On May 5, 2010, at 10:52 PM, robert bristow-johnson wrote:
Terry, i didn't originally intend to just pile on ...
On May 5, 2010, at 9:48 PM, Dave Ketchum wrote:
"a Condorcet winner can be a candidate that has the fewest first
preferences."
True in Condorcet, though not expected to happen often.
i dunno, but my derriere still hurts. in Burlington Vermont this
happens 50% of the time (we had two IRV elections and one of them
actually elected the CW).
Careful - the particular topic is "fewest" - for which IRV certainly
discards the CW. IRV discards the CW in many other cases.
Compared with each other candidate, the CW must win in each such
pair. Each such can have first preference over the CW as seen by
SOME voters.
it could be virtually *all* of the voters (but in Burlington in
2009, 23% chose the CW as their first preference). still doesn't
make the CW a bad candidate to elect.
Careful - "each" having a first preference means that voter ranked
"each" over the CW - enough such voters would mean that that supposed
CW could not actually be such.
IRV, looking only at first preferences when deciding [who to
eliminate, may eliminate] such a CW. It is IRV's discarding
without looking at all that the voters vote that makes many of us
desire to discard IRV.
for me, it's just that IRV does not necessarily elect the CW when
such exists. i am still convinced that it is fundamental in a
democracy where each citizen's vote counts equally, that if a
majority of voters agree that Candidate A (as in "Andy") is a better
choice than Candidate B (as in "Bob"), then Candidate B should not
be elected [unless perhaps when there is a cycle]. it's as simple as
that, and because that is not the primary function of IRV, that's
why it comes up short.
Setting up rules is tricky, but many of us choke when we look
carefully at IRV.
it's similar to the existence of the Electoral College in US
presidential elections. the E.C. doesn't do too bad when it elects
the same candidate with the popular majority, but when it doesn't
(like in 2000) it *never* brings legitimacy to the election result.
you don't hear people say "Whew! That was close! Boy am I glad we
have this Electoral College to protect us from the rule of the
population!" so the E.C. does well when it agrees with the popular
vote tabulation and not so well when it doesn't. it raises the
question as to why we should use the electoral vote over the popular
vote at all.
Before throwing rocks at the Electoral College it would pay to think
about how you would have managed campaigning for intelligent voting to
elect a President at the time the EC was created. What could and
should be done now is an interesting topic.
likewise with IRV and Condorcet. why bother with the IRV tabulation
at all when the best we can hope for it is that it *may* likely
elect the Condorcet winner, the candidate who is unambiguously
preferred by the majority of voters to any other specific candidate
when these voters are asked to choose between the two. this is, i
think, why Nobel Laureate Eric Maskin calls the Condorcet winner the
"true majority" winner.
Here those who think seem agreed it is simply time to discard IRV.
Terry, you and Rob and company still need to address this
philosophical deficit of IRV (besides all of the other anomalies
that result when IRV fails to elect the CW). i think that Tony
Gierzynski's analysis of the 2009 election was good only to the
point where it drew facts from Warren's quantitative analysis (and i
disagree specifically with Tony's conclusion where he says that IRV
is merely a technical solution to a political problem), but you and
Rob have *failed* to refute the identified pathologies of the 2009
election. because we have discussed this over tea, i still think
that you "get it", but i just cannot see that Rob (and Paul F) and
company "get it". IRV is repudiated and the trajectory doesn't look
so good for it. FairVote needs to reconsider its position on it
rather than just how best to market it.
Letting loose can be very difficult - but many of us are urging
FairVote to swallow the bitter pill.
--
r b-j [email protected]
----
Election-Methods mailing list - see http://electorama.com/em for list info
