Demorep1: Truncation. Approval.

Hugh Tobin htobin at redstone.net
Fri Dec 6 21:35:04 PST 1996


DEMOREP1 at aol.com wrote:
> 
> A Condorcet major truncation example--
> A 334
> B 333
> C 330
> CB   2
> 
> 2 of 999 voters made a second choice.   B wins a glorious victory and a
> mighty mandate to be El Supremo by beating A 335 to 334 and by beating C 333
> to 332.  Plain Condorcet supporters apparently would say this is a perfectly
> OK result.  [snip]

There is nothing wrong with the result reached by Condorcet (or IRO) in
this example.  "Majority" is relevant only with respect to the number of
voters who turn out to vote; nobody ever gets a majority of eligible
voters.  The way you turn out to vote in races that will matter if your
favorite candidate loses, in Condorcet or IRO, is to rank second and
lower choices.  In an actual runoff a "majority" is not computed based
on the larger base including those who voted in the earlier round; why
should it be different when rankings make the process more convenient
for the voter?  Your example seems implausible, and ignores a real
benefit of conducting all necessary races at once -- some Congressional
races are now undergoing actual runoffs where turnouts will be much
lower for the finals than they were in November, because people regard
elections as over, and would rather go Xmas shopping that to the polls
to vote on a single race.  This means not only that many who voted for
the candidates who have alreadly lost will not take the trouble to vote,
but also that some of those who voted in the first round for the
finalists will not show up, yielding a real minority winner, who would
have lost under any ranked ballot scheme that would guarantee a decisive
outcome.  The same perverse result could occur with an Approval system
if it required a new vote, particularly one involving only a state or
local race that does not get a lot of media attention.  
	As to strategic voting, of course Arrow shows that it cannot be
excluded.  I think Condorcet supporters should be open about the
possibility, and explore ways to minimize it, as has been done on this
list.  But approval introduces worse strategy problems, and if added to
Condorcet as was proposed may elect A when he would lose a head-to-head
race with B, which result is not consistent with any concept of majority
rule.

-- Hugh Tobin




More information about the Election-Methods mailing list