[EM] What it takes to give meaning to a criterion "failure"

[Michael Ossipoff]

I described a verbal scenario in which ICT fails the version of
Condorcet's criterion based on unimproved Condorcet's definition of
the verb "beat".

Jameson asked why that "failure" isn't meaningful. So, what does it
take to make it meaningful?

For it to mean something to fail version-1 of a criterion, when
passing a different version-2, version-1 must be importantly better
than version-2.

And for that to be so, when the two versions differ only in one
definition, version-1's version of that definition has to be
importantly more valid or justified than version-2's version of that
definition.

The definition in question is the matter of what it means to say that
X beats Y, when some ballots rank X and Y both in 1st place.
.
I've discussed the fact that ICT's version of that definition  is
completely consistent with the intent and wishes of the voter of that
ballot that ranks X and Y both in 1st place.

That makes it difficult to claim that the unimproved Condorcet version
of that definition is importantly more valid or justified.   Yes, I
know that many feel that _tradition_ is a justification. Sorry, I
disagree.

For that reason, ICT's "failure" of the CC based on unimproved
Condorcet's "X beats Y" definition isn't a meaningful failure or
violation.

"What about the scenario?", you say. Ok, in that scenario, X was CW,
by unimproved Condorcet's definition of "beat", and thereby wins.

Then, if we count by ICT:

But lots of the voters had ranked both X and Y in 1st place. When
those ballots are counted against the defeat of Y by X, then X no
longer beats Y, and both X and Y are unbeaten. With more 1st choice
ratings, Y wins.

...in violation of the CC version based on unimproved Condorcet's definitions.

Who has reason to be aggrieved by that? Surely you can't be aggrieved
about how I use my equal voting-power. The only person who has a right
to be aggrieved about the way ICT interprets those equal-top ballots
is the voters whose ballots they are.

Those are voters who have expressed indifference between X and Y. When
one wins instead of the other, they have no reason to be aggrieved.

What _was_ important to them all is that X or Y win, instead of
someone whom they ranked lower. And, as I've already discussed, by
unimproved Condorcet's equal-ranking interpretation, ranking X in 1st
place with Y can cause the winner to be someone you like less than
both (if you've ranked X and Y at top).

So, though those voters are indifferent between X and Y, they are not
indifferent between {X,Y} and the lower-ranked candidate whom
unimproved Condorcet might elect instead, because of their equal
ranking.

So, again, it would be difficult to claim that unimproved Condorcet's
"beat" definition is more valid or justified than that of ICT.

Therefore, a CC based on unimproved Condorcet's definition can't be
importantly more justified than one based on ICT's definition.

The CC based on unimproved Condorcet's definition of "beat" could be
called (for brevity) "unimproved CC").

ICT's "failure" of unimproved CC is only meaningful if you like
unimproved CC, and prefer it to improved CC. But that's purely your
personal feeling--to which you are, of course, entitled. But an
objective complaint about improved CC would require claiming that the
equal-ranking voters care about how we choose between two candidates
among whom they are indifferent, more than they care about electing X
or Y instead of some they genuinely like less and rank lower.


Yes, I know: "Tradition".

Mike Ossipoff