[EM] To ws, re: Condorcet vs Approval

[Michael Ossipoff]

ws wrote:

We show that approval voting strategic equilibria are closely related to 
honest
Condorcet Winners. There exists an approval equilibrium with a clear
font-runner F and runner-up R if and only if the F is the clear Condorcet
Winner and R the Condorcet runner-up. In contrast, we show that Condorcet
methods can elect a Condorcet Loser with non-zero probability when voters 
vote
tactically. With strategic agents, approval voting is better at electing
Condorcet Winners than Condorcet methods!

I comment:

As I've been saying here, in a series of elections in which voters base 
their strategy on previous results, Approval quickly homes in on the voter 
median candidate, the CW, and stays there.

I certainly don't disagree about the great merit of Approval.

But ws is being unfair to Condorcet. Yes, when Condorcet measures 
defeat-strength by any measure other than winning-votes, then Condorcet 
becomes a strategic mess. In fact, nearly all rank methods are a strategic 
mess. That includes Instant Runoff.

But winning-votes Condorcet and a few other good rank methods don’t have a 
problem with tactical voting.

With those methods, the only strategy problem occurs when offensive 
order-reversal strategy is attempted. We’ve discussed this at great length 
on this mailing list. Offensive order-reversal can be thwarted and penalized 
by mere truncation. Truncation is an easy and natural way of voting, in 
comparison to order-reversal. The defense is easier and more likely than the 
offensive strategy. Some will refuse to rank candidates that they 
sufficiently dislike. Some, in a devious voting environment, will try to 
avoid ranking anyone worse than the CW, or won’t rank candidates of voters 
likely  to order-reverse.

For deterrence against offensive order-reversal, it isn’t even necessary 
that the defenders know exactly where to truncate. The fact that voters 
might be making the effort will deter the order-reversal. And, without 
order-reversal, voters will have the benefit of those methods’ SFC 
compliance. You need only vote sincerely to ensure that no one worse than 
the CW will win, if a majority are voting with you. For that, you needn’t 
even know who the CW is.

The wv Condorcet method that I recommend for public elections is Schwartz 
Sequential Dropping (SSD). But MDDA and MAMPO share wv’s most important 
strategy advantages.

In conclusion,  Approval is as good as wv says it is, but Condorcet, in its 
best versions, along with a few similar methods, is a lot better than wv 
implies.

Though Approval will arrive at the voter median after a few elections, the 
best rank methods will typically get there in their first election. Also, 
with Approval, even when your favorite has outpolled your greater evil, it 
may take a little courage to stop voting for your lesser-evil. There’s some 
question about whether progressives will be able to. Never underestimate the 
bad judgment of a voter. For the reasons mentioned in this paragraph, I’d 
rather give to the voters SSD, MDDA or MAMPO than Approval. They make less 
demand on voters’ judgment.

Mike Ossipoff