[EM] Simpler, obvious, easily-used FBC, directly describing the relevant basic property

[Michael Ossipoff]

On Sat, Jul 28, 2012 at 9:04 PM, Jameson Quinn <jameson.quinn at gmail.com> wrote:

> Is this [FBC7] essentially equivalent to the normal statement of FBC?

It's roughly similar in strength, in the sense that the methods I know
to pass one also pass the other. But it's not just another way of
saying the same thing. The two criteria are different in significant
ways.


>Or is it
> meaningfully stronger in some way?

Yes, I think that it is. In one important way, FBC7 gives an important
guarantee that the original FBC, known as Weak FBC, doesn't give. Weak
FBC--even if the situation where that guarantee matters isn't being
encountered. Weak FBC guarantees that, without voting someone over
your favorite, you have _some_ way of getting as good a result as you
could get by voting someone over your favorite. But, as someone
pointed out: Maybe that way that doesn't need favorite-burial is a
complicated and difficult-to-find strategy.

That was what led me to look for a better FBC.

With methods passing FBC7, you know that it isn't some mysterious
unknown strategy that can get as good a result without
favorite-burial. You know that adding additional candidates to 1st
place isn't going to make any other 1st place candidates lose. Weak
FBC is more vague in what it promises. With FBC7-passing methods, You
cannot gain by favorite-burial.  ...at least not by defeating a 1st
choice.

Of course, by raising Favorite to 1st place, you might make Favorite
beat a Compromise whom you rank in 2nd place, who would have otherwise
won instead of Worst.

That isn't a Weak FBC failure, or a fault of FBC7, because you can
avoid that by voting Compromise in 1st place too, and, if the method
is ICT, you're then fully helping Compromise, in spite of Favorite
being in 1st place too.

If there's some set of candidates whom it's important to not elect,
then you can rank everyone else in 1st place, in ICT. That's the u/a
strategy in ICT--Rank all of the acceptables in 1st place. I suppose
that (if the method isn't Double-Ended ICT) you might want to rank the
unacceptables in reverse order of winnability, in order to maybe make
a defeat for one of them. With Double-Ended iCT, there's no need to
bother with that bottom-end strategy.

It's considerably worse with ordinary (un-improved) Condorcet, with
which you typically won't know what to do in a u/a election. Ranking
all of the acceptables in 1st place creates a danger that you're
letting one of them beat another, spoiling hir CW status and giving
the win to an unacceptable.

Approval and ICT have simple and easy u/a strategy--with Double-Ended
ICT being more like Approval in that regard, because it doesn't have
even the unacceptables-ranking strategy.

Anyway, in comparison to Weak FBC, FBC7 compliance is easier to
determine, and its guarantee is more concrete.

Mike Ossipoff