Talk:International logo contest/Vote on voting method

From Meta, a Wikimedia project coordination wiki

Where can I vote for the method to vote on the voting method?

And how exactly do we vote for the voting method by which vote for the method by which we choose an international logo? - good question.-Stevertigo 21:31, 8 Sep 2003 (UTC)

Please do take the opportunity to vote for a mind numbingly complex process that isn't used in any real world election.

One is tempted to rise to the challenge and design just such a process.

But the real search is for a simple fair procedure.

-- Eloquence (I think)

It is an illusion, and does not really exist. If voting is to be used for anything, we must understand the limitations of voting, and how to deal with those limitations. This includes knowing when voting is not the appropriate. Simple and fair can be mutually exclusive, given that "simple" is often driven by expediency.

First stage logo voting has been useful because we had two more candidates than are running for California governor. The purpose was to eliminate candidates that had absolutely no chance of winning. Setting the short list at exactly 10 candidates is arbitrary. If 10th and 11th place are separated by a single vote, it should not have serious consequences. If that is the case then the suggestions of vote cheating become meaningless. Making vote cheating meaningless is a lot easier on everybody than trying to develop elaborate systems to catch the cheaters. A final solution that results in a 51%-49% vote is seldom democratic. Eclecticology 18:03, 5 Sep 2003 (UTC)

How about a 5 to 4 vote? ^_^ -- Toby Bartels 01:30, 6 Sep 2003 (UTC)

Just a question about Condorcet voting...

Would this be the same as:

You have 10 points to allocate. You can allocate up to 4 points for any candidate. Once your points are used, you cannot vote any more.
EG: Four candidates. You can rank them 4,3,2,1 (pure and simple ranking) 4,4,2,0 (tie between 1 and 2, nothing for 4th) 2,3,2,3 (tie between 2 and 4, and 1 and 3) 4,0,0,0 (vote for 1, forfeit rest of points)
Most points at the end of voting wins.

This would allow a little flexibility with the voting, while keeping some restraint. You are forced to pick more than one candidate, but you don't have to give them a high ranking. If the voter chooses to forfeit the rest of their points, they can do so, but not to the advantage of one candidate.

Any thoughts on this as a voting method? (Not just for logo votes, but as a method in general).

No, this is not the same Condorcet. At first glance it seems most similar to en:Borda count, and suffers from the same problems, and probably some others, since voters are allowed to explicitly assign points. Consider the following outcome:
  • 3 people vote: 4,3,0,2 (Condorcet vote would be A>B>D>C)
  • 2 people vote: 1,3,4,0 (Condorcet vote would be C>B>A>D)
(There is never any reason not to give 0 points to at least one candidate.) Under your system the total counts are: A: 14, B: 15, C: 8, D: 6. The winner is B. However, the Condorcet winner is A (3 out of 5 preferred A to everyone else). In effect, the A>B voters diluted their vote by giving 3 points to B. My guess is that in practice most people would end up giving 4 points to everyone they would vote for under approval voting, and none to everyone else, thus mostly reducing it to approval voting.
Most voting systems have been thought of, and all of them have problems. Condorcet is generally considered to have the fewest. -- AdamRaizen

Taw, would you care to explain? I don't now what the correct term for "playoff voting" is but I know the method is regularily used irl. Neither do I understand your objects but I'm open to change my mind if you can explain it in a better way than "you don't know what you are talking about". BL 16:55, 7 Sep 2003 (UTC)

It's NOT being used in any real life situation I know.

The votings are not independent. Example. Let's say there are 4 candidates. So first round is:

  • 1 vs 2
  • 3 vs 4

But the voting isn't really about 1 vs 2, the options are:

  • whoever wins election 1 vs (3 vs 4)
  • whoever wins election 2 vs (3 vs 4)

So people who want 4 to win are going to vote not to their second preference - 1st or 2nd, but to one who has less chances of winning with 4 (if they think 4 is going to win against 3 and has chances of winning either against 1 or against 2). If they think 4 can win against neither, and they prefer 3 over 1 or 2, they'll going to vote for one who has smaller chances of winning against 3. If they prefer 1 over 3, but 1 has no chances of winning against, they'll vote on 2 if they prefer it over 3, no matter whether they prefer it over 1. Etc.

And it's not the real preferences that play role here, but strategies of other voters.

It's going to be lot of tactical voting. Probably even worse than FPTP, as people can understand FPTP easily, and in this kind of voting it's hard to guess one's optimal strategy. Taw 19:31, 7 Sep 2003 (UTC)