On May 20, 2004, at 4:46 PM, Ken Johnson wrote:

Date: Wed, 19 May 2004 23:24:29 -0700
From: Bart Ingles <[EMAIL PROTECTED]>

... In your 10 candidate, 1 issue trial, are you able to account
for why sincere CR, exaggerated CR, Condorcet, Borda, IRV, and Plurality
all yield exactly the same average across 100,000 elections? It looks
like top-two Runoff is within 0.1% of the same score.


I think it's simply the case that with 1 issue, all voters' CR profiles are precisely correlated (i.e., any two profiles differ only by a multiplicative scale factor), so all these methods become equivalent.

This doesn't make sense to me. If a voter has a random preferred stand on an issue, and a candidate has a random position on an issue, then every voter should have a different opinions of the candidates.


In my simulation I skipped the "issue" problem and designed voters to have direct utility for each choice. It was a uniform distribution, perhaps a normal distribution would have better simulated a multitude of issues combined in subtle ways.

Here's what I get for a similar test configuration: running work set 10 candidates, 100 voters, 100000 trial elections, +/- 0.000000 confusion error, 15 systems
NameAvg. Happiness
(higher better)
Reliability Std.
(lower better)
Consensus Std.
(lower better)
Max Happiness0.08889757460.03385158160.570981684
One Vote0.04655698350.05225507930.590699266
Instant Runoff Vote0.06917087060.04491003160.58041986
Acceptance Vote0.0769196220.04120600980.570201211
Rated Vote, raw0.08889757460.03385158160.570981684
Rated Vote, equal sum0.08743813140.03486418770.571236063
Rated Vote, maximized0.08852788830.03412523130.57094033
Rated Vote, 1..num choices0.08835597020.03423274820.571054417
Rated Vote, 1..100.08835597020.03423274820.571054417
Ranked Vote0.08484817880.03651659170.571332999
Ranked Vote, no neg pref0.08145755740.03880522550.576437401
Condorcet0.0799608580.03878132960.572688295
IRNR0.08330615680.03727257370.571565448
IRNR, positive shifted0.08465623530.03654333690.561307907
Random0.0002039741060.05803433370.573728578
NameAvg. Happiness
(higher better)
Reliability Std.
(lower better)
Consensus Std.
(lower better)


I suppose it's similar overall, but I think I see more variation in the primary statistic "Average Happiness", which makes me wonder if there's some artifact in your simulator.

Brian Olson
http://bolson.org/

Reply via email to