[EM] To Condorcetists:

[Juho Laatu]

On 14.5.2012, at 22.03, Michael Ossipoff wrote:

I wrote and you repled:

> I don't see any denial of Gibbard-Satterthwaite or other problems. My
> understanding is that many people like Condorcet methods because they think
> that their co-operation/defection problems are relatively small (although
> they exist at least in theory).
> 
> [endquote]
> 
> Nonsense. Can you justify that claim? I've showed a whole range of numerical
> examples, from the 27,24,49 example to the 33,32,34 example. I've told how
> the problem would come about, in Condorcet, just as well as in Approval.
> Condorcet is not strategy-free, or anything close to it.

Since you had some concrete examples I'll provide some concrete feedback in line with what I asked you to provide.

Sincere opinions:
33: A>B
32: B>A
34: C

In the U.S. these must be
A = Republicans (that hate C as you told)
B = Democrats
C = Socialists

Proposed strategic votes:
33: R>D
32: D
34: S

You seem to assume winning votes since you expect D to win. So, let's continue with winning votes based Condorcet methods.

First problem with the startegy is that it is unlikely that all Democats will vote strategically. If they vote 31:D, 1:D>R, R will win. Reaching 32:D is not probable. Republicans have no reason to worry.

Second problem. If Democrats really want to win, they could focus on making D look better than R. If that leads to one voter (= 1/99 of the voters) changing their opinion from R>D to D>R, D will win. On the other had, if they give a public recommendation to their supporters to vote strategically and try to cheat the victory from R, some voters that are close to the D/R border line might get upset and change their opinion from D>R to R>D. If that happens, the Deomcrat strategy will not work even if 100% of their voters will implement the strategy as told. In this set-up it seems that it would be a better strategy for Democrats to simply continue marketing their own candidate instead of starting to market strategic voting.

Third problem. If Republicans hate the third (large) party much more than Democrats do it is not probable that all third party voters will truncate. In real life votes are more heterogeneous. If two of the Socialist supporters vote S>D instead of S, D will win even without the strategy. I'd expect more than 0 Socialist supporters to have sincere preference S>D. If this is true, then Democrats don't even need a strategy. Or is there some other real life set-up where these sincere opinions would be plausible?

Since Democrats want all their supporters to vote strategically, I guess the strategy would include a public recommendation from the party to all their supporters to truncate / bullet vote and not to express their full preferences. Individual decisions and media guidance probably is not sufficient. Different political cultures may react in different ways to such messages. In some societies people would despise such attempt to falsify the results. In some other societies people might expect the party to reveal all such dirty tricks that they could use to fight and win by whatever means.

Based on this quick analysis, and in the absense of other more convicing arguments on how this strategy might actually work in real life elections, I tend to classify this method in the "theoretical vulnerabilities" category (not in th e"practical vulnerabilities" category). I mean that this example is a bit like a Turing machine that demonstrates that something is possible in theory, but doesn't say much about how well and efficiently this strategy (or program of a Turing machine) works in some real life environment. Am I correct? Can you make this type of vulnerability more plausible by changing the numbers, votes or explanation/mapping to some real life society?

Juho