27 March 2005

Complex fraud trials

I admit I don't understand what's so complex about fraud trials.

As I roughly understand the law, in order to prove someone guilty of fraud, you have to show:
  • They said or wrote something that wasn't true
  • They knew it wasn't true
  • They gained money or goods as a result.
The problem seems not to be "complexity", but rather a very large quantity of evidence intended to prove or cast doubt on each assertion. In the Jubilee Line case, the jury seems to have had all this evidence lumped on them over a period of getting on for two years, with the idea they would eventually be left to decide on it all.

The idea that generally crops up at this stage is to do away with juries, or, as suggested here, to change the composition of juries.

To my mind, a less drastic response would be to change the programme of a trial.

I, like many others, have been following the progress of SCO v IBM, via the Groklaw blog. Now that's happening in the USA, but the main difference is that it's a civil matter, and so far no jury has been involved.

As you follow the development of this genuinely complex case, what happens is that questions are answered one at a time. One side raises something they want to be decided, the other side submits arguments against, and the judge rules one way or the other.

I don't see why this can't be done in a criminal trial with a jury. The jury is there to determine the facts, why can't it be done one at a time? The prosecution allege that some detail is true, the defence deny it, both sides present evidence on that one point, and the jury decides. The prosecution, having got rulings on all their facts, make the case based on proved facts. The defence declares that there are other relevant facts that have not been covered, and those are decided in the same way. Finally, arguments are made, again based on facts that have already been considered to have been proved.

This has two advantages: First, the job is made simpler for the jury. Second, the same jury need not handle the entire case, if it takes four years or something.

Once again, I am extending my professional techniques beyond their normal scope: Earlier today I described BST as bad data modelling, now I am suggesting that the court system employ modularity. Every beginning programmer is taught that complex tasks are made simpler by separating any subtasks that can be handled independently.

It could be argued that this system would reduce the independence of juries, and would compromise our traditional liberties. Maybe so: it certainly would need very thorough study. But, as in the case of the anti-terrorist control orders, people are talking about abolishing traditional liberties, without really looking at less drastic ways they could be tweaked to achieve the same ends.

No comments: