Thread by @bettercount_us

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty This is an interesting case for talking about voting strategy, because the political zealotry is turned up to 11.

You think I’m talking about Brexit zealotry? Nah. Voting theory.
1/13

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty Sorry, don’t mean to be glib. I just don’t want people to assume they know what I’m going to say. I’ll be on all sides of the Twitter Voting Theory Wars here.
2/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty First: Gibbard-Satterthwaite proves that any voting method has strategy. But strategy≠insincerity. In approval voting, you set threshold strategically, but then sincerely approve based on that threshold.
3/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty Is rational approval strategy always bullet voting? Of course not. If your honest favorite has no reasonable chance of winning, the only rational thing to do is approve more than 1.
4/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty It’s trivial to show that if a majority Condorcet winner exists, and is mutually known—which is often true in practice—rational approval strategy elects them. No bullet voting problem.
5/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty But it’s not always that simple. There’s the chicken dilemma (CD), where two factions must cooperate to beat a common enemy, but the less-cooperative faction gets the relative advantage.
6/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty CD, combined with optimistic bias (“everybody secretly kinda likes my favorite”), can lead to excess bullet voting (that is, bad outcomes) in approval.
7/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty In the case of Brexit indicative votes, both CD and optimistic bias are at max. No faction wants to be the sucker. Also, vote is not binding, so signaling strategies dominate instrumental ones.
8/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty So Brexit is actually just about the worst case for bullet voting under approval. And even here, approval is MUCH BETTER than mandatory choose-one would be. But what’s the ideal?
9/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty AV/RCV/IRV (grr, names) would avoid the chicken dilemma, but replace it with center squeeze. Compromise options would be eliminated first, leading to a dangerous choice between extremes.
10/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty Score voting would risk the same strategic problems as approval. Condorcet would risk settling on a choice nobody took seriously, just because nobody hated it.
11/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty So I honestly think that something like 3-2-1 (electowiki.org/wiki/3-2-1_vot…) or STAR (en.wikipedia.org/wiki/STAR_voti…) would be ideal here.
12/

@TobiasNowacki @laderafrutal @benwansell @simonjhix @chrishanretty Of course, I can’t imagine Parliament doing a good job voting on how to vote. My criticisms of both RCV and Condorcet would be seen as advantages by some sides, and… meltdown. Oh well.
13/end