[EM] Oh yes, _that's_ why I rejected power-truncation :-)

[Michael Ossipoff]

I proposed power-truncation at EM a long time ago.

At some point during my current EM membership, I rejected it, found
out that there was a reason why I wouldn't want it.

Then, more recently, (as you know, because I've been talking about it
so much) it occurred to me that what ICT does at top end could be done
at bottom end too. It would confer LNHe compliance. And, it's
justified--compelled--by the correct and valid justification that
compels what ICT does at top-end.

That's Symmetrical ICT. I like it because it brings LNHe compliance.
But now it's occurred to me why I rejected it previously:

It spoils ICT's top-count, by which ICT gets rid of the chicken dilemma.

I'm not willing to trade getting rid of the chicken dilemma for LNHe compliance.

What gives Symmetrical ICT its LNHe compliance is that your equal
bottom ranking can cause two equal bottom ranked candidates to beat
eachother. So, each is beaten, and thereby disqualified from winning,
unless everyone is beaten.

That sounded great--until I realized that, in the chicken dilemma
examples, the C voters, by ranking only C,  can easily and reliably
make A and B both beat eachother. Then C is the easy winner, even if
the A voters and B voters all co-operate by ranking eachother's
candidates in 2nd place.

So: I withdraw my proposal of Symmetrical ICT, because its LNHe comes
at too high a price: the chicken dilemma.

ICT is a Condorcet method, which meets a more legitimately-defined
Condorcet Criterion, meets FBC, and gets rid of the chicken dilemma.
That's a lot of desirable properties.

But, in order to get rid of the chicken dilemma, by using the
top-count completion, it's necessary to use the traditional
illegitimate interpretation of equal rankings at bottom-end, thereby
giving up LNHe.

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
.
By the way, of course I didn't mean that Approval or Score directly or
literally implements Rawls' MinMax harm. But, by electing the
candidate who is acceptable to, or liked and trusted by, the most
voters, it does a better job in that regard than the ruthlessly
majoritarian methods, including all of the Condorcet versions (even
ICT, of course, because it is a Condorcet method).

That's another thing that counts for Approval or Score, vs ICT, for a
public proposal, especially since ICT's Condorcet Criterion compliance
might not matter so much if people use optimal strategy in a u/a
election.

Of course there's still the matter of ICT's luxury of not having the
chicken dilemma--except for the enactability and count-fraud
vulnerability rank-method problems that I've been talking about for
official public election method proposals.

But ICT's properties make it the best choice for informational
polling, to inform strategy in an upcoming Plurality election, by
finding the CW.

Mike Ossipoff


 Mike