Rkjoyce: "the people" are not a uniform blob

Continuing the discussion from Utilitarian Sum vs Monroe Selection:

Until the common people win, and then

truly develops. In fact, once it becomes the ideology contest, you know that the people have won. It is just that “the people” are not a hive mind: they have some interests that are similar and some that are different. We need a voting system that works for both contests. (I think simple score works well even in the ideological contest: centrists can win, but they do not have to.)

Again, we are not a hive mind. Even once the aristocrats lose, there will still be people who have fundamentally incompatible desires.

You are right. Replied as a new topic.

First of all, I never ever questioned the value of proportional representation. In fact I have written quite a lot about it here. There always exists a rather nebulous ‘majority’ camp, and there always exist smaller 'minority camps. Democracy is meaningless if the nebulous majority camp is not dominant in most instances, but there is great value in granting some power and ‘say’ to the minority camps as well.

There are two ways I know of to define minority camps: organizations or political parties, or possession of substantial albeit non-majority assertable electoral support. Now, significantly strong organizations and parties are always eventually co-opted and controlled by… aristocrats. So it is only reasonable to identify them by their possession of non-majority assertable electoral support. Ultimately they can only ever obtain any representation by dint of their ability to grant less votes that some number of the nebulous majority camp. The only straightforward way to enable that is simply to allow some candidates to gain office due to having been granted less votes than some other (nebulous majority camp) candidates. And this turns out to be very easy to accomplish with add-on (virtual) proportional representation.

The simplest solution is:

I also described this simple method in greater detail in a much earlier topic:

So it would in fact be quite easy to apply ‘add-on proportional representation’ to score voting. The results would be very random at first, but due to ‘paladin preservation’ would become very stable after a few election cycles. It should be obvious that no form of proportional representation can ever be ‘monotonic’ (really consistently responsive). This add-on method would not be ‘strictly proportional’ in the mathematical sense, yet would be entirely adequate in the political sense.

A method that requires such math must be more consistent and less random, otherwise it will be attacked by both math-phobes and those who understand the intent but seek better methods. This method would get destroyed by lobbyists and the parties, and worse, makes it harder to get rid of bad incumbents than it should be; one has to be strategic enough to score them low if they want them out, without taking the risk of reducing their own representation by not scoring them high enough. On top of that, a huge benefit of PR is that majority support among the people translates into majority support in the legislature; is there some way for you to show how your method would interact with ideological distributions of the people, as @Keith_Edmonds has with Sequentially Subtracted Score? It’s hard for me to tell if it will consistently give the majority enough legislators to support it on various issues, and I also don’t know if this will yield compromise as well as other cardinal PR methods.

With paladin preservation it will not be very random at all. And if some former paladin official betrays his or her constituency, she or he will never again be given enough votes to meet their tranche lower boundary, and they will be removed. The equation involved is a simple quadratic with only two parameters: the number of seats that need to be filled (which always equals the number of tranches), which is obviously known well in advance, and the number of voted granted to that strongest winner in the election. Nothing more is involved. The ‘ideological minorities’ are by definition those who have less votes to grant, and this system provides them with a voice simply because of that.

This system is certainly vastly simpler than the other convoluted ones that have been proposed.

But the results it gives are almost random, even if they may ultimately only select good candidates. I mean, how is “score your candidates 0-5, most points wins, every time a candidate you scored wins: you lose the number of points you gave them up to 5 points” with minor modifications more complicated than this? It certainly doesn’t involve anything quadratic, and is a lot more consistent and predictable in nature. Your method requires voters to be more trusting that the math checks out, and it almost seems like the kind of complexity in which maybe all the good candidates don’t quite make it into their “tranches” or something like that. For someone with no experience with voting theory, it will seem more an academic thought experiment than a carefully thought out method.

You must be kidding. With that, the portion of assertable support on each individual ballot must potentially be re-evaluated! This will have high complexity ramifications and very high information traffic. How can that be simple?

The very simple quadratic equation of add-on proportional representation can be solved by any junior high school student long after all the votes have simply been added up at the precinct level. There is absolutely no comparison, the add-on method is far, far, far simpler.

I’ll point one more problem out. Why would anyone trust a method where the results of the very first election will be almost entirely random? Who would want to make that leap, even if the results are great in the long run? Wouldn’t any activist or voter prefer something they don’t understand as well, yet which they feel confident in the results of in every election?

Those results could not really be all that random, since after all, a lot of people would have voted for them. Even if they were completely random, a very good argument can and has been made for outright random selection of officials (sortition). And changing the effect of every ballot one-by-one does not sound at all secure to me.

Even STV requires computers, no?

Asset is both simpler than your tranche method and actually proportional; just saying. Plus, it’s spoiler free.

Optional Asset (check a box to confirm transfer) may be safer from conniving politicians, while maintaining the simplicity; it’d be a mix of SNTV and Asset based on the population’s wishes. Optional Delegation paired with other methods is the ultimate, but a lot harder to obtain.

Two problems with making transfer optional:

  1. Surpluses (although you can easily address that with an exception)
  2. Ideologically similar candidates could play chicken with each other. Consider a case where A receives .9 quotas and their ideologically similar rival A’ gets .2 quotas that are nontransferrable. A would be under pressure to concede to A’, when it should be reversed.

It may be possible to somewhat counteract these by simply taking the top X winners, rather than demanding Hare Quotas be filled. If you want to go with Abd’s version though, then this does become a problem, though I wonder how intensely voters would pursue this strategy.

There are certain people who love to eat snails and climb up high rocky mountain cliffs. Perhaps these are ones who believe asset voting is a fine idea. My own perspective is that it is grotesquely convoluted and thus utterly hopeless. But perhaps that’s ‘just me’.

On the other hand, tranche based effectively proportional voting (maybe ‘TBEPV’) is very simple. Once you know the number of seats to be filled, you have a very simple formula for establishing the tranche boundaries. Then, once the number of votes received by the strongest winner is announced, you just plug that number into the simple formula, and the tranche boundaries can then be calculated by any junior high school student. Then you just install the strongest winner of each tranche, except when incumbents win by reaching the lower bound of their previously-won tranche. That’s it.

It’s not precisely proportional, but it is easily proportional enough. It’s not strictly ‘monotonic’ (faulty borrowing from set theory; I prefer ‘consistently responsive’). But presumably no party-free proportional system ever ultimately can be. My opinion is that there exists no system that is absolutely 100% spoiler free, although simple score can thoroughly disrupt kindred party lock-in.

But… ‘to each their own’, I suppose.

How is “vote for your candidate, he votes on your behalf” or “vote for your candidate, he can optionally vote on your behalf” grotesquely convoluted? Also, since it only takes points from you when you elect someone you voted for, Sequentially Subtracted Score ought to be monotonic.

2 Likes

Yeah rkjoyce is definitely a troll. No doubt about it.

1 Like

The ‘it’ referenced above is asset voting. One reason I think it’s convoluted is that (and this is all only to the best of my knowledge) its most typical versions require that every voter ensure that his or her votes total up to some specific value. That seems incredibly awkward. Another aspect of this method is that a number of ‘semi-finalist’ candidates are chosen, and these candidates then make ‘deals’ that determine who ‘gets to stay’. That seems downright weird to me. Isn’t that quite a bit like so-called ‘indirect democracy’? Am I wrong? Is that not rather convoluted?

Now, please note that I went on to briefly explain why I think it is ‘simple’. I always try to look for two kinds of absence of complexity (i.e. simplicity). They involve ‘immediate complexity’, and also ‘ramification complexity’. The game of chess is only mildly immediately complex, as one can learn to play it very quickly. However, is bears very high ramification complexity, as one can take a lifetime to learn to compete with the masters. When I consider, or claim, that something is ‘simple’ my habit is to consider both of these kinds of complication (i.e. lack of simplicity).

After years of contemplation I have concluded that tranche based effectively proportional voting tends to be relatively simple in both of these senses. Not ‘perfectly simple’, of course, but at least ‘surprisingly simple’. So this is my true opinion, which I was under the impression I had a right to.

Having done as much analytical work, and also real-world activism for as long as I have, one might suppose I would deserve a little respect. Look at all the careful analysis I have brought to this website. Perhaps it differs radically from what others have done.

Does that difference establish that I deserve to be denigrated?

Most people advocate vote-for-one Asset. With something like Optional Asset, you get the simplicity of SNTV with additional choice. It would be great if you could share the reactions you’ve gotten on tranche PR from regular voters.

Thanks for your question, AssetVotingAdvocacy. I like ‘tranche PR’! I am only just getting started with my local advocacy. As is of my nature, please let me bore you with a bit more useless detail.

Many of the things we call ‘systems’ are really comprised of a remarkably separated ‘frontend’ and a ‘backend’. Your radio has a ‘tuner’ frontend and an ‘audio’ backend. Voting systems do too. The key concept pertaining to the frontend is ‘expressiveness’, and the concept pertaining to the backend is ‘responsiveness’. Expressiveness is meaningless without responsiveness, and responsiveness is meaningless without expressiveness. If I gift you a lottery ticket you have vastly more expressiveness if I let you write in some number to be guessed. But still that ticket is of no more value to you if I insist on writing in that number myself. The random lottery system simply is not responsive to you in that regard.

Every election system has a frontend and a backend; these are the ‘balloting system’ and the ‘tabulation system’ respectively. Some example balloting systems are, score voting, ranked voting, approval voting, single-selection voting, and so forth. Suppose you have a ranked system ballot. The tabulation system (for the complete voting system) could be RCV/IRV. But the tabulation system could otherwise be ‘simple ranked voting’. With that, the first rank-place could be marked ‘10’, the next rank-place ‘9’, all the way down to ‘0’. With this type of tabulation you just add up the numbers that correlate with the candidates. It is just like score voting, but far less expressive, and so far less useful for the voters. Note: same frontend, different backend.

Now here is the really cool thing about (add-on) tranch PR: It will work just fine with just about any tabulation system! The balloting could be score, approval, single-selection, maybe even ranked for all I know. You can just add on the tranche PR backend and you are good to go! So now, you can righteously proclaim that any election system (including of course score) can easily become a proportional representation system. Cool!

My local election activism activity is only a few weeks old, and nobody’s mentioned the PR issue so far. We shall see…

Is there any way to simplify and make more consistent the tranche equation? It seems you could make it something linear (my math is bad), like say… this is a bit jumbled, but maybe you can do something like, if there are 5 winners, take the highest scoring candidate’s score, then take 1/5th of that, 2/5th of that, etc. up to 5/5th. Now, the candidates who come closest to each of these 5 scores, including the highest scoring candidate, get elected. This still ought to (maybe?) elect a majority of majority-preferred legislators, while offering some kind of minority representation. I’m sure the idea needs tweaking to work properly, though. Really, you want a system where the rationale can be easily understood while being precinct summable. I think SNTV is already not too far off from proportional, but if you want to use scored ballots or other better kinds, then maybe instead it could be take the highest 3 scoring candidates out of 5 seats, then for the last 2, find the two candidates closest to the 20%/40% of the winner’s score mark. Or you can base it off percentile of candidates i.e. the candidate who comes closest to the bottom 40th/20th percentile.