The prototypical legal “jury” consists of twelve random and anonymous citizens who are to passively hear ruling-relevant info while isolated from outside influences, to talk together in person for a few hours to a few days, and then to issue a binary ruling regarding which they have no obvious incentives.
Another concern is how this works in light of plea bargaining. The majority of cases plea out based on a defendants prediction of likely outcome at trial so the actual cases that go to trial aren't a representative sample.
Thus you aren't necessarily guaranteed convergence to the desired outcome. What you need is a mechanism to force a certain fraction of cases to be tried despite a plea bargain which seems unworkable since the defense no longer actually has an interest in the outcome.
Having said that I actually like the idea of denying all plea bargains in a random sample.
--
To give an example, suppose that the population of criminals consists of 99% rational actors who don't want to go to prison and 1% of actors who plead not guilty because they want the media coverage and believe they acted justly if illegally. Suppose juries (correctly even) convict 100% of people in that later category. One equilibrium for your system is the market predicts all juries convict so the actors in the 99% rationally choose to take the plea bargain rather than betting on the extremely slim odds of getting a real jury and winning leaving the only ones who go to trial the 1% which in turn vindicates the market's rating of 100%
It's not clear to me the extra predictive accuracy this kind of approach provides is desierable.
If this works you get a bunch of detailed analysis and produce models which are very good at predicting jury outcomes (and implicitly simulated jury outcomes). That essentially provides a blueprint for how to commit crimes in a way that doesn't result in punishment.
Yes, in some sense much of that data exists now buried in court records but the amount of effort required to extract it makes it non-worthwhile for criminals to do so.
If cost is not an issue, you are going to get a more accurate judgment by taking an ensemble of multiple diverse juries, instead of selecting a single "best" jury. https://en.wikipedia.org/wiki/Ensemble_learning
Instead of aggregating jury quality (ie the unique combination of individual jurors) how about jury selection via a standardized “prediction market test” where all potential jurors are provided tokens to bet on outcomes of historical cases given certain fact sets. Juries are then selected by best performers. Participation could be incentivized with rewards for being selected into a jury. This would be a meaningful reversal of how most people I know treat jury selection (purely downside!)
I think what you are proposing counts as a "jury mechanism" by my definition. So you could propose it in a tournament, and it could compete with all the others.
How to prevent pressures in jury selection? In actual court cases, the most predictive thing for case outcome is juror selection, not what is presented during the trial. Lawyers get a few strikes without cause, and can also strike jurors for cause. A big part of case strategy is trying to keep or move it to a district favorable to own side, vastly weighting jury pool. They are anything but random due to selection rules, and lack the kind of incentive for accuracy that you get with prediction markets. Maybe less true if marketsselect them, but this seems very noisy.
Using prediction markets to select them also requires the opposite of anonymity. A prediction market can make guesses with limited info better than a random dude, but when sufficiently limited, maybe not by too much. Some non law juries per post could have lots of useful public information. For those that don't, value might be limited.
Jurors also act beholden to judge instructions, sometimes inconsistent with law, because courts deliberately suppress awareness of nullification. This tracks to other jury variants too; you can break outcomes merely by limiting which discreet options they consider as a group.
Sure, and you can ask if that's a good idea but the prediction market is guessing at the overall outcome of that process so it essentially incorporates the entire process.
Agreed, but only insofar as prediction market has useful information about the process.
I guess my real doubt here is that presently, juries are not selected with the goal of truth or legal consistency, and that presumably whatever incentives drive that will both resist using prediction markets and attempt to undermine them.
Thus, I suspect using them is better than not, but that you might get more milage out of changing incentive structures in court. Especially for prosecutors and judges.
Also implies the idea will work better with juries in a non legal context, because you have functional incentives in more than just the market itself.
Maybe we could use a tournament to pick the best jury for any given case
and if not predicting its average distance from other tournaments entries
The winner, who gets a reward
Kind of similar to a mob, gladiator, or UFC ring. I like it.
Goes with Higher Intelligence -UFO- having domesticated us. Man, the zen of knowing just a bunch of monkeys Fighting over a banana.
You really want efficient jury? Use fMRI of Caudate Putamen, and Nucleus Accumbens brain scans to rate current population intelligence.
Which will just get you to breeding, which is waste of time as you can use AI orchestrated 'perfect' genome human, which will make laws superfluous as all will be efficient not needing disagreement resolution becasue super intelligent efficient.
Another concern is how this works in light of plea bargaining. The majority of cases plea out based on a defendants prediction of likely outcome at trial so the actual cases that go to trial aren't a representative sample.
Thus you aren't necessarily guaranteed convergence to the desired outcome. What you need is a mechanism to force a certain fraction of cases to be tried despite a plea bargain which seems unworkable since the defense no longer actually has an interest in the outcome.
Having said that I actually like the idea of denying all plea bargains in a random sample.
--
To give an example, suppose that the population of criminals consists of 99% rational actors who don't want to go to prison and 1% of actors who plead not guilty because they want the media coverage and believe they acted justly if illegally. Suppose juries (correctly even) convict 100% of people in that later category. One equilibrium for your system is the market predicts all juries convict so the actors in the 99% rationally choose to take the plea bargain rather than betting on the extremely slim odds of getting a real jury and winning leaving the only ones who go to trial the 1% which in turn vindicates the market's rating of 100%
I'm also okay with denying plea bargains in a random sample.
It's not clear to me the extra predictive accuracy this kind of approach provides is desierable.
If this works you get a bunch of detailed analysis and produce models which are very good at predicting jury outcomes (and implicitly simulated jury outcomes). That essentially provides a blueprint for how to commit crimes in a way that doesn't result in punishment.
Yes, in some sense much of that data exists now buried in court records but the amount of effort required to extract it makes it non-worthwhile for criminals to do so.
If cost is not an issue, you are going to get a more accurate judgment by taking an ensemble of multiple diverse juries, instead of selecting a single "best" jury. https://en.wikipedia.org/wiki/Ensemble_learning
But of course cost is always an issue. And an ensemble of juries *IS* still a jury by my definition.
Instead of aggregating jury quality (ie the unique combination of individual jurors) how about jury selection via a standardized “prediction market test” where all potential jurors are provided tokens to bet on outcomes of historical cases given certain fact sets. Juries are then selected by best performers. Participation could be incentivized with rewards for being selected into a jury. This would be a meaningful reversal of how most people I know treat jury selection (purely downside!)
I think what you are proposing counts as a "jury mechanism" by my definition. So you could propose it in a tournament, and it could compete with all the others.
How to prevent pressures in jury selection? In actual court cases, the most predictive thing for case outcome is juror selection, not what is presented during the trial. Lawyers get a few strikes without cause, and can also strike jurors for cause. A big part of case strategy is trying to keep or move it to a district favorable to own side, vastly weighting jury pool. They are anything but random due to selection rules, and lack the kind of incentive for accuracy that you get with prediction markets. Maybe less true if marketsselect them, but this seems very noisy.
Using prediction markets to select them also requires the opposite of anonymity. A prediction market can make guesses with limited info better than a random dude, but when sufficiently limited, maybe not by too much. Some non law juries per post could have lots of useful public information. For those that don't, value might be limited.
Jurors also act beholden to judge instructions, sometimes inconsistent with law, because courts deliberately suppress awareness of nullification. This tracks to other jury variants too; you can break outcomes merely by limiting which discreet options they consider as a group.
Sure, and you can ask if that's a good idea but the prediction market is guessing at the overall outcome of that process so it essentially incorporates the entire process.
Agreed, but only insofar as prediction market has useful information about the process.
I guess my real doubt here is that presently, juries are not selected with the goal of truth or legal consistency, and that presumably whatever incentives drive that will both resist using prediction markets and attempt to undermine them.
Thus, I suspect using them is better than not, but that you might get more milage out of changing incentive structures in court. Especially for prosecutors and judges.
Also implies the idea will work better with juries in a non legal context, because you have functional incentives in more than just the market itself.
Maybe we could use a tournament to pick the best jury for any given case
and if not predicting its average distance from other tournaments entries
The winner, who gets a reward
Kind of similar to a mob, gladiator, or UFC ring. I like it.
Goes with Higher Intelligence -UFO- having domesticated us. Man, the zen of knowing just a bunch of monkeys Fighting over a banana.
You really want efficient jury? Use fMRI of Caudate Putamen, and Nucleus Accumbens brain scans to rate current population intelligence.
Which will just get you to breeding, which is waste of time as you can use AI orchestrated 'perfect' genome human, which will make laws superfluous as all will be efficient not needing disagreement resolution becasue super intelligent efficient.