[EM] Condorcet vs RV

[Michael Ossipoff]

When advocates of Plurality, IRV, Borda, etc., or even RV, try to play-up 
Condorcet as having a strategy problem, it’s reminiscent of that scent in 
_How I Won the War_, in which the Lt., Michael Caine, is reviewing his 
squad, out in the Sahara desert. They’re lined up in a row for inspection. 
One of them is wearing a clown suit, and covered with blue dye. Caine walks 
past the clown, and stops at the next man, and says, “You’re out of uniform! 
You don’t have your collar-stays!”

Condorcet critics say that order-reversal is a problem. But it carries a 
penalty, an easily-enforced penalty. Defensive truncation by the intended 
victims. Offensive order-reversal can succeed only if its victims are trying 
  to help its perpetrators. Wouldn’t that make you proud of yourself.

Offensive order-reversal is an unnatural way to vote, voting  someone worse 
over a compromise that you probably need--and the Condorcet Winner (CW) is 
the ultimate compromise. Order-reversal, as a strategy, would require 
organization. Its victims would hear about it in advance. On the other hand, 
defensive truncation is natural. In fact truncation needn’t even be 
strategic. Principled voters won’t rank someone whom they don’t believe to 
deserve a vote.

Critics liken the Order-reversal/truncation game to a game of chicken. But 
it’s one in which the reversers are at a disadvantage. They have more to 
lose, and their position is plainly the immoral one.

What if there’s no significant amount of order-reversal? If 
preference-falsification doesn’t occur on a scale sufficient to change the 
outcome, then SFC and GSFC apply. As I’ve said before, SFC & GSFC are the 
pinnacle of the promise of rank-balloting. With SFC-complying methods, if a 
majority agree with you that some greater-evil is worse than the CW, and 
falsification doesn’t take place on a scale sufficient to change the 
outcome, you and your majority need do nothing more than rank sincerely in 
order to defeat the greater-evil.

Some Condorcet critics fear that Condorcet will elect a low-SU CW. But, if 
voting is spatial, based on issues, and if we use city-block distance (as 
I’ve argued we should), then the CW is always the SU maximizer. Even if we 
use Pythagorean (Euclidean) distance,  the CW is the SU maximizer under the 
conditions used in simulations, such as a multidimensional normal 
distribution of voters, or uniform distribution.

Some argue that scandals could make a candidate who is low-SU for 
non-spatial reasons. But surely the CW region of issue-space will be 
well-populated, with candidates who don't have nonspatial disutility such as 
scandals.

Some fear that people will vote someone up as a compromise merely because 
they know nothing about him/her. That's their choice. They're adults. Maybe 
they know something about another candidate, and it isn't good. Warren's 
dark-horse worry is probably a revival of that fear.

A few years ago, Kevin proposed 2 methods that meet FBC & SFC, among other 
criteria. The 2nd one was reasonably problem-free. I liked it because, by 
meeting FBC, it offers something for the very most timid voters. Now I feel 
that I was a bit cynical, and I’d rather have GSFC than FBC, appealing to 
hope  rather than to fear.

But Kevin’s method is simpler than Condorcet, and that could be a decisive 
advantage.

But now even Kevin no longer advocates that method, so I don’t usually bring 
it up here. I do claim, however, that Kevin’s FBC & SFC method should be 
proposed to CVD, because it’s simpler than Condorcet, and simpler than IRV.

Schwartz Sequential Dropping (SSD) is my best proposal for public elections. 
RV and Approval are good if people feel that SSD is too complicated, as is 
Kevin’s method.

RV, like Approval (which is the simplest RV) meets FBC & WDSC. That’s a lot 
of benefit for such a simple method. In general I prefer Approval, partly 
because of its greater simplicity, and partly because a sincere ballot and a 
strategic ballot are pretty much the same. But, with members of the public, 
I haven’t had good reactions to Approval. Everyone considers it illegal 
Plurality, a violation of “1-person-1-vote”, a rule which encodeds 
Plurallity, and maybe IRV. Because of that confusion, RV is an easier 
proposal. RV is already familiar and popular, and probably has a better 
chance of enaction than does Approval.

Of course the way to offer Approval is to first offer RV, and then, when 
there’s agreement about RV, point out that Approval is the simplest RV, the 
0,1 RV. Maybe that’s the most practical route. Propose RV, and then offer 
Approval as the simplest, most easily-implemented RV, one that doesn’t even 
need new ballots (except that “vote for 1” needs to be replaced by “vote for 
1 or more”. Two new words printed on the ballot. That’s all the change that 
Approval needs.

So that’s what I like for a public single-winner proposal:

1. SSD
2. Kevin’s method
3. Approval (introduced as the simplest RV)
4. RV

RV isn’t a bad method, and it’s very winnable, but let’s not get carried 
away and say that RV is as good as Condorcet.

Mike Ossipoff