[EM] Why I Think Sincere Cycles are Extremely Unlikely in Practice

[robert bristow-johnson]

On Nov 12, 2010, at 3:57 PM, Juho wrote:

> Sincere cycles are probably not very common in real elections. There  
> have already been many ranked ballot based elections with reasonably  
> sincere ballots, but at least I'm not aware of any top level cycles  
> in them.

i haven't examined other jurisdictions running STV (or "IRV" or "RCV"  
or whatever you draw from the alphabet soup), but the two elections we  
had in Burlington VT (in 2006 and 2008) using IRV indicated no cycle  
at any level.  there was a clear CW and then if you remove that  
candidate from the race, there was another clear CW with the remaining  
candidates (who i would call the runner-up), and again if you removed  
that 2nd-placer.  both elections were perfectly ordered in a Condorcet  
fashion from the CW to the CL.  the only remarkable note was that in  
2009, the CW didn't make it to the IRV final round which is the only  
way for IRV to elect someone other than the CW.  in 2006 the CW, IRV  
winner, and plurality winner were all the same candidate and in 2009,  
they were three different people.

i doubt that Condorcet cycles would be common if used in political  
elections because for the issues that matter the most and that  
motivate voters the most, it's a pretty clear spectrum from "left" to  
"right".  relatively, just not very many Nader voters would choose  
Bush over Gore nor many Bush voters would choose Nader over Gore.


On Nov 12, 2010, at 2:06 PM, fsimmons at pcc.edu wrote:

> then the best Condorcet method is the one that most
> effectively discourages artificial cycles.


i think that Markus Schulze might agree with this maybe, but i think  
that while cycles would be rare, i'll bet anything that cycles with a  
Smith set of more than 3 candidates would be the most minutely  
common.  it just would not happen at a frequency higher than Black  
Swan events (like almost never).  that being the case, since Schulze  
and Ranked-Pairs pick the same winner for a Smith set of size or less,  
i would suggest that RP might be a "better" Condorcet method *only*  
because it's conceptually simpler.  if there is no cycle, it's easy to  
explain why the CW should be elected (the reason in a single sentence  
is "If the majority of voters prefer Candidate A to Candidate B, then  
Candidate B should not be elected").  if there is a cycle with a Smith  
set of 3, it's easier to explain why the same winner chosen by either  
Schulze or Tideman should be elected from the simpler POV of Tideman  
ranked-pairs rules.

transparency is important for the legitimacy of elections and if some  
rules are opaque or difficult for the average brain-dead voter (like  
searching for the beat path with the least defeat strength), then too  
many of them say "fuck this shit" and then they join the Tea Party and  
start screaming "get the government outa my Medicare!" and voting for  
empty suits like Michelle Bachmann or Sarah Palin.  sorry, since  
November 2008, i haven't been having a pleasant time in any of the  
elections, and have been mostly disgusted for the past decade.  so i  
feel like ranting.

--

r b-j                  rbj at audioimagination.com

"Imagination is more important than knowledge."