<html><head></head><body>Forest Simmons wrote:<br>
<blockquote type="cite" cite="mid:Pine.HPP.3.95.1010514145208.427C-100000@orion.cc.pcc.edu"><pre wrap="">Here's an inconsistency of IRV that I wish somebody had told me about<br>before I submitted my article to the Green Voice.<br><br>It is possible for a candidate to "win" every precinct without winning the<br>election.</pre>
</blockquote>
Guilty. I had thought of this problem once before when thinking about IRV's<br>
failure to meet the Summability Criterion, but it wasn't on my mind when<br>
I reviewed your article. Every elimination in IRV means going back to the<br>
original ballots and eliminating the loser of the round, then recounting (you<br>
can sum the ballots into bins for each possible combination of votes such as<br>
A, AB, ABC, ACB, AC, ACD, etc., but this array gets very large very<br>
quickly as the number of candidates goes up, as we saw here a few weeks<br>
ago). That means per-precinct vote totals are useless. You have to do the<br>
elimination globally, not locally, and that means you can't predict the<br>
results from local data. Even if all localities pick the same winner.<br>
<br>
So IRV is a great randomizer, but not a very good tool for democracy.<br>
<br>
Another consequence of the summability failure is that reporting IRV<br>
results will be very complicated. At least for Condorcet you could<br>
publish the overall pairwise matrix (and also the pairwise matrices for<br>
individual counties or precincts or whatever the desired resolution is).<br>
<blockquote type="cite" cite="mid:Pine.HPP.3.95.1010514145208.427C-100000@orion.cc.pcc.edu"><pre wrap="">Note that (lone mark) plurality does not suffer from this inconsistency.<br><br>Why is IRV considered better than plurality when it fails this consistency<br>test and also fails monotonicity?</pre>
</blockquote>
This is a myth that IRV proponents have concocted. They want to fool<br>
the public into believing that it's the best method, and for those who aren't<br>
as easily fooled, they want to be able to claim "Even if Approval or<br>
Condorcet is better, IRV is still an improvement over the system we have<br>
now, and besides it is the method with the political momentum behind it."<br>
Supporting IRV because it has momentum is a persuasive idea only if IRV<br>
is in fact better than Lone-Mark Plurality, which it isn't.<br>
<br>
The problems with IRV can be summarized as follows:<br>
<br>
When IRV declares a winner, the only evidence supporting that winner is<br>
that the winner is preferred by a majority to all other candidates remaining<br>
in the last round. There is no evidence that the winner is preferred by a<br>
majority to the candidates that were eliminated in early rounds. Consequently,<br>
for every round prior to the decisive round, IRV is throwing away<br>
information.<br>
<br>
The other problems with IRV (including the local/global consistency<br>
problem) are manifestations of the way it trashes information.<br>
<br>
Richard<br>
<br>
</body></html>