[EM] Getting carried away and claiming RV is as good as Condorcet...

[Abd ul-Rahman Lomax]

At 02:20 AM 2/9/2007, Scott Ritchie wrote:
>On Fri, 2007-02-09 at 01:54 -0500, Abd ul-Rahman Lomax wrote:
>
> > One of the characteristics of Range is that there is never an
> > incentive to reverse preferences. Strategic or insincere voting in
> > Range is limited to equating candidates at the extremes when, in
> > fact, you do have a preference between them. Warren calls this
> > partially sincere. I've called it "magnification;' the gain on an
> > amplifier has been increased until the output pegs the meter. The
> > contrast on an image has been increased until some detail is lost in
> > total black or white. It isn't insincere in the same way as
> > preference reversal is.
>
>So, wait, was half the Condorcet electorate strategic voting by doing
>order reversal?  Are we making the assumption that strategic voting is
>exactly as common in range and Condorcet in these simulations?

That would seem to be an unjustified assumption, of course. However, 
the *only* way to vote strategically in Condorcet is to reverse order.

In Range, it makes no sense to reverse order.

I'm not going to comment on Warren's exact methodology. He's created 
a tool for analyzing election methods, and I'm quite sure there are 
details to be added.

>That seems a bit strong, exactly because the risks are different and the
>information required is greater for Condorcet.

One thing that Warren's work seems to have done is to answer the 
common objection to Range that sincere voters will be harmed by 
strategic voters. That doesn't appear to be true, with a necessary 
qualification. "Harmed" must mean "Significantly harmed."

Further, as I've noted, the meaning of "strategic voting" is 
different under Range. It could only refer to "magnification." Which 
isn't order reversal, it's quite a different animal, and, it can 
easily be argued that it is *not* insincere.

Now, there is a strategy which some voters might use, which is 
actually foolish, which is true reversal. That is, the voters sincere 
preferences are A>B>C>D, and the voter believes that B has a 
reasonable chance at defeating A. So the voter, apparently, wants to 
defeat B, and so rates B at zero. But, not being concerned about C or 
D, the voter keeps a sincere rating for them. This is order reversal.

It would seem.

But my argument would be that if the voter is so desirous of A 
winning that the voter would maximally downrate B, the voter *does* 
have a strong preference. What actually doesn't make sense here is 
the C>D>B ratings. Rationally, the voter should vote A 100, B,C,D 0. 
It's a sincere vote.

I don't know to what extent Warren has considered all this; my 
understanding is that he believes the tool needs refinement.

The fact is that, though Range is commonly asserted to fail the 
Majority Criterion, a majority of voters cannot fail to elect their 
preference unless they *also* consider another candidate as an 
acceptable winner. It's true that you can construct scenarios that 
appear to contradict this statement, by making the majority margin 
very small, and by arranging that the contrary voters all vote 
strategically. It's actually a warped expectation.

I've debated this matter of the Majority Criterion here. What I 
noticed was that Approval is asserted to fail the MC, but Approval 
does provide a means to vote strict preference. If a majority of 
voters vote to express a strict preference for a candidate, they 
cannot fail to prevail. No method can satisfy the Majority Criterion 
if the voters don't indicate their strict prefrerence on the ballot!