(Fwd) Re: New Condorcet-ish method: "weak & strong preferences

[Steve Eppley]

------- Forwarded Message Follows -------
From:          Mike Ossipoff <dfb at bbs.cruzio.com>
Subject:       Re: New Condorcet-ish method: "weak & strong preferences"
Date:          Thu, 6 Mar 97 4:19:32 PST

The weak preferences option is similar to the "circular tie option"
that I used to talk about. Circular tie option:

A voter has the option to indicate on his ballot that, in the
event of a circular tie containing alternatives above & below
a point in the ranking designated by him, the alternatives below
that point shall be dropped from his ranking.

***

I too considered allowing the use of the circular tie option
at several points in one's ranking, which makes it more like
your suggestion, but it seems to me that it turned out that
there wasn't strategic advantage to using it at more than one
point in one's ranking.

***

I consider it a good addition to Condorcet's method, for
devious electorates, because it allows one to use fully effective
strategy, while still including every alternative in one's ranking.
Defensive strategy becomes using the option rather than voting
a short ranking. As you suggested, it doesn't have potential
for offensive strategic use, but is only a way to facilitate
defensive strategy, & make it even less drastic than Condorcet(EM)'s
defensive strategy already is.

But I don't propose the circular tie option, and merely mention
it as a possibility for later, because devious-electorate refinements
probably won't be needed, especially at first, and so it's
best to simplify by just proposing ordinary Condorcet(EM) or
Smith//Condorcet(EM).

***

One thing should be added to my wording of the circular tie option:
It should only go into effect when everything in the circular tie
is beaten with a majority against it, because those are the only
conditions in which order-reversal could otherwise be accomplishing
anything.
 
***

By the way, to try to make the circular tie option automatic,
I've talked about a "conditional circular tie option", where
the dropping of alternatives from one's ranking is triggered
by a combination of everything in the Smith set having a majority
against it, and some other condition detected by the counting
computer, some condition that reveals a likely order-reversal.

Of course, if the method were plain Condorcet, then instead
of "everything in the Smith set", it would just be "everything".

I so far have 2 proposals for what that other condition could
be. My best proposal is that you can indicate that if some
rankings are inconsistant with a "spectrum oredering" specified
by you (on your ballot), and if, without some of those inconsistant
rankings, the winner would be something that you like better than
the existing winner, then you want dropped from your ranking
that existing winner. It might be desirable to run this procedure
iteratively till it is no longer dropping anyone from anyone's
ranking.

My other proposal was about "unanimity". It's based on the
fact that supporters of some particular candidate won't really
cheat unanimously, and, if they do, then surely the publicity
&/or discussion needed to organize that mass cheating would
tip-off the intended victims. 

***

Mike