[EM] Score MCA, or MCA Score

[Michael Ossipoff]

I don't know which name would be better.

First, some background:

The trouble with rank methods, and with all of the defection-resistant
methods (so far as I'm aware), is that they're too labor-intensive for
a count-fraud-secure handcount. Additionally, rankings can be counted
in innumerable ways, with the result that there's never any agreement
(and probably never will be) regarding how to count rankings.

As I've been saying, Approval is the obvious proposal, for
single-winner voting system reform, because Approval is nothing other
than Plurality done right--Plurality without its ridiculous
forced-falsification rule. It's the minimal improvement on Plurality,
but it's a remarkable powerful improvement.

To explain what I mean by that:

Plurality is a points rating system. On your Plurality ballot, you
assign points ratings of 0 and 1. But the problem is that you are
required to assign 0 to all but one of the candidates, even if that's
entirely contrary to how you'd prefer to rate them.

When the balloting-rule requires you to assign points ratings that are
contrary to your preference, then that balloting rule is requiring you
to falsify your rating preferences. Expressing a rating preference
that isn't really your preference amounts to falsifying your
preference.

Forced falsification is anti-democratic, and has no place in a
democracy's voting.

Approval is nothing other than the repeal of the forced-falsification
requirement.

Approval lets you give 0 or 1 to whatever candidate(s) you want to.

Approval is Plurality with honesty permitted, with freedom allowed.

The wrongness and obvious anti-democratic-ness of forced falsification
is such that its repeal qualifies as a voting-rights court case.

Anyway, but Approval, being simpler and more easily-counted, as
compared to ICT, also understandably requires more strategy.

People here sometimes (often) complain because Approval requires strategy.

Strategy comes with simple methods. But how bad is that strategy
really? And remember that even with the best, most deluxe and
strategy-free method, like Symmetrical ICT, its strategy-free-ness
won't do you much good when the election-result is decided by
count-fraud, and the winner chosen by whoever is running the
vote-count.

Besides, those who say that you have to use strategy in Approval are
wrong. My best Approval-voting advice is to just approve the
candidates whom you like &/or trust. It's really quite simple.

But there are also suggestions for strategy, if that's what you want.
For instance, if there's a compromise whom you feel you need, to beat
someone worse, then approve hir, in addition to those whom you like.

Approval's simple strategy can be stated by saying, "Approve the
candidates who are better than what you expect. In other words,
approve optimistically." The election results will reflect that
optimism. The winner will be the candidate who pleasantly surprises
the most voters.

But Approval has been criticized for the co-operation/defection
problem. In fact, that problem is the subject of the "Approval
bad-example" that I've discussed at EM.

That problem can happen, but Forest Simmons proposed an excellent
solution to it. Forest suggested what I call "strategic fractional
ratings" (SFR). In the Approval bad-example where the A voters and the
B voters greatly prefer A and B to C, and despise C, but want their
own candidate to be the {A,B} candidate who wins, and if you're an A
voter, and you know that the B voters would like to take advantage of
your co-cooperativeness if you approved B, then you could give to B a
_fractional_ rating, by probabilistically approving B. Approve B with
probability that is effectively like giving B a fractional Score
rating. Draw a numbered piece of paper from a bag, for example.

The idea is to give to B just enough so that B will be able to beat C
and win, only if the B voters are more numerous than the A voters.
After all, if B is bigger, then you don't mind helping B win for you.
B is then the rightful winner in {A,B}. But you don't want to help B
win if B is smaller than A, and the less-numerous B voters are taking
advantage of your co-cooperativeness. Hence SFR.

SFR could be done unilaterally, or it could be done by agreement
between the A and B voters.

The trouble with A and B voters agreeing to give eachother a full max
rating is that, then, A and B will end up with the same point
score--unless some voters defect. And the winner will be the candidate
who has the most dishonest voters. Dishonesty is rewarded. So, even
with a vote-trading agreement, SFR is better than whole maximum rating
sharing.

Of course SFR is a lot easier in Score than in Approval.

By the way, that puts Score in a whole new light: It used to be said
that the strategy in Score is to give only max and min points. But,
for defection-deterrence, SFR is a better strategy.  That means that
Score's facility in that regard is genuinely useful and important,
contrary to what I (and probably some others) had believed. In the
light of defection-deterrence, Score is looking a lot better than it
did before.

So anyway, probabilistic voting, for SFR, is a good way to deal with
the defection problem in Approval or Score.

...And Score is a good strategic enhancement of Approval, to deal with
the defection problem.

Another way that we at EM have discussed to partly deal with that
problem is to first look for a majority winner, before doing the
Approval count.

I believe that it was Forest who first proposed MCA. Later I suggested
MTA. And, somewhat after that, I proposed MTA2, which differs from MTA
in requiring a certain condition before your ballot gives Middle
ratings, when there is more than one majority-top candidate. I'd have
to look it up, but it seems to me that MTA2 said that your ballot
doesn't award any Middle ratings if any of your top candidates has a
top majority.

MCA says:

A majority candidate is a candidate who has Top rating from a majority
of the voters.

1. If there is 1 majority candidate, s/he wins.
2. If there are no majority candidates, then the winner is the
candidate with the most Top + Middle ratings.
3. If there are more than 1 majority candidates, then the winner is
the one of them with the most Top ratings.

MTA merely substitutes, in line #3, "Top + Middle" for "Top".

MTA2 stipulates that, in line #3, your ballot doesn't award any Middle
ratings if one of your Top rated candidates is a majority candidate.
(At least I think that's what MTA2 says)

As I said, MTA2 makes more sense than MTA, and is better than MTA.

I don't know that MTA2 is as good as MCA. Maybe (probably?) MCA is
better.It's certainly simpler. I think I'd propose MCA instead of MTA
or MTA2

Anyway, then, the MCA approach has its way of somewhat. sometimes
avoiding the defection problem, when it looks first for a majority
winner before looking at the Approvals.

I think that Score, and Approval, using SFR (probabilistically with
Approval) deal with it in a more generally applicable way.

Of course you could probabilistically use SFR for the Middle ratings
in MCA, MTA or MTA2 if you wanted to.

But why not allow you to give a fractional Middle rating in MCA, MTA or MTA2?

You could also regard that as having MCA's majority search before
looking at the Score ballots.

Hence the names Score MCA or MCA Score.

MCA isn't quite as simply-counted as Score. It requires a little
back-and-forth between the central count and the precincts, or at
least requires the central count to do more than just one simple sum.

...But maybe not unduly count-complicated or count-fraud-vulnerable.

I feel that if it were necessary to choose between Score and MCA as
the enhancement for Approval, Score is probably the simpler and more
generally helpful. Maybe, as a next improvement for later, Score MCA
(or MCA Score) would be good.

Mike Ossipoff