[EM] Markus reply

[Russ Paielli]

> And whenever someone submitted a definition for WDSC,
> SDSC or FBC in terms of cast preferences and asked you
> whether his definition corresponds with your intention
> of this criterion, you always refused to answer.

What we have here, it seems to me, is confusion cause by a failure to 
distinguish between two fundamentally different classes of criteria. 
Consider the basic voting process. It starts with the voters' true 
preferences, then the votes are cast, then the votes are tallied and the 
winner is determined:

true preferences --> votes cast --> winner determined

Let's call the process represented by the first arrow the "voting 
strategy" and the process represented by the second the "tally rules."

All voting system criteria that I have ever seen, excluding those that 
originated with Mike, involve the tally rules. They consider only the 
votes cast and make no reference whatsoever to the true preferences of 
the voters.

At some point Mike came along and changed the paradigm fundamentally, 
probably without ever explaining that he was doing so. Naturally, this 
causes confusion. To minimize the confusion, I suggest we distinguish 
between "normal" criteria and "Mike-style" criteria.

I'll get to the value of Mike-style criteria shortly, but first let me 
say something about the old "Technical Evaluation" page at 
ElectionMethods.org. It had a set of 8 or 9 criteria and a compliance 
table for several major election methods.

Some of those criteria were Mike-style criteria and some were normal 
criteria. However, looking back at it, I see that we had the Condorcet 
criteria defined in terms of true preferences, with the stipulation that 
the voters voted "sincerely." Stipulating that the voters vote sincerely 
simply eliminates the voting strategy and essentially converts a 
Mike-style criterion to a normal criterion. But it involves an 
unnecessary step that only confuses the matter. In other words, we had a 
normal criterion bollocksed up to make it look superficially like a 
Mike-style criterion. Now that Mike is out of the picture, such 
embarrassments have been eliminated. If and when I ever post a new 
webpage on criteria, it will have none of that crap.

Mike-style criteria involve a model that normal criteria don't need. 
They need a model of voter preferences. I had stated on the "Technical 
Evaluation" page that we assume each voter has a sincere personal 
preference list, a sincere ranking of the candidates (that may or may 
not be truncated). That seems reasonable to me. Another model could be a 
sincere cardinal (numerical) rating of the candidates, but that would be 
a more specific model.

What, then, is the value of Mike-style criteria? Voting strategy is a 
very important topic, of course, but I think that Mike just confuses the 
issue by trying to turn it into pass/fail criteria.

Take SFC, the "Strategy-Free Criterion":

"If an Ideal Democratic Winner (IDW) exists, and if a majority
prefers the IDW to another candidate, then the other candidate should
not win if that majority votes sincerely and no other voter falsifies
any preferences."

(IDW was my name for the "sincere" Condorcet winner.) The only complying 
method is (a good variation of) Condorcet. Now, what does this 
Mike-style criterion tell us? It tells us that a majority can use a 
strategy to thwart a minority. So why, then, is it called the 
"Strategy-Free" criterion? It is called that because the strategy 
doesn't require any reversal of true preferences. But it *does* involve 
insincere truncation strategy, which the criterion itself does not 
state. So the criterion name itself is misleading.

A more useful criterion is the normal (as opposed to Mike-style) 
criterion taken from Blake Cretney's website:

Name: Secret Preferences Criterion: SPC
Application: Ranked ballots
Definition:
If alternative X wins, and some of the ballots are modified in their 
rankings below X, X must still win.

Condorcet does not pass this criterion, which tells us that voters have 
incentive to truncate in some cases if not routinely. OK, then, lets 
determine when and how the voter should truncate rather than waste time 
making a useless criterion out of it. Could it be that good voting 
strategy in Condorcet involves truncation at the Approval cutoff point? 
Now *that* would be interesting.

Oh, by the way, did I mention that Mike-style criteria are pedantic?

--Russ