How was this sausage made?

What was the one development without which the Post Office scandal could not have happened? In a bitter irony, tinkering with a law enacted following a serious miscarriage of justice – the Confait case – the Inquiry report (yes, another one!) is here) enabled what is now the worst miscarriage of justice in English history. The law is the Police and Criminal Evidence Act 1984 (“PACE”); the tinkering is to S.69 – its removal and replacement by – well, nothing. 

Yes – it is time to look at how the Law Commission reached its recommendation. And, in the second part of this (here), the scrutiny MPs gave to the Commission’s recommendation.

This matters for three reasons:

1. What the Law Commission did and how and Parliament’s role are outside the scope of the Williams Inquiry;
2. There are decisions now to be made about how to correct this, decisions which Ministers are seeking to avoid; and
3. This is not just a miscarriage of justice, a cover up, an IT failure, corporate failings by Post Office managers. It is a failure of governance by Parliament and  Ministers at every stage. It continues now.

Section 69 

This dealt with the admissibility of evidence obtained from a computer. To use it, a prosecutor (or defendant) had to prove the computer was operating properly (and if it wasn’t, this did not affect the accuracy of the document provided) and had not been used improperly. Important to note two points – because they lie at the heart of what the Law Commission got wrong:-

(1) admissibility relates only to whether something can be evidence (an exception to the hearsay rule);

but

 (2) it says nothing about the reliability, accuracy or completeness of that evidence.

The latter is a factual assessment of what a particular computer system is designed to do, how accurate it is for those purposes and whether that purpose and level of accuracy are sufficiently reliable in a criminal trial.

The Experts?

In 1997 the Law Commission recommended its removal. This meant that computer evidence put forward by the prosecution would be accepted as admissible and true unless the defendant could prove otherwise. This had a number of important and serious consequences:-

1. It shifted the burden of proof onto the defendant in any case reliant on computer evidence;
2. Defendants would need access to the necessary expertise and the money to pay for it if they were to rebut this presumption; and
3. The prosecution would have to be scrupulous about disclosure relating to the workings of the computer systems from which it obtained evidence. 

Surely this last could be taken for granted? No: many of the miscarriages of justice in previous decades were because of police disclosure failings (e.g. the failure to reveal alibis in the Guildford Four case). Hence the creation of the CPS. But what of prosecutors not part of the CPS? 

The Commission’s analysis has been eviscerated in two articles here and here. In summary, it showed a woeful ignorance of how computer-based systems, particularly complex ones, work and the difference between hardware, software and data. It failed to understand the concept of technical reliability of complex computer systems (depending on their purpose), margins of error and the difference between this and the legal reliability of data produced. It consulted very narrowly; and it misrepresented what those computer experts it did consult said, so badly that it recommended the precise opposite of what they had advised. This is a very serious failing by a body whose expertise is the law not computer science. 

There is worse. One of the bodies lobbying hard for this change was the Post Office. It stated that this provision made it hard for it to prosecute its staff because of the difficulties of showing that its computer systems were working properly. The Commission appears not to have noticed that this was an entirely self-interested argument. Nor did it consider the Post Office’s obvious conflict of interest. Nor does it seem to have worried that, if a prosecuting authority no longer had to prove its systems were working properly, it had no incentive to ensure they were in fact doing so – at least to the standard required by the criminal courts. Nor did it ask itself whether it was right to shift the burden of proof to the defendant in something so important and complicated as computer evidence, given the then widespread use of computers and the likelihood of this increasing still further. Nor did it state how, if this shift were made, defendants would in practice be able to prove that the computer systems were not working properly, given the need for expert evidence and detailed technical disclosure. 

It did ask the question:

“what sort of evidence must the defence adduce, and how realistic is it to suppose that the defence will be able to adduce it without any knowledge of the working of the machine?“

(Note the confusion between hardware and software. And the description of a complex computer system as a “machine”.) 

It airily gave this de haut en bas answer:

“a party cannot be required to produce more by way of evidence than one in his or her position could be expected to produce. It could therefore take very little for the presumption to be rebutted, if the party against whom the evidence was adduced could not be expected to produce more.“

There is so much wrong here it is hard to know where to start. 

• First, what is necessary as evidence is what is needed to satisfy a jury to the right standard. The prosecutor says the computer works fine. The defendant says: “The computer system is wrong but I can’t prove it” (much as subpostmasters did ad nauseam). What on earth is a jury going to make of that? The strength and validity of evidence is not determined by a defendant’s ability or resources. 
• Second, the belief that only a “very little” would be needed is a pious hope. It is not set out in law as the low threshold which must be passed. 
• Third, it completely ignores the practical problems of meeting even such a low threshold, especially in complex systems, unless full disclosure is made of all the problems and defects in such a system. It ignores the reality that, even if disclosure is made, expert evidence would be needed, a cost which most defendants cannot afford and which legal aid won’t often pay for.
• It did not properly consider what sort of disclosure would be necessary by the party relying on computer evidence nor how to ensure that this should be produced in a way fair to all parties. 
• Finally, the Commission seemed unaware of the concept of plea bargaining or defendants feeling forced into admitting guilt if they could not afford to mount a defence. 

It was bad enough for the Commission to get technical matters wrong as this undermined the recommendation it made. But not properly considering these fundamentally legal questions and presenting lofty answers untethered to the reality of the criminal courts was utterly inexcusable.

The Myth of Expertise

Why did the Commission take such a flawed approach? One answer which has been given is that it fell into the all too common trap of believing what it wanted to be true then finding the opinions justifying this.

Another is the myth of expertise. It is not uncommon for experts in one area to think that they can become experts in another without too much difficulty. It is a trap which lawyers all too easily fall into. Their ability to understand non-legal material, query it, explain it to juries, use it to make legal points can give the illusion of being an expert in it. But being able to question something, however expertly, is not the same as being an expert in it. Lawyers are not computer scientists. The Commission’s report was based on a superficial, faulty understanding of a very different discipline. 

In thinking that S.69 could be removed with no problems arising or unintended consequences, the Commission was, ironically enough, making the same mistake IT engineers sometimes make when fixing a bug in software – assuming (or hoping) that one change would not have other unwanted, unintended or unknown consequences. Complex computer systems are dynamic; so are legal systems. The criminal justice rules are like Jenga: if you pull one of the blocks out the whole thing can come tumbling down. The judgment is knowing what you can remove or change without undermining the tower.  

The issue should not have been what to do with S.69. Rather it should have been how to ensure the courts could properly assess evidence from computer systems, particularly complex ones, in a fair, effective, realistic way, taking into account the realities of such systems, disclosure obligations requirements and the need to have a fair trial. This was not done. 

It is unsurprising that one of the experts consulted – Alistair Kelman, an IT engineer turned barrister – wrote this in 2006 in his book “The Computer in Court”:

“Today there are virtually no controls over putting computer evidence before the court….. The opportunities for fabricating computer evidence and …. destroying or altering audit trails are growing.

Although ….. I believe that nobody has yet been convicted in the UK of a crime they did not commit owing to inaccurate computer evidence ….. well publicised miscarriages of justice such as the Birmingham Six and the Guildford Four show how dangerous it can be to accept technical evidence in the courtroom at face value.“

Prescient words. At the time he wrote this, many subpostmasters had already been wrongly convicted on the basis of inaccurate computer evidence, though this was not yet widely known.  The reference to the Irish miscarriage of justice cases is also key. The Law Commission might not have joined – or even be expected to join – the dots between what happened there and what would happen if S.69 were removed. But Parliament surely would have looked at the issue more widely.

Wouldn’t it?