>>283991039 minutes in, Leone Castar says he looks for collaborators do develop a chapter
how a postcapitalist judiciary would work. Well, I don’t use Discord or Zoom and I prefer to remain anonymous, so I will just drop my answer in this thread.
JURY SYSTEMTo get inspiration, we start with a brief look at a
very simple analytical model: There is a group of people trying to answer some question about reality by voting between two options, one true and one false. Each person in the group got more intelligence than zero, so their chance of picking the right option from the two is a bit better than 50 %. Assume the same percentage for each person, with no particular correlation between people. That’s the model. What follows from this? Three things:
- The banal thing: The best approach to finding the truth here is to just go with the option that gets more votes, not requiring a super-majority or people voting for representatives who then vote.
- The beautiful thing: We can get arbitrarily close to being correct 100 % of the time if the group is big enough. These two conclusions are known as Condorcet’s Jury theorem.
- A tricky detail: If you have a group size that is odd and you just add one person, the probability that the group gets the right decision just stays where it is. So the group size should be odd.
When it comes to punishment, I’m thinking of something a bit more complicated than just two options. I think the topic fits into a
one-dimensional spatial model. I think such a model fits because we can think of prison time as a quantity shown on a line, and everybody agrees what more or less means here. This agreement is necessary, but not sufficient for a really good fit. It is also necessary that people see themselves as standing on a point on that line. A counter-example: A parliament discusses different proposals for military spending and there might be consensus on how to order these plans from small to big, but the politicians may not all see themselves as standing on one point on that line, but rather with each foot on one end. On that issue, one may be of the opinion: “Go big or go home.”
The punishment in a criminal case is to be decided not by a judge, but by the interplay of a
sortition-selected jury and laws from a
sortition-selected legislature.
Each jury member expresses where they individually stand on the line. For the aggregated jury voice we take not the arithmetic mean, but the
median. That is, we order the ballots on a line from smallest to biggest stated punishment, and then we walk over the ballots. When we have just walked over half the ballots, we stop, and we have the answer at our feet. A jury majority wants no bigger punishment and a jury majority wants no smaller punishment.
(By the way, there have been some estimation contests, like guessing the weight of an ox or the amount of marbles in a jar, and the median has usually been the more accurate estimate than the arithmetic mean. Francis Galton wrote about this.)
Picture yourself as a jury member who believes to know better than the others what the proper size of punishment should be. If the arithmetic mean were taken as the aggregated jury voice, you would have an incentive to estimate the votes of the others and then to
strategically exaggerate your difference on your ballot to get a result closer to your ideal. Not so with the median. (Aside: If the number of voters were even, there would be the need for an extra rule to deal the two middle votes in order to avoid exaggeration incentive, like flipping a coin or always taking the lower one.)
When the legislature designs a law describing some kind of wrongdoing, the people in the legislature vote with two ballots, one for the
recommended minimum sentencing and one for the
possible maximum sentencing. We don’t want that the legislature can easily turn the juries into a powerless appendix, so we use
super-majorities here. We saw with Condorcet’s Jury theorem that the majority has a special meaning. Requirements for super-majorities usually seem rather arbitrary, but here we use something meaningful. We order the ballots for the minimum recommendation, and walk from the end with the biggest stated value towards the other end, stopping once we have passed over ¾ of the votes. Likewise the other way around: We order ballots for the possible maximum sentencing and walk from the smallest stated value towards the other end, stopping once we have passed over ¾. These super-majorities of over ¾ have their basis in protecting an “ordinary” majority: Doing the two counts this way, over half the people voting will see both their voted values included within the two aggregated values.
For either act of voting in the legislature a voter not stating a value that is equal to the aggregate answer will either land on the lower-value or the higher-value side; and when we counter-factually assume a scenario of that voter voting with a ballot exaggerating the difference to the aggregate result, we can easily see that such behavior would not have changed the aggregate result, just like with the jury’s median aggregate.
What punishment does the defendant receive in the end: The jury cannot make a harsher sentence real than the possible maximum set by the legislature, but the jury can always make real something below the minimum recommendation of the legislature, even
zero punishment.
The issue of the received punishment is not quite identical to what the jury can express: The jury can always express a wish for a harsher sentence than the maximum set by the legislature, even though the actual punishment gets capped at that maximum. This is valuable feedback for the legislature to re-evaluate things. If the jury’s aggregate voice going out of bounds happens often enough with a particular type of wrongdoing, the update of the recommended minimum and possible maximum could even be automatic. (But an increased maximum cap for a particular type of wrongdoing should not be used to
retroactively increase sentencing.)
Is the question of punishment really always about an amount of an obvious “punishment substance” expressed as cardinal data though? Why is it that a one-dimensional and highly abstract approach to punishment was so obvious to me in the beginning of writing this? Perhaps it is because I am living in capitalist society where almost anything is measured in a one-dimensional abstract way. Perhaps a new society with more focus on concrete use-values will also react to the many different ways of wrongdoing with many different concrete approaches, and more focus on what offenders should do for their victims. An intriguing thought.
The voting procedures that use scales
can actually handle combinations of different types of punishment if these are expressed as a
ranking of punishment bundles, with a bigger bundle compared to a smaller bundle always containing it as a strict subset. But people on a jury might come up with rather diverse ideas that don’t fit into an obvious ranking.
How about this: In addition to using the scale procedure described in this essay, a jury dealing with a concrete case can also discuss alternative ways and vote with an
approval ballot for these. Any alternative proposal that gains approval by a majority of jury members is then submitted to the defendant together with the scale procedure’s result. The defendant chooses.
And with that addition we will also have more feedback data and a really self-critical system.