Talk:Asset voting

From Electowiki
Jump to navigation Jump to search

Asset voting or Delegated voting[edit source]

It's interesting seeing the history of the "Asset voting" page on Wikipedia. It appears as though this is the chronology:

  1. 2008-2009: w:Asset voting is created and nominated for deletion three times (one, two, three) on Wikipedia. The outcome of the first was "keep", but the second and third were both "delete".
  2. December 2009: just before conclusion of the third nomination, two things happen:
    1. w:User:Homunq merged the article into the w:Proxy voting page (see the "Asset voting" section of the 2009 "Proxy voting" Wikipedia article).
    2. an anonymous editor creates Asset voting on Electowiki.
  3. January 2010: w:User:Robert_Philip_Ashcroft deletes the "Asset voting" section, calling it "linkspam"
  4. May 2010: w:User:Homunq fixes the w:Asset voting redirect to point to the "Delegated voting" section of the Wikipedia "Proxy voting" article (where it points today).

It would seem that "w:Delegated voting" is how it is referred to on Wikipedia. That seems like a perfectly cromulent name to me.

I discovered all of this looking when I was replying to a thread on reddit. As I'm about to say there, I think this method is really interesting as a way of dealing with crowded primaries. However, I'd like to use preferred terminology to refer to it, and it appears as though "Delegated voting" is winning out over "Asset voting" as the preferred term. -- RobLa (talk) 04:38, 26 February 2019 (UTC)

"Delegated voting" has lots of meanings, though, including Liquid Democracy, SODA, PLACE, etc. Sounds like Homunq moved it there out of desperation, not because anyone actually calls it that.
Also it's not clear to me why Asset doesn't suffer from center-squeeze effect, since it seems that center candidates with low numbers of votes will be compelled to give them to other candidates nearby. — Psephomancy (talk) 02:49, 2 March 2019 (UTC)

Smith Set[edit source]

Asset always picks a winner or winner set that is in the Smith Set of negotiators' preferences if the negotiators are given enough time to negotiate

I don't see how this could be true, since it's a single-mark ballot and suffers from vote-splitting/center-squeeze. I find it hard to believe that a candidate who received zero votes would win under Asset, even if they are the Condorcet winner and sole member of the Smith set. — Psephomancy (talk) 17:40, 15 December 2019 (UTC)

There are two Smith Sets to consider: the Smith Set of voter preferences (which is what Psephomancy addressed above) and the Smith Set of negotiator preferences (whoever the negotiators can give more votes to than any other winner or winner set.) To explain, if there are four candidates, and the voters' ranked preferences of them are:
  • 49 Liberal>Moderate>Conservative>Kim Kardashian
  • 10 Moderate
  • 41 Conservative>Moderate>Liberal>Kim Kardashian
The Moderate is the Condorcet winner of voter preferences here. But suppose the voters submit a single-mark ballot, and now the votes are: 49 Liberal, 10 Moderate, 41 Conservative, 0 KK. Let's say the negotiators are corrupt and got paid off to elect KK; then the negotiator preferences now are:
  • 49 KK>L>M>C
  • 10 KK>M
  • 41 KK>C>M>L
So the Condorcet winner, and sole Smith Set member, for negotiator preferences is KK. - User:BetterVotingAdvocacy
Ok, so ignoring the fact that the negotiators' preferences may be completely different from the voters' preferences, if one of the negotiators is a Condorcet winner (among the negotiators), but has zero first-preference votes to barter with, how, realistically, are they going to be selected by the other negotiators as the winner? — Psephomancy (talk) 23:01, 15 December 2019 (UTC)
Suppose at every point in the negotiations, the negotiators are keeping track of the candidate (candidates equalling the number of seats, in the multiwinner case) with the most votes committed to them, and consider such a candidate "the one to beat", because this is the one who will win if the negotiators do nothing. Then, all it takes for Asset to elect a CW with no votes is for the other negotiators to keep looking for candidates who they a) prefer over "the one to beat" and who b) can get more votes to be committed to them. As an example:
  • 26 AB
  • 26 CB
  • 48 D
It's clear that even without any 1st choice support, if no negotiator preferences change here, eventually A and C's votes will go to B, because there's no other way to beat D, and because once B becomes the one to beat, there's nobody that can get more votes committed to them and is preferred over B by those committing the votes. If, let's say, A and C made no move to elect B (even despite having a weak preference for B), then we could say their functional preferences (their mental preferences modified to fit their negotiating behavior) were
  • 26 A>B=D
  • 26 C>B=D
in which case B isn't the CW or in the Smith Set of negotiator preferences.
I did add to my claim that the negotiators must be honest in their negotiating moves; in other words, they can't promise to give their votes to someone at the end of negotiations and then not do so, since that breaks this mechanism.
Also, it's worth pointing out that the more candidates there are, the more negotiation is required to find the Smith Set, since the negotiators are in essence doing pairwise comparisons. Even if given enough time to theoretically reach the optimal negotiating outcome, they may simply stop after a while when they've reached a satisfactory enough outcome, which is conceptually similar to an optimal PR algorithm getting a very good bit not best winner set because of long computations. Note that this still would be an outcome in the Smith Set, since the negotiators' functional preference is to equally rank the outcomes they would prefer negotiating for with the outcome they reached, or said differently, they're equally ranking the candidates they prefer over the ones in the final winner set with those ones in the final winner set. Some of the negotiators might even rank the final outcome as being better than what they earlier preferred, perhaps because they really want to finish the negotiations quickly.
Here is a visualization of Asset done algorithmically if it helps support my claim: https://www.reddit.com/r/EndFPTP/comments/eac87u/demonstrating_condorcet_pairwise_counting_with_an/ BetterVotingAdvocacy (talk) 14:53, 16 December 2019 (UTC)
So then it's just an oligarchy where the opinions of the negotiators are the only things that matter, and the votes given to them by the populace are irrelevant? — Psephomancy (talk) 16:56, 16 December 2019 (UTC)
Probably, though this view must be balanced by the fact that if Asset is used as a PR method with a hundred or more seats with the entire country voting as one district, and the candidates are forced to be elected whenever they receive a Droop quota (and perhaps the voters have the option of going with a normal party list rather than picking an individual candidate, if they want to), then the voters will have much more of a say in who wins and who doesn't. Also, suppose that you did Asset but with ranked or rated ballots, and where the negotiators' only role is choosing who within the voter-preferred Smith Set (either for a single-winner method or multi-winner method) wins; that would be also be more democratic.
Anyways, do you still find the claim that Asset elects from the negotiators' Smith Set dubious? The point of that claim is to demonstrate that Asset, as a concept, is really just a reframing of the Condorcet principle (which can also be applied to PR), and when applied appropriately (i.e. if a small group election is held where the voters themselves negotiate, so that they actually elect the voter-preferred Condorcet winner, or at least begin to figure out who that is, then they start to understand that the Condorcet winner is like an equilibrium outcome that's iteratively reached if everyone is maximally strategic) or used only as a concept (as an algorithm, or for demonstrating how Condorcet pairwise counting works, for example) can help explain that principle. In other words, the simplicity of Asset can be a segue to understanding voting theory. Consider that understanding Asset PR is a lot easier than understanding CPO-STV or Schulze STV, even though the concept (elect from the Smith Set of winner sets of some group of voters or negotiators) is almost the exact same; once someone understands those finer details of how Asset can get proportional results, they can then use that insight to understand Condorcet PR in general, and then perhaps use that to better understand cardinal PR as well. BetterVotingAdvocacy (talk) 17:20, 16 December 2019 (UTC)

Content from deleted Wikipedia article[edit source]

Content added by Psephomancy in this edit (6650):

Asset voting or candidate proxy is a voting system which provides proportional representation by allowing all candidates to negotiate the outcome, using the votes cast for them as proxies in subsequent rounds. It is unconventional in the sense that not only voters, but also candidates, can directly affect the outcome.

Candidates may use, distribute, or redistribute votes they received in the election, negotiating with each other to put together a coalition of enough votes to win office. This was proposed in a pamphlet published by Charles Dodgson (Lewis Carroll) in 1884.[1] Though famous for his Alice stories, Dodgson was also a mathematician who pioneered voting theory.[2] The mathematical economist, Duncan Black, analysed his work and described the process: candidates receiving votes may treat them as "if they were their own private property" for purposes of creating proportional representation.[3] The term "asset voting" for this was coined by Warren D. Smith of the Center for Range Voting in 2004[4]. Mike Ossipoff (in 2000) and Forrest Simmons (in 2002) earlier referred to a similar idea as "candidate proxy".[5][6]

This voting system may be used as a model to coordinate multiple heterogenous autonomous systems which have a mission with multiple objectives. The autonomous systems are allowed to influence the weights assigned to each objective. Results from a simulation demonstrated a level of overall mission success comparable to that obtained in an ideal centralised mechanism.[7]

References

  1. Charles Lutwidge Dodgson, Francine F. Abeles (2001). The Political Pamphlets and Letters of Charles Lutwidge Dodgson and Related Pieces. Charlottesville: Univ. Press of Virginia. ISBN 0930326148. 
  2. Duncan Black, Iain McLean, Alistair McMillan, Burt L. Monroe, Charles Lutwidge Dodgson (1996). A Mathematical Approach to Proportional Representation. Boston: Kluwer. ISBN 0792396200. 
  3. Template:Citation
  4. Warren D. Smith. "Asset voting – an interesting and very simple multiwinner voting system". The Center for Range Voting. Retrieved 29 May 2008. 
  5. Yahoo! Groups
  6. "Candidate Proxy Methods". Election Methods mailing list. 2 Dec 2002. 
  7. Template:Cite conference

Content from deleted Wikipedia article -- unsigned comment by Psephomancy

Can we remove the old Wikipedia content above from the Talk: page? We can add it in an appropriate spot on this wiki if we find a place that makes sense and if it belongs on Electowiki, but I made a rudimentary attempt to find the original content in the history over on enwiki, and wasn't able to find it. We should make sure we keep track of the provenance of the text that gets added to this wiki, and this takes a big block of this Talk: page, but isn't really part of a conversation. -- RobLa (talk) 01:44, 14 January 2020 (UTC)
It's from https://en.wikipedia.org/w/index.php?title=Asset_voting&action=history which you can't see unless you're a Wikipedia admin. I posted it here in case there's some content that can be rescued from the deleted article. If it's all redundant, then delete it from here, too. — Psephomancy (talk) 02:05, 14 January 2020 (UTC)
Thanks for the prompt response, Psephomancy. I appreciate your taking the time to pull things that were aggressively deleted from English Wikipedia (enwiki). I'm an enwiki admin too, so I can at least see the old revisions. What is the oldid and/or timestamp of the revision you pulled this from? We might want to cite the Wayback Machine version of the article as the public source. Ultimately, someone should probably do a plagiarism-free rewrite of the parts we want to keep. -- RobLa (talk) 19:09, 14 January 2020 (UTC)

Asset voting discussion over on reddit[edit source]

User:Psephomancy started a very interesting discussion about asset voting over on the EndFPTP subreddit. The discussion: https://www.reddit.com/r/EndFPTP/comments/eniyfl/i_dont_understand_why_anyone_likes_asset_voting/

I haven't had the chance to really read through it, and wrap my brain around the pros and the cons, but it looks like it prompted many great explanations of asset voting that could be used to improve this article. -- RobLa (talk) 20:01, 14 January 2020 (UTC)