[EM] Need for criteria, desirability of criteria. Organizations.

[Michael Ossipoff]

Discussion of voting systems has mostly been about official public
elections in the United States. That's why I've emphasised FBC. The
strategy situation here is abnormal and pathological, consisting of
thoroughly dishonest and disinformational media, and a public who
believe whatever those media say. Those pathological conditions make a
special demand on the voting system. Ii refer to those U.S. public
official election conditions as "current conditions".

But other than for U.S. official public elections, just speaking of
voting in general, I make no claim that FBC is necessary. FBC is
always desirable, certainly, but not normally necessary.

Even without the special situation described in the first paragraph,
above, there's a case for FBC. It's just that there's also a case for
other advantages instead. Obviously it would be nice if there could be
no favorite-burial incentive. The drawback, however, is that, even
with the FBC-complying methods, fully protecting a compromise, while
top-voting your favorite, requires voting the compromise equal to your
favorite. Under different conditions, where voters aren't inclined to
do that, there's no need for those FBC complying methods, and that
makes some ambitious properties available.

Approval and Score, the best proposals for official public elections
under U.S. current conditions, would also be fine for general use, in
all voting applications.

But, when the serious _need_ for FBC isn't present, there are other
things that we can ask for. For instance, there's the rank-balloting
ideal, the goal of strategy-free-ness. The goal of being able to fully
act toward making the winner come from a certain preferred set (which
you can do in Approval and Score), while, at the same time, _choosing_
within that preferred set.

When voting system reform advocates criticize Approval, they criticize
it because that isn't possible with Approval.

If you want that rank-balloting ideal for all of the voters, that
isn't attainable. But it's attainable for a particular set of voters,
an innermost mutual majority (MM). That can be achieved by compliance
with the Mutual Majority Criterion (MMC). But MMC loses its meaning if
there's a chicken dilemma. Therefore, MMC only has its full power in
methods that don't have a chicken dilemma. Such a method is IRV.
...and the various IRV hybrids that i've discussed, under the
collective name "IRV etc.", abbreviated IRV&c, or I&c.

Without FBC, favorite-burial need, under current conditions, is a
certainty, and FBC is essential. But under the different "Green
scenario" conditions that I've discussed, voters aren't deceived, for
the reasons that I've described in previous posts. Then, even with an
I&c method, there's no particular reason for voters to assume that
they aren't in a MM. In fact, I suggest that most people will believe
that they're in a MM--especially if they actually are. For example, I
claim that progressives are a MM.

So the situation is entirely different when we don't have thoroughly
dishonest disinformational media and a public who believe whatever
those media say. Then, we can have much more ambitious goals for the
voting system--complete strategy-freeness for an innermost MM.

...and not just in the Green scenario. I claim that this is feasible
for any voting other than "current conditions". That includes
organizational voting and most polls.

I feel that IRV, MM//Benham, and MM//Woodall might be too adversarial
for amicable organizations. Benham, Woodall, or maybe AIRV seem
better. AIRV, of course, would be easier to handcount.

As I said, Approval or Score would be fine. I disagree with FairVote's
claim that Approval is only for non-contentous elections. Approval
would work fine even when strategy is needed. But when an organization
has to vote on a set of very different alternatives, including some
that are considerably disliked by some voters, then the I&c methods'
strategy-freeness for a MM would be a very nice luxury.

Approval or Score might be the more desirable choice when the
alternatives are all ok to each voter. Then, Approval's social
optimizations are easily available--the election of the alternative
that is liked by the most voters, for example. But I emphasize that
Approval is still quite adequate even when that isn't so--it's just
that, then, the I&c methods provide a tempting luxury.

Benham is better than Woodall if brief definition is important.
Otherwise, Woodall has the advantage of slightly better social
utility.

For small organizations, Schwartz Woodall might be better than
Woodall, because the Schwartz and Smith sets can differ when there are
pairwise-ties. The Schwartz set is more exclusive than the Smith set.
You can get into the Smith set by tying one member of it. Not so with
the Schwartz set.

So, I suggest, for organizational voting, Benham or Schwartz Woodall,
depending on whether the members are willing to hear Schwartz
Woodall's longer definition.

The reason why traditional Condorcet methods like Beatpath would be
inadequate is that they have the chicken dilemma, and, when they do,
their MMC compliance is meaningless.

If the alternatives aren't so different that some of them are
repugnant to some voters, then I'd suggest that Approval, or maybe
Score, would be better.

Michael Ossipoff