[EM] New Criterion
cbenham at adam.com.au
Tue May 20 11:23:38 PDT 2014
I've been meaning to remind you: for IRV and Benham (and Woodall and
similar) I'm strongly opposed to allowing voters to do any equal-ranking
apart from truncating because it makes Push-over strategizing much less
risky and more likely to succeed.
Two versions of ER-IRV have been discussed, one where an A=B ballot
gives a "whole vote" to each and one where it gives half a vote to each,
ER-IRV(whole) and ER-IRV(fractional). The problem I referred to is
much worse for the former and so I consider the latter to less bad.
But if we insist on allowing above-bottom equal-ranking and don't mind a
lot of extra complexity, I have this suggestion:
*Before each elimination, order the candidates according to their
ER-IRV(fractional), (so that among continuing candidates a ballot that
ranks n candidates give 1/n of a vote to each).
Then assign each of the ballots that equal-top rank more than one
candidate to whichever of them is highest in that order.
Then eliminate the candidate with the fewest ballots assigned to hir.*
So in this example of Forest's, to create the initial order the 34 A=B
ballots give half a vote each to A and B, to give the scores
B (31+17=48) > C35 > A17.
B is above A in this order, so all of the A=B ballots are assigned to B.
This gives the scores B65 > C35 > A0. A has the lowest score so
A is eliminated and B wins.
On 5/20/2014 4:45 AM, Forest Simmons wrote:
> Chris and Mike,
> your combined comments gave me an idea for a more practical version of
> A method satisfies Semi-Sincerity Relative to MAM or SS(MAM) if and
> only if a semi-sincere modification of the sincere preferences leads
> to a strategic equilibrium ballot set from which the method elects the
> the sincere MAM winner.
> This criterion recognizes the superiority of MAM under ideal
> conditions while allowing the method in question to comply with CD,
> for example.
> Suppose our method is Benham, and sincere votes are
> 34 A>B
> 31 B
> 35 C
> A semi-sincere ballot modification results in a Nash equilibrium for
> Benham that elects B, the MAM winner of the sincere ballot set (not to
> mention the modified set).
> 34 A=B
> 31 B
> 35 C
> Date: Sun, 18 May 2014 14:46:30 -0400
> From: Michael Ossipoff <email9648742 at gmail.com
> <mailto:email9648742 at gmail.com>>
> To: cbenham at adam.com.au <mailto:cbenham at adam.com.au>,
> "election-methods at electorama.com
> <mailto:election-methods at electorama.com>"
> <election-methods at electorama.com
> <mailto:election-methods at electorama.com>>
> Subject: Re: [EM] New Criterion
> <CAOKDY5DYQZtEnaoxoV8NG3OxyFBsw+59x2vpZU8maFZZEuJiqA at mail.gmail.com <mailto:CAOKDY5DYQZtEnaoxoV8NG3OxyFBsw%2B59x2vpZU8maFZZEuJiqA at mail.gmail.com>>
> Content-Type: text/plain; charset="utf-8"
> One comment that I can make right away is that FBC is almost surely
> incompatible with CD + MMC. ...just as FBC is incompatible with
> So, in Green scenario or ideal majoritarian conditions, FBC would
> be too
> costly. So, if the 2nd of Forest's criteria, too, is incompatible
> with the
> criteria desirable for the Green scenario, that's favorable to a
> of that criterion to FBC. Sure, the differences are great too..
> Of course you have a point about the desirability of sacrificing one's
> favorite in order to save the winner under sincere voting.
> It could be argued that the thing being measured for is the
> _possibiity_ of
> easily (without reversal) preserving the sincere winner, whether
> or not
> it's always desirable, and that that's a matter of interest, just
> it _could_ be desirable.
> -------------- next part --------------
> An HTML attachment was scrubbed...
> Subject: Digest Footer
> Election-Methods mailing list
> Election-Methods at lists.electorama.com
> <mailto:Election-Methods at lists.electorama.com>
> End of Election-Methods Digest, Vol 119, Issue 18
> Election-Methods mailing list - see http://electorama.com/em for list info
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Election-Methods