[EM] U/P voting: plausible heuristics

[Jameson Quinn]

Here are a few plausible heuristics you could use to vote under U/P:

H1: Upvote 25% of the candidates, rounded up. Downvote 50% of them, rounded
down.

H2 base: Call your preferred candidate X, and the two candidates you'd
expect to be plurality leaders aside from your preferred candidate, Y and
Z, where Y is the "lesser evil" you prefer over Z.

H2a: If you think X is "between" Y and Z (that is, if you expect that over
half of the combined Y and Z voters would put X as second choice among
those three), then upvote X, and downvote Y and Z and anybody you like less
than either of them. Otherwise, upvote X, and downvote Z and anybody you
like less than them.

H2b: If you think that there are more Z>X>Y voters than Z>Y>X ones, then
upvote X, and downvote Y and Z and anybody you like less than either of
them. Otherwise, upvote X, and downvote Z and anybody you like less than
them.

H2c: Upvote X, and downvote Z and anybody you like less than them.

H2d: Upvote X. Downvote Z and anybody you'd like less than a 50/50 lottery
between X and Z.

H2(a-d)+x: Use one of H2a-d. Then, if you expect X to be upvoted by less
than 20% of voters, and you're not downvoting Y, also upvote Y and anybody
you like more than them.

Any of the above heuristics will do pretty well. H2c is still subject to
center squeeze in a scenario like the following:
35: A>B>>C
25: B>A>>C
40: C>>B>A

But H2a or b would handle that, and H2d might handle it if you change over
10% of the second faction to B>>A>C.

Although I've described these heuristics in complicated terms, I think that
the intuition for each is clear and obvious enough that some nonnegligible
fraction of voters would use the heuristic without coaching.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20160919/c3358136/attachment.htm>