y/n strategic instability. 6-Candidate example.

Mike Ossipoff dfb at bbs.cruzio.com
Sun Nov 17 19:50:08 PST 1996

DEMOREP1 at aol.com writes:
> 1. Plain Condorcet only makes relative rankings no matter how much is the
> "absolute" support for any candidate.

Why do you keep picking on plain Condorcet? EM's main recommendation
was Smith//Condorcet, not plain Condorcet.

Thank you for pointing out that Condorcet's method uses rankings
rather than "absolute support". We knew that. Relative preferencs,
the kind that are expressed by a ranking, are what the LO2E 
problem is about. And your absolute disapproval, and especially
your absolute approval, are not only completely irrelevant to
LO2E, but they're also often less exact, less easy for the voter
to know for sure that s/he feels that way. And they're completely

from what we've been trying to do with sw reform. Sometimes what
is completely different is also incompatible. Recent discussion
has brought that out in this case.

I've admitted that I'd have used "No" against Clinton & Dole
& nearly all others in the '96 election. The fact that I'd use
it doesn't mean it would be a good thing to have. We've talked
about what I've referred to as a "wholesale mutual massacre"
resulting from strategic instability, when "No" is used.

Part of "No"'s incompatibility with the pairwise relative preference
count & LO2E is mitigated by the fact that, if I & others help the
Republicans disqualify Clinton, the Nader voters will know tht
didn't happen because they didn't vote Clinton in 1st place; they'll
know it happened because of me & those like me. Still, though, it
will mean that sw reform is sosmething of an empty bag, when it
has an added feature, foreign to its purpose, which will take
away that lessser-evil that sw reform is supposed to protect.

You want to sabotage the lesser-evil? Fine. You can do that in
any method. No method can protect against it. For instance,
in Condorcet, you could do it by order-reversal. If I order-
reverse, then either my favorite (Nader) will win, or Dole
will, assuming enough others like me try order-reversal. But
don't blame Condorcet--as I said, no method can protect against
sabotage. (We've talked about how order-reversal for immediate
strategic gain, as opposed to sabotage, is well deterrred in
Condorcet). Anyway, the point here is that you can sabotage
the compromise without having that option offered to you
on the ballot--no matter what the method is. So don't feel
to deprived if we don't want to include the incongrouos,
foreign & incompatible absolute "No" vote as an express ballot
> 2. The competitors (top 2 runoff, approval voting, instant run-off) to plain
> Condorcet all can routinely produce a majority winner (who in many cases will
> not be the Condorcet winner).

So? None of us are advocating those other methods. And presumably
what you mean by "majority winner" is "false majority winner", because
that's the kind that Runoff & Instant Runoff will often elect.

> 3. Plain Condorcet has a major problem with the strategies involving
> candidates having a high plurality of first choice votes (or candidates
> having a nearly equal number of first choice votes) and truncated votes.
> Examples

Please excuse me if I postpone dealing with these examples. I have
yet to answer Don's ER letter & Demorep's 7-voter letter. One
thing at a time. I'll answer this example after those other
2 letters.

> 1.   A 49, B 50, C 1 (an extreme example)
> 2.   A 45, B 41, C 14
> 3.   A 34, B 33, C 33 
> Each of the high plurality or nearly equal candidates will tell their
> supporters not to vote for any other candidate while attempting to get votes
> from the candidate(s) who have few first choice votes.

As has been explained to you so many times (like talking to a wall),
the only candidate whose voters have reason to "bullet vote" is
a candidate likely to be middle Condorcet winner, in a
(most improbable) devious electorate situation, where order-reversal
has been discussed or is considered a possibility.

> Regarding circular ties and truncated votes- See Subj:  Re: Circular Tie
> Percentages???  Date:  Sun, Nov 17, 1996 2:48 AM EDT by Mr. Ossipoff. 
> 4. A plain Condorcet tie breaker winner will by definition have been defeated
> by at least one other candidate (if the tie breaker is the fewest votes
> against in his/her worst defeat).  

Wrong. Not "if the tie-breaker is fewest votes against in his/her worst
defeat". In a circular tie the winner will be someone beaten, because
everyone is beaten. Don't blame that on Condorcet's method.

> Is Mr. Ossipoff ready to take the total political heat from the media, the
> defeated candidates and the voters who voted to defeat such winner if and
> when such an event occurs ?

Yes I am. The other candidates have even more people voting against
them, and so they're not in a position to complain. To the voters
who voted against the winner, it can be pointed out that if
no one with a pairwise defeat can win then in that election 
there'd be no winner--and in that power vacuum, Ghengis Khan,
Hitler & Stalin might take over :-)

If you want to say that the election is in-valid when no
one beats each of the other, or when no one is unbeaten,
then you're making it easy for truncators to make it impossible
to get a result, and easy for them to perpetuallly prevent
there from being a President (& maby a Senate, etc.).

> Is there no strategic instability in plain Condorcet ?

Yes there is none. We've discussed that at great length. I hope
you have copies of the earlier letters about that. If not,
then I refer you to the archive in which is kept all the
letters of this list. We've thoroughly covered that topic.

> .-


More information about the Election-Methods mailing list