A Place to Stand

   In 2009 I wrote a Big Engineering article on offering a prize for an open source computer programme able to function as a judge on matters of international law - a relatively simple field because there isn't much of it.

   Big Engineering 40 World Peace By Prize

   With Moore's Law (computer capacity doubles every 18 months) it has now increased some million times since 1980 & 250 times since 1997 (now 10 million or 2,500 times). Time for something bigger. The golden aim of such computer science is something that can pass the Turing Test - be able to carry out a conversation indistinguishable from that of a human (& more technical way of saying a computer with human level intelligence). I will aim for something lower - or higher.

I propose a series of prizes should be put up for a process culminating in the establishment of an open source computer programme which could determine the legality of national actions under international law.

The prizes should start with something capable of rendering a simulated decision in a war game atmosphere which both sides agreed was satisfactory. Then a larger prize for something used in a real situation & ultimately for one which
successfully acts as a judge on the International Court of Justice, or if that option is refused, is able, on its own to render judgement in a years worth of different cases with judgements not agreed by a worldwide panel, to be inferior to the ICJ rulings. It is obviously necessary that, though the copyright remains in the designer's hands, the programme be open source so that it can be checked & run by anybody who needs to be able to trust it.Also here 

Law is a very computer like system with either/or decisions, set rules, logic & great importance laid on previous examples. As such it would be much easier for a computer programme to impersonate a superior judge than an ordinary human being.

It is overstating to say this would provide peace on Earth but it would provide a framework for it. Conflict usually occurs when both sides have convinced themselves they are in the right. Even where it isn't it is usually important for bystanders to be able to make a decision on that order.

The fact that a third of the judges believed in the reality of the boy – this is a very significant figure. Under the terms of the test, developed in 1950 by the legendary scientist Alan Turing, it can be considered for the program passed if at least 30 per cent of judges are confident in communicating with the person."

    Of course those who don't want to accept it say that 30% approval is to low a bar and so on, and they may have a point, but it doesn't matter. Moore's law is continuing and progress is going to continue.

    Ray Kurwzeil here call it a premature announcement; that he thinks the test somewhat fixed and he "has seen better". This is probably true however being a 13 year old boy, on all subjects, is clearly a much more complicated task than being a Supreme Court judge discussing only the subject of the logical processes of the law. Whether a true Turing test or not it is clearly enough for a judge programme.

   There can now be no reasonable dispute that that proposal is currently practical. And not much, that we can go further.

   Which is why I am now proposing that the same process be applied to interpretation of constitutions. The whole point of a constitution is to say what government can, cannot & possibly must do and government itself is obviously an unsuitable entity to interpret this. Which is why we have "separation of powers" and judges, nominally independent of government deciding - except that, being human appointees there can be no certainty of them being independent.

   But an open source computer programme has to be because it can be tested by anybody (well anybody who knows how)  and while many will dislike some decisions, because they will dislike the constitutional assumptions (republicans in Britain, gun opponents in the US) they will have to be consistent, whoever is making the case - that's what impartial justice means.

    We should design a programme to act as a Supreme Court for interpreting constitutional law. In turn new laws, or indeed a new Scottish constitutional settlement can be tested before it is enshrined. making laws mean what we want them to mean is most of constitutional law. With this programme that can be done quickly and easily. For example we have little idea how the redefinition of marriage will affect other laws, and currently we have no way of definitively knowing until after, perhaps decades after, passing it.