[EM] Margins vs. Winning Votes

[Juho Laatu]

Hello Paul,

On Jul 25, 2005, at 01:42, Paul Kislanko wrote:

>  Juho Laatu wrote in part:
>> (P.S. Number of "1000 supporter parties" could be also higher
>> than two,
>> and number of candidates in each party could be higher than two, and
>> the results/problems would stay the same.)
>
> I'd be very careful with generalizations like this one. The
> three-alternative case is qualitatively different from the 
> two-alternative
> case. The example itself depends upon their being an even number of 
> voters
> so the split can result in a tie. With the same 2000 voters and a third
> candidate, you can't have a tie since 2000 is not 0 mod 3.

Ok, I'll try to be more careful with my definitions. I'll give an 
example to clarify what I was thinking. I the attached example one vote 
seems to be able to pick any winner - in this case the last vote picks 
B.

1000:  A>B>C>D
1000:  E>F>G
1000:  H>I
1000:  J
1:     B>F