectively).

Altman justifies it in these terms:

Although there are issues about Mr Jenkins' independence and objectivity (with which I deal below), I am content to rely on Mr Jenkins' witness statements (based as they appear to be in whole or in part on the Horizon integrity reports) for these purposes as providing a reasonably adequate, and almost certainly accurate, summary of the Horizon system.

Altman sets out points of relevance that include:

·         The sense that Horizon online was, "a complete re-implementation of Horizon,” which, “utilised a central database to hold details of all transactions rather than the MessageStore used by the original Horizon system." The scope of the CK sift is confined to 2010 onwards because Horizon online is felt to be different from legacy Horizon. This is a fundamental assumption which should have been thought about and subject to critical scrutiny.

·         The timing of migration from original Horizon system (again critical to the temporal scope of the review).

·         Descriptions of how Horizon stored and replicated counter data, captured data failures, created and securely stored audit files against tampering and corruption (para 19) detected lost records (para 20) and rendered failures visible to the user (para 21). The description is extensive and would, to most readers without knowledge of the case in the light of history, convey a sense of a system that was inherently accurate and secure.

A particular, and important point that arises from reliance on Jenkins is that I cannot see how one could come to view that the time-period of the review is correct without understanding the changes made in Horizon in 2010 and the way the Horizon before and after this date. Jenkins is seen as unreliable, and it is clear they have not been able to identify an alternative expert. Indeed, as we have seen above, Altman has spotted that Seema Misra’s case suggests the post-2010 problem was indeed apparent before 2010. The point is simply not addressed save through Altman’s oblique references to the potential for new material coming to light over and above the Second Sight and Rose reports through hub meetings and so on.

So I have doubts about the appropriateness of the scope test and think Altman, though suggesting his ad hoc approach, and his awareness of Seema Misra’s claim, may have doubts too. The temporal scope is one of two absolutely critical substantive judgements made on each prosecution review by Cartwright King. The second is whether once a case is reviewed disclosures needed to be made. The issue there is whether Cartright King thought there was material that needed to be disclosed because it raised doubts about the safety of convictions.

We’ve seen here that the temporal scope test was justified on flawed grounds, but a logical justification was then offered. That the test itself needed to be flexed on an ad hoc basis, but there is no evidence here that it was. And that an understanding of Horizon was relied upon which was relevant (I think) to this scope test (and the safety test) and emanated from a tainted witness. Tomorrow’s post will look in depth at the application of the second test: the disclosure test.