[EM] Condorcet completed by SC reversed-rankings IRV elimination

James Green-Armytage jarmyta at antioch-college.edu
Mon Jan 19 00:08:02 PST 2004


Dear Chris Benham, et al,

	Here is my first reaction to the CCSCRRIRVE method.
	First of all, it's interesting. I appreciate your ability to suggest
thought-provoking new methods. (I note in passing that you seem inclined
toward methods which are double-reversals from existing methods, such as
your Condorcet loser elimination idea, and now a reverse IRV winner
elimination.) Also, your method is Condorcet efficient, which I like of
course.
	As far as being resistant to burying, I'm not so sure... maybe though.
You aren't claiming that the method is immune to the burying strategy, are
you? No, I don't think that you're making that claim. What you have done
is shown particular examples where successful burying strategies are more
difficult using CCSCRRIRVE than WV. This, in itself, doesn't prove a lot.
It is quite possible that CCSCRRIRVE is in some ways more resistant to
burying than WV, but I'm not quite sure how to go about showing that.
	However, more to the point, the CCSCRRIRVE method will likely introduce
more (perhaps quite complex) strategic incentives that are not found at
all in WV minimax / sequential dropping, ranked pairs, or beatpath. 
	For example, CCSCRRIRVE seems to belong to the class of Condorcet methods
where the completion method taken on its own (without an explicit
statement that a Condorcet winner should be selected if one exists) is not
Condorcet efficient. (Um, SCRRIRVE *isn't* Condorcet efficient, is it?...)
Thus it is in a class with Black, where the Borda winner is selected when
no Condorcet winner exists, and whatever it's called when you take the IRV
winner when there's no Condorcet winner. Let's call this Condorcet / IRV
for now.
	The problem with these methods, as I mentioned in my proposal, is that
first of all you have the strategic incentives for voters to prevent a
Condorcet winner from occurring if the completion method winner is more
likely to be a favorite of theirs... and then you have the strategic
incentives inherent in the completion method in itself.
	Condorcet / IRV is arguably more resistant to burying than WV Condorcet
methods. It is not an awful method, and it is actually the method that I
chose to propose to my school for student government elections before I
left there in December. However, it can give incentives for people to
prevent a Condorcet winner from being found if their candidate is a likely
IRV winner. Then, it has some of the strategic problems of IRV, such as
its relatively more prevalent favorite betrayal incentive, and perhaps the
"pushing-over" strategy as well, which as far as I can tell takes
advantage of IRV's lack of monotonicity (or mono-raise, I think, in
Woodall's terminology).
	Likewise, in order to analyze the strategic incentives in CCSCRRIRVE, I
think you would want to first analyze the strategic incentives in
SCRRIRVE. 
	Anyway, this is why these "hybrid" Condorcet methods haven't sparked my
interest yet. I prefer a "parsimonious" voting method, because it seems
that the more complexity there is in the tally rule, the more convoluted
the strategic incentives will become.
	Anyway, I'm sorry to send such a crotchety and close-minded message.
Perhaps some of it is that I have come to my own conclusion about how to
fight the burying problem in Condorcet, and I am defensively skeptical
about other solutions. Sorry again if that's what I'm up to. Sorry also if
I've made any goofy errors. Anyway, I hope I've been helpful nonetheless.

my best,
James





> I propose and reccomend this single-winner  Condorcet  compliant method:
>Plain ranked-ballots, equal preferences and truncation ok.
>1: Eliminate all candidates who are not members of the Schwartz set.
>2: If  more than one candidate remains, then based on the symetrically
>completed (SC) and reversed rankings,
>eliminate the candidate picked by the Alternative Vote (aka IRV).
>Repeat steps 1 and 2 until only one candidate (the winner) remains
> 
>In terms of criteria mentioned by Woodall, it has in common with Winning
>Votes  that it  meets the Plurality Criterion,
>and because of this (combined with meeting  Condorcet) also fails
> Mono-add-top, Mono-raise-random, 
>Mono-sub-top, Mono-raise-delete, and Mono-sub-plunp.  
>(These criteria are defined here:  [
>http://groups.yahoo.com/group/election-methods-list/files/wood1996.pdf
>]http://groups.yahoo.com/group/election-methods-list/files/wood1996.pdf  )
>Unlike  WV, this method  meets  Symetric Completion, and  I  believe that
>that allows it to meet my 
>Decisiveness Fairness Standard, which means means meeting Kevin Venzke's
> "Earlier-no-harm" and "Earlier-no-help"
>criteria  (introduced here: [
>http://lists.electorama.com/pipermail/election-methods-electorama.com/2003-December/011480.html
>]http://lists.electorama.com/pipermail/election-methods-electorama.com/2003-December/011480.html
>)
>
>This method makes use of IRV's great resistance to Burying  (aka
>"offensive order-reversal"), so that in this respect  I believe
>(with some evidence) that it performs  better than Winning Votes.
>Some examples (I copied from somewhere):
>Sincere preferences are:
>44: A>B>C
>14: B>C>A
>14: B>A>C
>28: C>B>A
>100 ballots. B is the CW.
>
>The A voters try to "Bury" B:
>44: A>C>B
>14: B>C>A
>14: B>A>C
>28: C>B>A
>and it backfires.  A is eliminated and  C wins. Schulze, Tideman,
>Simpson, Raynaud, LeGrand all pick A.
>C has the highest Borda score. (Borda is not fit to be used, but is
>allowed to comment.)




More information about the Election-Methods mailing list