Re: ELECTION: problems/solutions

From: hal@finney.org
Date: Mon Nov 13 2000 - 18:29:48 MST


Damien B writes:
> Posit an electronic poll with cumulative vote tallies posted
> instantaneously to the net and other media. In a dispersed time zone nation
> like the USA (or Oz, although our least populous states are over on the
> west), this will presumably see enormous chaotic fluctuations among
> undecided voters, increasingly feverish efforts to get out the vote as the
> long day wears on, etc. Naively, this is surely a good thing (assuming that
> voting is not intrinsically wicked); in practice, maybe it would provide
> information giving some moeities more advantage than others and more
> incentive for targeted bribes/dirty tricks etc.

I like this. Combine it with Eliezer's idea to let people vote more
than once (each vote superceding their earlier ones) and you might
see some interesting effects.

Now eliminate the fixed time limit, and let the vote be settled once
there is a consensus, that is, one candidate has had the highest score,
stably, for some extended period of time.

This might give you some of the effects of Borda count type systems,
where you'd end up with a candidate who was acceptable to all even if
not the first choice of most.

Or do you think you'd see constant fluctuation? Like A beats B, B
beats C, and C beats A, so we constantly cycle through them? I wonder.
It would seem kind of stupid for people to get stuck in a cycle like that.

Hal



This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:50:21 MDT