Let's found an organization to oppose IRV

Bart Ingles bartman at netgate.net
Thu Nov 16 20:20:39 PST 2000

> [...] When Merrill did his SU simulation study, he found that the
> tendency of candidates to crowd the center greatly worsens IRV's
> already poor SU performance.
> When candidates crowd the center, or when there are many candidates,
> IRV does worse at picking sincere CWs in Merrill's simulations.

I took another look at the graphs in Merrill's book, and it looks to me
as though the 'ABC' methods are relatively unaffected by candidate
dispersion, while IRV/Runoff/Plurality are all significantly worse when
the candidates crowd the center.  So it seems to be more a case of IRV
punishing centrists, than of the other methods rewarding centrists.

I meant to ask if there were any suggestions for a better term than
'social utility'. 

I think 'utility' is probably acceptable when talking about individual
voters, since we *are* generally talking about Von Neumann-Morgenstern
utilities.  Otherwise, when talking about ratings (as in a number which
a voter would consciously assign, in the absence of some cost/risk
comparison), I usually use the term ratings (generally as 'sincere
ratings', since I don't advocate using ratings in an actual system).

But for the aggregate value, 'utility' probably is a misnomer.  I think
the concept is mainly useful when near its maximum or minimum values (or
at least when comprised of voters' utilities near the max. or min.)  I
tend to discount middling values, except that comparing a very high or
low value to a middling value might be useful.  'Consensusness' seems a
bit awkward.  Probably no more so than my wording in the last couple of
paragraphs.  It's been a long day...

More information about the Election-Methods mailing list