[Election-Methods] YN model - simple voting model in which range optimal, others not

Warren Smith warren.wds at gmail.com
Wed Mar 26 08:13:42 PDT 2008


>  Then have THE SAME voters vote Range and Condorcet.  I would expect
>  comparable vote counts - if that does not happen, tell us why.

--I *do* use the same voters for Range & Condorcet.  In the example answer e,
the range voters elect the best winner YYYY,
and the Condorcet voters elect the worst winner NNNN
(or at least they prefer NNNN over YYYY pairwise by a large margin: 70 to 30).

>  > --well, the voters were not biased toward Ns - they were biased toward Ys!
>
>  Quoting from the web page that I printed 3/23:
>  "e.If 70% of the voters each have ideologies consisting of 70% Ns ..."

--yes.  And that means there are 70%*70%=49%  Ns versus 51% Ys, which
yields a Y-favortism, on every single individual issue, by 51% majority.

The problem is when these issues get AGGLOMERATED into voters with
different stances in an unhappy manner, the result is that Condorcet
and other voting
systems malfunction, whereas range continues to function fine.

The other systems are not "self consistent" under such aggregation,
that is my whole point.

To make an analogy, it is like gerrymandering.   With gerrymandering,
it is possible
(and common!) for Republicans to win a majority in Congress, despite
the majority of
voters being Democrat and all votes honest.

With issue-agglomeration, it is possible in many voting systems
for the worst candidate NNNN to win, despite the majority of voters being
pro-Y on every individual issue, and despite 100% honest voting.
Range voting does not exhibit this problem.

Now it might be that in real life, gerrymandering is common and hence
a serious problem, while
this kind of issue agglomeration is uncommon and hence not a serious problem.
That seems at least plausible, but I do not currently know if it is
true.  The examples
I constructed were intended to maximally dramatize the situation - but
the problem
will also arise in less-dramatic forms.

-- 
Warren D. Smith
http://RangeVoting.org <-- add your endorsement (by clicking "endorse"
as 1st step)
and
math.temple.edu/~wds/homepage/works.html



More information about the Election-Methods mailing list