[EM] Amateur peer-reviewed "journal" for voting methods, criteria, and compliances?

Michael Ossipoff email9648742 at gmail.com
Sun Sep 30 12:52:59 PDT 2012


First, thanks for putting in some good words for Approval, at EM yesterday.

But I note that, in your message at or to wiiipedia, as part of your
proposal of MJ there, you referred to Approval and Score as "inferior

Didn't we make a pledge to support (at minimum, that means not attack)
 the voting systems that were recommended in the Declaration?

Now, if you'd merely given a properties-chart, that would have been better.

So I suggest that, in keeping with the Declaration pledge, you remove
from your wikipedia message the claim that Approval and Score are
inferior methods. Replace it with an objective properties-chart.

As I previously mentioned, I've read that MJ fails Participation and
Consistency. Approval and Score pass Participation and Consistency.
Don't forget to include those properties in your properties-chart.

I seem to have forgotten by what property MJ is superior to Approval
and Score. Oh yes, it minimizes the effect of strategy, right? The
dastardly strategy of extreme-rating, as opposed to sincere
rating."Sincere rating" refers to the assumption that people would,
when not being devious, rate in proportion to the alternatives'
utilities to the voter. I've described reasons why a Score voter, or a
probabilistic Approval voter, would sosmetimes have good reason to
fractional-rate. In those methods, that fractional, inbetween, rating
will have a known, voter-chosen, measured result, unlike in MJ.

This notion of minimizing the effect of strategy contains an implied
assumption that there is something devious, dishonest, unfair, about
voting in one's best interest, by extreme-rating when it's in your
best interest.

But this was all covered, in the (likewise 1-sided), EM discussion a
week or a few weeks ago, comparing MJ to Score..

Whatever you think of extreme-rating, it's a useful reference point
for comparing Score and MJ. It's how to max-rate a candidate, or
maximize help for one candidate over another. So your minimizing of
the effect of extreme rating (with respect to inbetween rating) can
also be described as  giving an inbetween rating an effect equal to
that of an extreme rating--though you often won't know _which_ extreme
rating your inbetween rating will emulate.

I've discussed the much more complicated, necessarily probabilistic,
SFR of MJ, as compared to that of Score or Approval.

And I've mentioned that MJ's inbetween ratings will more likely hurt a
compromise candidate when the candidate is polling higher, and is more
likely to help the compromise when s/he is polling lower--the opposite
of what is desired for SFR, whose goal is to help Compromise beat
Worse, only when it is Compromise that is polling better than

So, taking extreme rating as a standard reference, the difference
between Score and MJ is that MJ's inbetween ratings are unpredictable
in the direction of their effect, and act contrary to the goal of SFR,
and make SFR much more complicated to apply, and necessarily

MJ has all that, in addition to a much more wordy and elaborate
definition, with its elaborate tiebreaking bylaws.

...and its failure of Participation and Consistency.

You see, regarding those 2 criteria: One thing that can be said for
Approval, maybe for Score as well, is that it is _only_ an improvement
on Plurality. In no way does it do anything worse than Plurality. It
doesn't take anything away from what we have now, with Plurality. It
doesn't do anything ridiculous that Plurality won't do.

You can't say that for MJ.

That means that those two "embarrassment criteria" can be used against
MJ in a public enactment campaign.

"Look what their proposed method will sometimes do! Let's stick with
our current voting system--It won't do that."

Yes, that could happen to Condorcet proposals too. But at least they
give Condorcet Criterion (CC) compliance. ...and maybe
defection-resistance too, in addition to (for some Condorcet versions
but not others) the Approval and Score properties of FBC and
Later-No-Help (LNHe).

Of course, I admit that there's ample precedent for violating the
Declaration pledge, when Richard made his public attack on ICT, at DC.
ICT is a Condorcet version (IC stands for Improved
Condorcet--Condorcet with an improvement).

Mike Ossipoff

More information about the Election-Methods mailing list