...free to think freely


13th January 2024

Another legislative mistake?

It seems the government has decided to use primary legislation to acquit the postmasters convicted on the flawed evidence of the Horizon computer system. This is well-meant and urgently needed, of course, because the sheer number of appeals which would otherwise need to be processed would overwhelm the courts for years to come. However, constitutional lawyers warn this could set a dangerous precedent for the future of freedom and democracy, that people can be legislated into guilt or innocence without the need for due process.

This danger is real as well, for it empowers governments in theory to convict and imprison anyone they like without that person having to undergo any recognisable trial. That is why we have yet to see how the bill will be worded as the government takes advice on minimising this effect.

Is the government in danger of repeating the historic mistakes which led to the current mess of the Equality Act 2010? That resulted from specific acts outlawing particular bases for unfair discrimination, ultimately consolidated in a law based, not on principle, but on arbitrary differences and characteristics and giving no protection to those not referenced by those specific categories. It is the consequence of a legislative process too concentrated on specific symptoms of unfairness and lacking in a desire to work out the underlying principles which made the treatment unfair.

Is a specific law for the current cohort of victims similarly flawed. Was the problem simply that one particular organisation with one particular set of software caused massive repeated miscarriages of justice, or is there an underlying problem in the way computers and the information they provide is trusted? In other words, is the problem the law needs to address and correct specific or generic?

That matters because if a law is passed which simply addresses the current set of victims it will be a one-use law which will never be useful again. Next time, another law will be needed, and after that another and another until far from being exceptional this sort of thing could become routine. That really would change the nature of law and the relationship between power, law, and the citizen, making the law an arbitrary and unprincipled thing with inherently retrospective effects.

It might be better to consider that miscarriages of justice caused by trusting untrustworthy data could become frequent as society increasingly relies on software to process information. There will inevitably be further cases where a computer error results in false evidence being put before courts. Maybe, rather than a specific law exonerating the postmasters in this particular case, we need a generic law addressing an automatic or fast-tracked procedure for quashing convictions later found to be based on faulty information and for ensuring in future that computer evidence must meet some sort of standard to certify it is forensically reliable. That, naturally, would introduce new barriers to the use of evidence, but just as the word of a single uncorroborated witness might be questioned on cross-examination, so we need to ensure computer evidence is subjected to some equivalent process of scrutiny. We can no longer assume machines cannot be mistaken. This becomes even more a concern with the appearance of generative AI with its ability to create convincing false images and speeches.

A generic law would not only safeguard people in future and continue to be useful, but it would also not be a judgement by Parliament on one specific group of criminal cases. Rather it would address the issue of quashing unreliable convictions in general and avoid a constitutional clash between legislature and judiciary.

Surely that would be better.