[EM] Condorcet Strategy Reply, part 1 (of 3)

MIKE OSSIPOFF nkklrp at hotmail.com
Fri May 28 22:33:01 PDT 2004


James--

You wrote:

	For public elections, I'm recommending the following procedure.
1. Ranked vote. Pairwise tally. If there is a Condorcet winner, they take
office.
2. If there is no Condorcet winner, non-members of the Schwartz set...

I reply:

In public elections, where pairwise ties are vanishingly rare, the Schwartz 
set is the same as the Smith set. And the Smith set is a lot briefer and 
simpler to define. So, for public election proposals, I suggest changing 
"Schwartz set" to "Smith set".

You continued:

...are
eliminated from further consideration.

I reply:

I repeat here my suggestion that there's no reason to hold a 2nd balloting 
for order-reversal protection unless the circular tie has all majority 
defeats. That's because a circular tie won by a majority-beaten candidate 
must have all majority defeats in the circular tie.

You continued:

In addition, there is a period of
time between votes where any member of the Schwartz set has the option of
dropping out of the race and removing their name from the second ballot.
3. A second ranked vote takes place. The Condorcet winner or completed
winner of the second vote takes office.

I reply:

Aside from my suggestion to only hold the 2nd balloting if the circular tie 
is all majority-beaten, I don't have any objections to that proposal. As you 
know, I prefer Approval for the 2nd balloting, but either Approval or 
Condorcet would be fine for the 2nd balloting. Others too have told me that 
they'd prefer Condorcet to Approval for the 2nd balloting.

You continued:

	At this point I think that the problem is very serious, easily serious
enough to justify another balloting if such a second balloting would help,
and possibly serious enough to render single-balloting Condorcet
critically unreliable as a voting method.

I reply:

...unreliable compared to what? IRV will jump to extremes because of its own 
idiosyncracy. Condorcet will jump to an extreme only if someone sucessfully 
uses the risky offensive order-reversal strategy. Purality can fail to elect 
a CW if sufficiently many people don't bury their favorite.

Two main differences between Condorcet and IRV or Plurality:

1. IRV & Plurality spontaneously let an extreme beat the CW, unless 
defensive strategy is knowledgably used. That happens in Condorcet only if 
the risky offensive order-reversal strategy is successfully used.

2. In all of these methods the CW can be protected by defensive strategy. In 
Condorcet that defensive strategy is equal ranking, or defensive truncation 
which, though not electing the CW, will deter offensive order-reversal. In 
IRV & Plurality, that defensive strategy requires burying one's favorite.

These are things that I've been telling you all along, but you continue to 
speak of Condorcet's strategy problem in a vacuum, as if Condorcet were the 
only method with a strategy problem. Maybe you'll just keep on repeating 
your claims without replying to these objections to them, but, if so, I just 
want everyone to know that that's what you're doing. You're repeating 
without listening.

All 3 methods, Condorcet, IRV, & Plurailty, can elect an extreme candidate 
instead of the CW. But that's where the similarity ends. These 3 methods 
differ in the two numbered ways that I described above. Of course they also 
differ in their strategy criteria compliances, the strategy guarantees that 
they offer.

You continued:

I'm not entirely sure about
this, but that's the way it seems to me now. As for whether a second
balloting would significantly mitigate this problem, I'm not sure, but I
*hope* so.

I reply:

A 2nd balloting would help. As would all the strategy enhancements that I 
described here.

You continued:

	(Actually there is also a third, namely: is there a way of mitigating the
strategy problem that is both easier (i.e. one ballot) and more
effective

I reply:

James, I posted a list of strategy enhancements for 1-balloting Condorcet 
that reduce its aready relatively minor strategy problem.

You're requiring perfection from Condorcet. You remind me of that scene from 
_How I Won the War_, in which the sergeant is inspecting the squad, out in 
the desert. He walks by the row of assembled soldiers. One of them is 
wearing a clown-suit, and is dyed  (maybe blue, green or red) from head to 
toe. The sergeant walks by the clown, and then stops at the next man, and 
says "You're out of uniform! You don't have collar-stays!".

No, you won't find perfection in any voting system. But if you compare 
Condorcet's strategy to that of other voting systems, instead of comparing 
it to perfection, you'll find that the others are worse.

You continued:

	Many people seem to disagree with me on one or both of these questions...
let's see if we can get a meaningful and productive discussion going on
the topic. Myself, I'm open-minded about the issue, and I hope that you
can be open-minded as well.

I reply:

That's big of you. You can start being open-minded by looking at Condorcet's 
strategy in comparison to the strategies of other voting sytsems, instead of 
expecting it to meet standards that the other methods fail worse than 
Condrocet does.

A supporter of the CW, wanting to deter offensive order-reversal strategy 
against his candidate, doesn't have to know who is going to attempt 
offensive order-reversal, or even if anyone is going to. All s/he has to do 
is not rank anyone except hir candidate. Then, any offensive order-reversal 
against that candidate will backfire.

You complain later in your posting that s/he can't know that hir candidate 
is CW, and that hir truncation could take hir support away from the real CW. 
Yes, James, that's the whole problem, with all the methods' strategy 
situation: We don't have perfect information. I said that in my previous 
posting. I find that I'm repeating these same answers, in reply to 
successive long postings of yours, because, in those long postings, you just 
repeat your claims, quite oblivious to the fact that they've just been 
answered.

As I said before, the difference is in 1) what it takes to bring that 
strategy situation about; and 2) What it takes to protect the CW, or to 
protect majority rule. By both of those 2 standards, Condorcet does much 
better than IRV or Pluraity. Approval beats IRV & Plurality by the 2nd 
standard.

You continued:

Sincere preferences
27: A>B
25: B>A
24: C>A
24: C>B
Pairwise comparisons
A:B = 51 : 49
A:C = 52 : 48
B:C = 52 : 48
	A is a Condorcet winner. But if just a fraction of the B voters reverse
their preferences, they can steal the election for B.
27: A>B
21: B>A
4: B>C
24: C>A
24: C>B
Pairwise comparisons
A:B = 51 : 49
A:C = 48 : 52
B:C = 52 : 48

I reply:

You originally brought up this co-operation/defection dilemma in connection 
with Approval. I answered it recently on EM, bringing it up as a Condorcet 
problem, but also discussing it in terms of Approval.

I guess it's necessary to repeat some of that: That's IRV's shining example. 
For the A & B voters.

If A or B is taken advantage of by the other, it isn't a majority rule 
violation. Though strategy problems are undesirable, we should measure them 
for comparison. In contrast, IRV's failure is often a majority rule problem.

You're looking at it from the point of view of the mutual majority, the 
{A,B} voters. What about the C voters in IRV? Say they prefer one of {A,B} 
to the other. Say they can pretty well tell that C can't win, because the A 
& B voters are a mutual majority. That means that the C voters gain nothing 
by sincerely voting C 1st in IRV. Their best strategy is to vote their {A,B} 
favorite in 1st place, burying their favorite.

So what, you might say, C can't win, and shouldn't be able to. Sure, but 
what if they're mistaken in their belief that they should use 
favorite-burial compromise strategy? That's the whole problem, you know, 
when voters don't have complete information, and mis-estimate, and either 
give away the election or fail to protect a needed compromise. You talk of 
Condorcet voters who don't have perfect information. What about those C 
voters in IRV who don't have compelte information.

So every mutual majorilty example of the kind that you use as a Condorcet or 
Approval co-operation/defection dilemma is also an IRV failure example for 
WDSC & FBC, two criteria met by Approval and by all the wv versions.

You exaggerate the benefit of defection: It will result in the loss of the 
support that you'll need from your victims in the next election. In the next 
election, you can't expect them to co-operate again. If anyone co-operates 
then it will be you. I mentioned the Tit-For-Tat strategy in my previouis 
posting. The voters who have been victimized by defection might refuse to 
co-operate with their victimizers ever again, or they might just use TFT, 
whereby they copy the most recent strategy of the other "player". 
Co-operation/defection games aren't the hopeless disaster that you might 
think they are.

Aside from that, since we're now discussing Condorcet, the ATLO option would 
solve your co-operation/defection dillema: A voters or B voters rank 
sincerely, but apply ATLO below their favorite. If no one defects, either A 
or B wins, depending on which have more voters. If one defects, making a 
strategic circular tie, then the circular tie triggers ATLO, and the 
co-operating side withdraws its support for the defecting side, and C wins. 
Both side know that will happen if they defect. Neither side defects.

I'm gonig to stop and mail this now. I'm posting this reply in 3 parts, 
because last time my reply was so long that it almost didn't post, and its 
posting was delayed because of its length. Part 2 will be along tonight or 
tomorrow.

Mike Ossipoff

_________________________________________________________________
FREE pop-up blocking with the new MSN Toolbar – get it now! 
http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/




More information about the Election-Methods mailing list