![[Home]](http://www.bayleshanks.com/cartoonbayle.png)
"The group recommended reforms that include increasing the number of seats on the court, term limits and having smaller panels of justices hear cases as opposed to all nine presiding over each one, a practice that's employed by the federal appellate courts that the task force says would "disrupt static voting blocks" on the bench...The task force called for adopting screening committees to recommend candidates for judicial nominations at all levels of the judiciary to thoroughly vet potential judges. And it recommended changes to how the court selects cases, its ethical standards for justices and its lack of transparency." -- https://thehill-com.cdn.ampproject.org/v/s/thehill.com/regulation/562028-watchdog-groups-calls-for-supreme-court-reforms?amp=&_gsa=1&_js_v=a6&usqp=mq331AQIKAGwASCAAgM%3D#amp_tf=From%20%251%24s&aoh=16257711930478&csi=1&referrer=https%3A%2F%2Fwww.google.com&share=https%3A%2F%2Fthehill.com%2Fregulation%2F562028-watchdog-groups-calls-for-supreme-court-reforms the actual report is: https://www.pogo.org/report/2021/07/above-the-fray-changing-the-stakes-of-supreme-court-selection-and-enhancing-legitimacy/
---
some potential simplifications:
---
in addition to the 60%/50% threshold difference for increasing/decreasing org power, you could require multiple chambers to vote on increasing but only one to vote on decreasing
---
you could have equity-weighted voting in one chamber (e.g. the board) but not in another (eg the forum)
---
"How do you avoid perverse incentives, runaway complexity, endless bikeshedding, or stagnation due to "vetocracy" like what exists with California housing? How do you prevent the seemingly natural formation of an oligarchy? So far I don't think democracy has ever existed except at tribal scale (below Dunbar's Number). All former and current attempts are oligarchies with a degree of democratic veto power or a democratic facade. I think this problem is closely isomorphic or maybe even identical to the open problem of efficient and secure fully decentralized computing and global consensus in distributed systems without hidden centralization or brute force approaches like Bitcoin proof of work. (... and Bitcoin PoW? is in reality an oligarchy if you look at the largest pools ...) " -- [1]
---
mb a way to simplify triscore voting:
give up on exact proportionality, and instead of deweighting ballots that 'got their way' with a previous candidate, just eliminate them. This might not be good for electing slates of large numbers of seats, but we are using it to just elect slates of 3 seats.
So something like this:
That is perhaps still too complicated. Here's something simpler:
An issue with these systems is that it rewards strategic voting where a party computes how many votes it needs to elect a candidate and then splits its members into a group of that size, and another group, and tells the other group not to vote for the candidate so that their votes are still available for someone else.
I guess you could call this a semiproportional system. It doesn't exactly try to match factional proportions, but it does avoid a situation where the most numerous faction controls all the seats at once. It also probably tends towards centrism; a candidate who is no one's first choice but everyone's second choice should be competitive with extremists who excite their base, because although the extremist will get more +1s, the inoffensive candidate will get less -1s. In fact, in a sense it's better to win by getting less -1s than it is to win by getting more +1s, because when you win, those ballots who gave you a +1 are set aside, whereas those who merely did not give you a -1 remain to affect the selection of other seats.
What if every member of every faction just votes +1 for their preferred candidate and -1 for everyone else? The first winner is the largest faction's candidate, the second winner is the second-largest faction's candidate, and the third winner is the third-largest faction's candidate. So that's not so bad.
A way this could be upset is if, for example, the largest faction's candidate is offensive to all other factions but other factions' candidates are not offensive to each other. In this case, if the largest faction is small enough, it's possible their candidate would not be elected at all.
Consider if the members of the largest faction apply restraint and vote -1 for other candidates but 0 for their own candidate. If they are sufficently large, they will still win a seat, and they will also have some influence over the 3rd seat. So, in these situations, there is an unpleasant amount of potential for strategic voting.
I guess the main property that i'd like to prove here would be:
Say a faction has 1/3 support. If its supporters support it, we'll define that to mean that its candidate should get at least 1 score more from each of its supporters than those supporters give to any other candidate. So, (i think) scorewise that's equivalent to saying that its supporters give it +1 and everyone else +0 (which is conservative in that it's worse for this faction than if they gave everyone else -1 and gave it 0). So, normalizing votes by the numbers of voters, now this candidate has a score of +1/3. The worst case is if everyone outside of the faction gives its candidate -1, so 1/3 + -1*2/3 = -1/3 is the final score of this faction's candidate. So, if all of the other 2/3s of the voters just vote 0 on each of 3 other candidates, then those candidates beat the 1/3 faction.
What if the 1/3 faction votes -1 for everyone else and +1 for their own candidate? Now, again, their candidate has a score of -1/3, but so does everyone else (before the other people's votes). In order to pull ahead of the 1/3 faction's candidate, other candidates must receive some +1s. Say one other candidate receives three +1s, another one gets two +1s, and another one gets one +1, and none of those voters give +1 to more than one candidate, and say that none of them get any -1s aside from those coming from members of the 1/3 faction. Now all three of those other candidates are elected.
So, i think that property doesn't hold. So maybe the method needs some work.
otoh, i think it works if the faction has >40%:
if the faction has exactly 40% (2/5ths), then it gives its candidate +1s which give it a score of +2/5ths, and the other 60% gives the faction -1s for a score of -3/5ths, for a total score of -1/5. The faction gives -1s to the other candidates for a score of -2/5ths. The other 60% splits into 3 20% subfactions, each of which give +1 to one of their candidates, and 0s to their other candidates. Each of the 60% faction's three candidates therefore gets +1/5 from these +1s, for a total score of -1/5 (=-2/5s + +1/5; recall that the -2/5s are from the 40% faction's -1s). So now we have 4 candidates each with a score of -1/5. If the 40% faction had more than 40% it would be able to do better, and guarantee that its candidate were one of the top 3 candidates.
So, the cost of being simpler than a fully proportional method is that a faction needs >40% to guarantee one of three seats, instead of >=1/3.
or how about this, which is even simpler except with the addition of the excess transferrable vote:
This is kinda like https://en.wikipedia.org/wiki/Sequential_proportional_approval_voting except (a) instead of D'Hondt method (Jefferson method), we have elimination, which is sorta even further along the scale of D'Hondt/Jefferson -> Webster/Sainte-Laguë, and (b) we have the excess transferrable vote mechanism to remove the incentive not to approve of someone you think will win anyways without your vote.
This is simpler to explain, remember and compute than sequential proportional approval voting because (a) you don't have to justify or remember the formula for the divisor, (b) there is less computational work because each ballot is eliminated if any of their candidates won, rather than having a graded weight that depends on how many of their candidates won.
---
Draft email to voting methods list:
Sequential semiproportional approval voting with elimination and transferrable excess vote
I am looking for multiwinner voting methods that (a) are simple to justify, remember, and compute, (b) are not party-based, (c) don't have the spoiler effect, and permit inoffensive 'compromise candidates' who are everyone's second choice but no one's first choice, to win, (d) don't just hand all of the seats to the dominant faction.
Has the following method been named and studied?
I'm imagining it might work well for small numbers of seats (where there are not enough seats to fully realize the unproportionality of flat-out eliminating ballots which get their way, while still having the property of ensuring that a single dominant faction doesn't take all the seats).
This is kinda like https://en.wikipedia.org/wiki/Sequential_proportional_approval_voting except (a) instead of D'Hondt/Jefferson method, we have elimination, which is sorta even further along the scale of D'Hondt/Jefferson -> Webster/Sainte-Laguë, and (b) we have the excess transferrable vote mechanism to reduce the incentive not to vote for someone you think will win anyways without your vote.
This method is simpler to justify, remember, and compute than sequential proportional approval voting because (a) you don't have to justify or remember the formula for the divisor, (b) there is less computational work because each ballot is eliminated if any of their candidates won, rather than having a graded weight that depends on how many of their candidates won.
---
yknow what, EXCESS doesn't help with Burr's dilemma; if you have two candidates who are neck-and-neck and in the same party, it is still unstrategic for all of their supporters to vote for both of them rather than to allocate votes between them. So just eliminate it:
---
Draft email to voting methods list:
Multiwinner approval voting via elimination
Has the following method been studied yet?
This is kinda like https://en.wikipedia.org/wiki/Sequential_proportional_approval_voting except that instead of D'Hondt/Jefferson method, we have elimination, which is sorta even further along the axis of D'Hondt/Jefferson -> Webster/Sainte-Laguë (this method can be thought of as attempting to give as many factions as possible one representative on the committee, rather than attempting to apportion seats of the committee proportional to the strengths of the factions -- although with strategic voting (assigning different subsets of voters to vote for different candidates of one party) a faction could gain seats in proportion to their strength anyway).
I am looking for multiwinner voting methods that (a) are simple to justify, remember, and compute, (b) don't just hand all of the seats to the dominant faction, (c) don't have the spoiler effect, and permit centrist/inoffensive 'compromise candidates' who are everyone's second choice but no one's first choice, to win, (d) are not party-based. This method is simpler to justify, remember, and compute than sequential proportional approval voting because (a) you don't have to justify or remember the formula for the divisor, (b) there is less computational work because each ballot is eliminated if any of its chosen candidates win, rather than having a graded weight that depends on how many chosen candidates won. This method might be good when diversity is required but proportionality is not; such as for very small committees (say, a 3-person committee), where all you really need is to prevent the largest faction from taking all of the seats.
---
Because of the elimination, it may not be very proportional for large numbers of seats; without strategic voting, all of the supporters of a party would get their ballots eliminated after winning a single seat, with no chance to take additional seats in proportion to their strength. But for small committees (say, a 3-person committee), this method would at least prevent the largest faction from taking all of the seats on the committee (unless the largest faction is much much stronger than all of the others). So this method might be good when diversity is required but proportionality is not. In a sense, elimination is at the extreme end of the D'Hondt/Jefferson -> Webster/Sainte-Laguë axis, maximizing diversity -- this method can be thought of as attempting to give as many factions as possible one representative on the committee, rather than attempting to apportion seats of the committee proportional to the strengths of the factions -- although with strategic voting (assigning different subsets of voters to vote for different candidates of one party), a disciplined faction can 'appear as' many factions and achieve a number of seats proportional to their strength.
---
i dunno, do i even think that the loss of proportionality (from elimination rather than reweighting) is worth the simplicity here? Well, it might be, for the case of the Chairs (which is what i'm thinking of using it for) -- because there you do want diversity more than you want proportionality.
in the legislatures we are using asset voting (for the Board), which is proportional, and transitive proxy voting (for the forum), which has elements of direct democracy but insofar as it is like representative democracy, is similar to asset voting and hence proportional (i think). So, actually, multiwinner approval voting via elimination might be okay for the chairs. In fact, due to the individual powers of the chairs, and their negative function, diversity may actually be better than proportionality in this one case (although note that with strategic voting, proportionality is the result anyways).
of course, i would prefer less strategic voting, but at least here the type of strategic voting (The Excess Method: A Multiwinner Approval Voting Procedure to Allocate Wasted Votes calls it 'subset manipulation') encouraged by 'Multiwinner approval voting via elimination' seems kind of desirable; it encourages a large party to give various subsets of its voters their own 'local' representative, and forces parties to try and give name recognition to many candidates rather than just to their leaders.
yeah, so i think i like it, for the chairs.
---
i guess two things not to like about the above system:
so i guess i see six ways to go here; from simplest to most complex:
another variant of sequential approval voting is that each seat could be chosen by 321 voting and then those ballots who approved of the winning candidate are eliminated
---
if we use this system, it's even simpler than the proportional voting required for the Antenna project.
that's okay; perhaps a governance system needs to be executable by humans using 'stone age' technology, whereas Antenna is in the information age (can use computers).
---
in a trial:
---
anyone who said something that was recorded in debate should be able to retroactively 'revoke' what they said (that is, to assert that they may no longer agree with it) at any time in the future, even after they lose whatever office or role or privilages allowed them to participate in debate. The idea is similar to a cryptographic revocation key. A cryptographic revocation key can be separate from the private key needed to do normal stuff, which is why here, similarly, the speaker can't actually amend or comment on what they said, they can just revoke it. Revocation only applies to speeches, and doesn't have any actual effect; it just provides notice to the reader that the speaker no longer agrees with something that they had said. In online systems actually using public key cryptography, this has the additional use that if someone steals someone else's private key, the victim can revoke the key going forward, and in addition if the thief had used the key to impersonate them, although the victim can't undo the decisions taken, at least they can provide notice to the world that the things the impersonator said wasn't said by them. Should an online system actually stop showing speech which was revoked, or should it show it along with a message that they took it back? I'm leaning towards the latter, since this is a formal "legislative" platform. Maybe you can also revoke actions, with no effect, except again an indicator. The word 'revoke' makes sense because of its cryptographic roots, but perhaps 'regret' would make more sense in this context. 'Unsay' doesn't make sense because the effects aren't actually being undone.
---
