[EM] SITC vs [what?]

[Juho Laatu]

On 25.9.2012, at 7.56, Michael Ossipoff wrote:

> You said:
> 
> Minmax(margins) can elect outside the top cycle if such a candidate is
> closest to being a CW (measured in number of required additional
> votes)
> 
> [endquote]
> 
> Now,  you see, that's exactly what I was talking about. Now you're
> back to Dodgson again, aren't you.

I think I'm still at Minmax(margins). But you are right that the difference is not important.

> ..quite aside from the fact that I've told you why SITC does well when
> people rank sincerely.

I still have not heard you claim that SITC could be used as a definition of the best sincere winner. Is it an ideal definition of an ideal winner, or is it a practical method that performs almost ideally also when voters use strategies?

> With Approval, Score, or SITC, the voters will decide
> that for themselves.

Note that in multi-party countries representative bodies are typically elected using a multi-winner proportional method, not using a single-winner method in singe-member districts. The latter approach tends to maintain a strong role of the largest parties. When I talked about the possiility of keeping a two-party system, I thought that the latter approach would mean 50% interest to maintain a two-party (or few-party) system. In typical multi-party systems presidential and parliamentary election methods are normally very different (single-winner vs. multi-winner).

>> I see the sincere winner criterion and strategic concerns as two separate topics.
> 
> What is the sincere winner criterion? The methods that I advocate are
> the most likely to encourage sincere voting, or relatively sincere
> voting, in comparison with other methods.

Throughout this mail stream I have tried to talk about two separate topics: "who would be the ideal sincere winner" and "what method to use in practice". Strategic concerns may infuence the selection of the latter, but not the definition of the former. If your methods "define sincere voting" then they are "sincere winner criteria" themselves. If they encourage sincere voting by some additional tricks like properties that disourage strategic voting, then those modifications/tricks probably cause a deviation that means that the ideal winner will not be always elected with sincere votes (since the method differs from the ideal sincere winner definition).

> And that _is_ a strategic topic. That's because certain strategy-needs
> are what can and does distort sincere voting--the only thing that can
> distort and prevent sincere voting.

This sounds like certain strategy defence means are in place, and the method has therefore not been designed based on the ideal sincere winner criterion only, and therefore it does not always elect the ideal winner with sincere votes. The alternative explanation would be that the actual method and the ideal sincere winner criterion happen to coincide (which sounds unlikely). It may well be that the method elects more often or more ideal winners than a method that would implement exactly the ideal sincere winner definition (because of the strategic votes or increased number of sincere votes).

> You think that strategy-freeness and good sincere results are
> mutually incompatible.

No, I say that a method whose behaviour has been tweaked so that it performs well in strategic environments, normally does not always elect the ideal winner (in whatever way one defines that) with sincere votes.

>>> If there will be defection in situations like the chicken
>>> dilemma examples, then can you still advocate Beatpath,
>>> MinMax(margins) or Dodgson over SITC, by saying they will get sincere
>>> rankings?
>> 
>> You have to pick the method so that strategic concerns will be properly addressed. I don't want to take position if one of those is >absolutely better than others (since that is not relevant to my claim).
> 
> I don't know what that means.

Nothing important, just restating the oblious fact that practical methods must be selected based on practical requirements, and that my intention was not to estimate the level of chicken dilemma problems in various methods but just to discuss the relation and differences between ideal sincere winner definitions and practical methods.

>>> 1. What makes you think that MinMax(margins), Dodgson, or Beatpath
>>> won't have a chicken dilemma?
>> 
>> I already said that I do believe that basic Condorcet methods are not very prone to this problem. I know that you disagree. Maybe you'll find one day a proof that will convince me.
> 
> I'm going to repeat this all over again for you:

I'll comment this winning votes example together with the margins bease example in the other chicken dilemma mail.

Juho