[EM] Bad Condorcet winners?

[Martin Harper]

Tom Ruen wrote:

>> Votes             Candidate (rating)
>> -----     ---------------------------------
>>  49%       A(10)                 B(1)  C(0)
>>   3%       B(10)  A(9)                 C(0)
>>  48%       C(10)                 B(1)  A(0)
> 
> 
> This shows an example where Condorcet demands more care from voters in lower
> rankings.
> Voters for A and C should simply bullet vote if they hate the centrist.

They don't hate the centrist as much as they hate their worst choice, 
though. Bullet voting for A could lead to C winning instead of B, which 
is a worse position from their point of view.

> They can also consider between the centrist and the opposite party which
> they prefer if their candidate can't win.

According to the example, they rate the centrist 1, and the opposite 
party 0. That means they prefer the centrist.

> In Condorcet, this should be: A=49, C=48, BA=3.
> Then A is Condorcet candidate.
> 
> If they're really worried about B: AC=49, CA=48, BA=3.
> This more powerfully resists the centrist.

This is also an insincere set of votes. And again, if only one party did 
this but not the other, they could end up replacing B, who they dislike, 
with the opposition, which they dislike more.

But doesn't every method get hit by this, provided you have strategical 
voters and good information?

For example, in approval, if everyone uses a 0-info strategy, the A camp 
bullet vote A, the B camp vote AB, and the C camp vote C, resulting in a 
win for A. But the C camp could change this A to a B by voting CB - and 
this is a more preferred outcome from their viewpoint.