[EM] Chris: Your Losing-Votes proposal

[Michael Ossipoff]

Chris:

You wrote:

I propose the following as a reasonably practical, "summable", Condorcet method:

*Voters rank from the top however many candidates they wish.
Equal-ranking is allowed.

The result is determined from a pairwise matrix. On that matrix,
ballots that rank above bottom
any X=Y contribute one whole vote to X>Y and another to Y>X.

[endquote]

Wait a minute:

So you're saying that if you rank X and Y both in 1st place, you want
to vote for both of them to beat eachother? You want to make your
favorites beaten? Why?

The only candidates whom we can be sure that you want to vote to beat
eachother are the one's you've voted at bottom.

And yet you're saying that the bottom-voted candidates are the only
ones that you wouldn't count as beating eachother when ranked equal to
eachother.

You can't mean that, as the final completed and perfected version of
your proposal.

I, too, have wondered if there's a way of combining ICT with Beatpath
or Ranked-Pairs, to get FBC, CD, 0-info LNHLe, but also some of the
Beatpath and Ranked-Pairs properties not available in ICT and
Symmetrical ICT. I refer to MMC and Clone-Independence. But mostly
MMC.

But I don't think that those Beatpath and Ranked-Pairs properties are
compatible with FBC, CD, and 0-info LNHe. But I don't know that for
sure. It would certainly be nice if there's a combination of
Symmetrical ICT and Beatpath or Ranked-Pairs, that could provide it
all.

Is it possible to have FBC, CD, the (my broadly-defined) Condorcet
Criterion, 0-info LNHe, and MMC?


You said:

Ballots that truncate both X and Y  have no effect on the X>Y and Y>X
entries in the pairwise
matrix.

[endquote]

Ok, you want to encourage people to rank all of the candidates, and
there's certainly a case for that goal.

You said:

Instead, in the "acceptables versus unacceptables" situation it has
the more natural zero-info
strategy of just equal-top ranking the acceptables and truncating the
unacceptables.

[endquote]

ICT and Symmetrical ICT have the u/a strategy of top-ranking all of
the acceptables, whether the election is 0-info or not.

In Symmetrical ICT, there's no need to rank unacceptbles. If it's
really 0-info, then that's a certainty.

In ordinary ICT (as in Beatpath and Ranked-Pairs) it seems to me that
it's best to sincerely rank the unacceptables, unless you have
improbably-detailed predictive information.

If it's 0-info u/a, I'd expect the optimal TUC(wv) strategy to be
equal-top-ranking of the acceptables. But if it isn't 0-info, the the
TUC top-end strategy becomes unknown.

Mike Ossipoff