And why do verdicts have to be binary–i.e. either “guilty” or “not guilty”? What if jurors were allowed instead to rate or score the plaintiff’s case, just like people rate movies or restaurants. In our most recent working paper, Why don’t juries try ‘range voting’? (pun intended), we propose just that: a simple “range voting” method for juries in which jurors would rate or score on a scale of zero to ten (or some other specified scale) the evidence presented by the parties at trial. The jury’s verdict would then consist of a numerical value, either the average or the sum total of all the individual scores. A plaintiff would prove his case only if the average value or sum total of the jury’s collective score exceeded some critical threshold. Since the jury’s numerical verdict would then be a function of this range voting procedure, we refer to such a numerical verdict as a range verdict. (By the way, we also discuss how range voting improves jury accuracy AND solves several problems endemic to juries, including holdouts, strategic jurors, and ignorant jurors.)
Our little paper is still a very rough proof of concept, so your critical comments and suggestions, especially destructive criticisms, are much appreciated by prior probability.
prior probability 
Although I am a fan of range voting I’m not sold on this idea at all.
A jury has characteristics which make voting method almost irrelevant.
Namely:
* a limited number of people
* together in one physical space
* able to deliberate for a sufficiently long time
* decision(s) are binary in nature – eg. find in favor of the prosecution or not for one or more charges
These characteristics mean that the jurors can determine whether the burden of proof was made or not in a free-form deliberative manner where the binary question each juror has is a boolean ‘was the burden of proof for X met’ question.
—
Another criticism is that the paper does not discuss metrics for determining whether a ‘range jury’ gets better results than a ‘traditional’ jury or not. You could describe experiments that would validate:
* how often a range jury comes to a different decision that traditional jury
* how quickly a range jury comes to a decision vs a traditional jury
* how often a range jury is ‘right’ vs ‘wrong’ as compared to a traditional jury
One experiment framework would be to have 2 jurys exposed to the same trial that operate independently and compare how they do. This would need to occur with many juries over many trials. There would also need to be a control framework to identify what level of variation happens at baseline due to having multiple (traditional) juries.
This is, of course, very expensive to do with actual trial juries. You might be able to devise easier social experiments that can get at the same conclusions though. For example a ‘trial’ where all the testimony and evidence is in written form, knowledge of what the actual verdict should be known ahead of time and jurors participating in a time boxed forum. Forum could be physical or virtual for experiments. (My sense is that losing the physical presence would change the jury process too significantly though.)
Awesome points … all of them well-taken … like a good Bayesian, I will go through them carefully and “update” my priors (and my working paper) accordingly!