[EM] Simulation results (Approval, utility, Schulze "efficiency)

[Richard Moore]

Kevin Venzke wrote:

"Message: 3
"Date: Thu, 4 Mar 2004 07:44:50 +0100 (CET)
"From: =?iso-8859-1?q?Kevin=20Venzke?= <stepjak at yahoo.fr>
"To: election-methods at electorama.com
"Subject: [EM] Simulation results (Approval, utility, Schulze "efficiency)
"
"Hi all,
"
"I've been sitting on this for a while, but I'm thinking I'll post it 
"now:
"
"
"Here are some results from the simulation I recently wrote about:

<snip>

Kevin,

Can your sims compare approval results for cases where the ranked 
voting produces cycles, vs. cases where there are no cycles?

"It's surely a fluke that "Two Evils" outperforms "Zero-Info" here.
"I have to doubt that random information could be better than none at
"all.

It comes as no surprise that zero-info performed about as well as 
Schulze in terms of average utility. Also in average utility terms, 
"two evils" performed worse than zero-info, also no surprise. The 
results you refer to simply mean that the former will pick the Schulze 
winner more often than the latter, but when either one misses the 
Schulze winner, zero-info strategy is more likely to result in an 
average utility improvement over Schulze than "two evils" is.

I expect a majority of voters to favor the Schulze winner over at 
least one of the "two evils", except on occasion when there are cycles 
of 4 (or larger). But I also expect "two evils" to cause fewer voters 
to approve the utility maximizer (and fewer voters to disapprove the 
utility minimizer) than zero info would, when the utility maximizer is 
not the Schulze winner.

I hope that makes sense.

  -- Richard