A Data Science Central Community
Do you think that it is possible to design a system that automatically sentences convicted persons, using historical data and legal precedents set by courts? What would be the challenges in implementing such a system? Could it be used to accelerate (say) 50% of the cases showing up in court? Would it remove some arbitrariness and reduce costs?
In other words, could we outsource some of the judge's tasks to a computer program, to allow them to focus on more complicated cases?
- I've seen something like it already, in traffic "fines";
- It DOES accelerate the process;
- It fills a document with all the parameters (in the law) it can automatically, and those who can't jurist will fill them in a "User-friendly" way;
- Last decision is human thought (e.g. a jurist), and that's the way it should be;
- A person can apeal in court the sentence;
The problem lies in the number of appeals a person can do, but worst in the time periods (contemplated in the law) for each phase of the process in each appeal. Leaving crucial parameters who need to be "filled" "unfilled";
Just my 0,02 cents;
If it's sentencing only system (e.g. it doesn't involve judgement of guilt or innocence) ... sure! It can be done because, in the states, most judges work within sentencing guidelines ... formulas ... a rules-based system. And in that regard this kind of system in already is in place.
From there, it gets a bit more problematic. This elementary kind of system wouldn't involve the discretion of judges within guidelines, which are often in play. To include discretion would require a much more involved measurement system that would need to involve thinking based upon an almost infinite array of evidence possibilities in addition to the defendant's history ... a daunting challenge, I think.
It would be fun to look at actual conviction patterns of judges to see who the outliers are ... and why.
It would certainly be interesting to look at applying analytics to sentencing, at least as an academic exercise. Of course, judicial discretion is important to get to just outcome in some cases. However, removing judicial discretion might sometimes increase the justness of trial outcomes. For example, Jonathon Levav and colleagues suggest you are up to 3 times more likely to be released if your case comes up early in the day, as compared to late in the day. As another example, we have all read about differential sentencing based on race. Analytics solutions could be blinded to time of day, and could produce race-neutral sentencing -- or take care to account for bidirectional relationships between race and sentencing. The point is that benefits of an analytics approach may extend beyond improvements in efficiency to reductions in unintentional biases in sentencing.