[EM] Re: serious strategy problem in Condorcet, but not in IRV?

[Eric Gorr]

At 9:27 PM -0400 8/18/03, James Green-Armytage wrote:
>Dear Eric, et al
>
>>Voting sincerely is always in the interest of an individual. For in a
>>collection of people, it is the group preference that is important.
>>If enough individuals are not prepared to accept a legitimate loss,
>>the voting system will not matter.
>>
>>Only in those cases of dictatorships, etc. do pure individual
>>interests become more important.
>
>
>	I don't think that I agree with this. For one thing, there is the
>possibility than an individual actually does have a very clear personal
>interest in the election of a particular candidate. Or that they perceive
>themselves to have such an interest.
>	Another possibility is that they are so sure that they are right about
>what is good for society, and the supporters of the other candidate are
>dead wrong, that they will decide that the ends justify the means, and use
>offensive strategy.

Or, in other words, they wish to appoint themselves dictator and 
decide the election for themselves. I think you have argued my point 
for me.

>  >I still don't understand why it would be so unsettling from the
>>aspect of an election method.
>>Considering that all elections methods can be manipulated in some
>>form or another, what good is it to remind us that society can break
>>down and make the election method irrelevant by subverting the
>>process?
>
>
>Well, I find that fairly disturbing in itself, of course. But perhaps what
>worries me about this particular type of scenario is that IRV turns out to
>be less manipulable that Condorcet.

and IRV is more manipulable in other scenarios...

What is the fundamental aspect in common with all these 
scenarios...they each assume enough voters are no longer be 
interested in finding the legitimate winner and that they now have 
the power to select anyone. When that happens, the election method 
becomes quite irrelevant.

Of course, I have to mention again with respect to IRV, that even 
when you have a perfect election with people expressing only sincere 
preferences, IRV can select an obviously wrong winner.

>  >Of course such flaws should not be ignored, but the flaw that James
>>was pointing out actually lies utterly outside the scope of an
>>election method and inside of the scope of what it means to be a
>>moral/ethical participant in an election.
>
>This is not a dichotomy that voting systems designers have the luxury of
>making. If you put in place a system that works great as long as people
>vote sincerely, and it works worse in practice than another system that
>doesn't produce as-good results given sincere votes, then the people who
>implemented the voting system are partly to blame for their lack of
>foresight.

I believe it to be impossible to design a voting system devoid of all 
possible manipulation. When manipulation becomes the sole desire for 
a sufficient number of people, it will happen regardless of how 
strategy-proof it is.

I also believe that among the most important questions a designer can 
ask is would the method, given a perfect election, select an 
obviously wrong winner.

>Take cardinal ratings, for example. It would be a pretty decent voting
>method if you could expect people to vote sincerely in any meaningful way,
>but that is just not realistic. A rational voter will tend to give any
>given candidate either the highest or the lowest possible score to
>maximize their voting power, and so the end result is supposed to be
>roughly equivalent to approval.

Well, it would appear to me that Cardinal Ratings fail because it 
merely adds undo complication one what would simply turn out to be 
either an approval ballot or a ranked ballot.

But, I note that you have stated that rational people would vote 
sincerely and in a meaningful way...at least as meaningful as the 
Approval method allows.

>I'm not saying that approval is so bad, but the point is that any serious
>voting system theory (for public elections, at least) expects voters to
>try to maximize their voting power, and hence vote strategically. Voting
>systems are designed around the principle that they should produce good
>results if used this way.

Again, I believe it to be impossible to design a voting system devoid 
of all possible manipulation. When manipulation becomes the sole 
desire for a sufficient number of people, it will happen regardless 
of how strategy-proof it is...

...and manipulation will always be possible, regardless of legality.

>The process of American government now is hardly characterized by very
>scrupulous people who refuse to unfairly take advantage of loopholes in
>the system. There is no reason to expect that it would suddenly become so
>under Condorcet.

Of course not, however, under a ranked method, there would at least 
be the option again for people to express sincere opinions.

At the moment, the election method is part of the problem because 
there simply are more then two choices in the vast majority of 
elections.


-- 
== Eric Gorr ========= http://www.ericgorr.net ========= ICQ:9293199 ===
"Therefore the considerations of the intelligent always include both
benefit and harm." - Sun Tzu
== Insults, like violence, are the last refuge of the incompetent... ===