Truncation with Condorcet

[Mike Ositoff]

> 
>  


In reply to the below-copied letter: Alight, the A voters gained
by voting un-felt preferences. But in your example, that
strategy didn't defeat a Concorcet winner, and wasn't needed
to protect one. So, though I must admit that your statement
is true, your example isn't about a really important situation.

Arrow, Gibbard & Satterthwaite have shown that no method is
ideal. Gibbard & Satterthwaite have shown that they're
all manipulable. The important thing is: When is strategy
needed to protect a Condorcet winner. I emphasize that I'm
just saying that that's what's important to me. Examples
with chaotic natural circular ties, where there's no 
really obviously right winner (the Condorcet winner) don't
bother me. As bad-examples, they don't seem so bad, to me.

Your letter is copied below, but the above concludes my reply to
it.

> What about this example?
> 
> 24 A B C \  Sincere preference A > B=C
> 24 A C B /
> 23 B A C
> 29 C B A
> 
> The A 1st voters took my advice and marked the other 
> candidates randomly instead of truncating.  They came 
> out even.
> 
>        A    B    C
>   A    X    48   71
>   B    52   X    47
>   C    29   53   X
> 
> Max loss
>        52   53   71
> A wins
> 
> Now here's what happens if they vote sincerely and 
> truncate
> 
> 48 A
> 23 B A C
> 29 C B A
>        A    B    C
>   A    X    48   71
>   B    52   X    23
>   C    29   29   X
> 
> Max loss
>        52   48   71
> B wins
> 
> So unless I've made some kind of error here, it can 
> sometimes hurt to use truncation instead of my random 
> fill strategy.
> 
> The question is, is there any situation where this can 
> backfire?  That would give a reason not to always use 
> my random fill strategy.
> 
> 
> 
> -----== Sent via Deja News, The Discussion Network ==-----
> http://www.dejanews.com/  Easy access to 50,000+ discussion forums
> 
>