[EM] Criteria

[Michael Ossipoff]

[quote]
Exactly what kind of criteria you would use would depend on the voter
population.
[endquote]

Quite.

Regarding that matter, I'll define two terms here:

"Current conditions" means:

The conditions existing now, with 1) Media that rarely if ever mention
non-Republocrat political parties, platforms, policy-proposals or
candidates, and which continually promote the belief that the winner
will always be a Democrat or a Republican.   ...and the belief that
Republocrat corruption, the boughtness of candidates, is a fact of
political life and is therefore acceptable because it's inevitable;
and 2) An electorate that believes those media claims.

"Green scenario" means:

Conditions that would exist after our electorate has, via Plurality,
elected Green Party candidates to the presidency and most of Congress
and state legislatures.

As I've mentioned before, any electorate that could do that would be
one that doesn't believe the old media disinformation. Additionally,
according to the GPUS platform, a Green government would mandate a
vastly more open and honest media system, without corporate control,
and with the opportunity of extensive public participation and freedom
of expression for all, including a much wider range of political
parties, candidates and policy positions.

In those ways, the Green scenario would be very different from current
conditions.

The choice of a voting system, the desirability-comparison of voting
systems, depends almost entirely on the conditions, such as those
conditions described above.

[quote]
 If the voters are very strategic
[endquote]

...as they are, under current conditions.

Of course one could always speculate that maybe they'd be sincere with
a different method. I've discussed that i, in the case of traditional
unimproved Condorcet (TUC).

[quote]
, strategic resistance
criteria is [He meant "are"] important or you get garbage-in garbage-out.
[endquote]

Exactly. Except the use of the word "resistance" reveals confusion
about strategy. It implies that strategy is something of an offensive
nature that voters might attempt, in order to subvert the voting
system results, and that therefore the method needs to be able to
resist those nefarious attempts by voters.

Wrong. The significant and meaningful strategy problems consist of
strategy-needs. The more drastic and preference-distorting the
strategy, the worse the strategy-need is.

Sure, methods that fail the Chicken Dilemma Criterion (CD) are
vulnerable to an offensive defection strategy. But the importance of
that comes from the strategy problem that that vulnerability puts on
the preferrers of a candidate who is likely to be defected-against.

Thus, vulnerability to offensive strategy can cause strategy
need/dilemma for other voters, and _that_ is the reason why
vulnerability to offensive strategy matters.

But the actual strategy problem is the strategy need or strategy dilemma,
which may or may not be a result of offensive strategy vulnerability.

[quote]
 If they are
not, then accurate translation criteria are more important
[endquote]

In the Green scenario, FBC isn't nearly as important as it is under
current conditions. . But CD still matters, because CD failure can
take away most of the meaning and value of Mutual Majority Criterion
(MMC).  For instance, it does so in the case of Beatpath or Bucklin.

Approval has "accurate translation" under all conditions. Even under
current conditions, Approval will accurately measure likedness,
trustedness and acceptability, and will elect the candidate who liked
or trusted by, or acceptable to, the most people.

Approval, unlike any Condorcet method, meets Consistency, and all of
the monotonicity criteria, including Mono-Raise, Participation,
Mono-Add-Top, and Mono-Add-Unique-Top. These must be regarded as
"translation criteria".

Condorcet methods (all of the that I'm aware of) fail all of the
above-listed criteria except for Mono-Raise. Instant Runoff fails
Mono-Raise, but doesn't fail Mono-Add-Unique-Top, though Condorcet
fails that criterion.

Traditional Unimproved Condorcet (TUC), under current conditions, does
poorly by "accurate translation", due to its preference-distorting
strategy needs.

Under Green scenario conditions, TUC isn't as bad, because, as I said,
its FBC failure won't matter as much. But its CD failure will still
matter, and would reduce TUC's value there too, by compromising any
MMC compliance that a TUC method has--as is the case with Beatpath.

Symmetrical ICT, under current conditions, at least doesn't have
favorite-burial need, or a chicken dilemma. But, under current
conditions, Symmetrical ICT will still give strategic incentive to
top-rank all of the acceptable candidates (TUC's strategy in that
regard isn't even known, and would have to be guessed at by voters).
So Symmetrical ICT isn't claimed to encourage completely sincere
rankings.

Under Green scenario conditions, due to that method's MMC failure,
there would still be incentive to top-rate all the acceptable
candidates, even within a mutual majority.

Under current Conditions, IRV's FBC failure, a particularly flagrant
one, quite disqualifies IRV.

Under Green scenario conditions, if people want strategy-freeness, an
easy, strategy-free choice among
their preferred set of candidates, IRV is the very best there is.

But I hasten to advise that, if you aren't sure that you'll be in a
mutual majority,
then you definitely should oppose IRV.

Speaking for myself, I _don't_ need easy, strategy-free choice among
the set of candidates that I prefer, who, I expect,
would likely be a mutual-majority-preferred set. In Approval, I'd
approve all of them, and wouldn't consider there to be a need
to choose among them. Therefore, I'd prefer Approval to IRV, unless I
were sure that the progressives would be
a mutual majority that is voted as such. I've written to advise the
Greens regarding the considerations for judging the desirability
of IRV. That's what my "opposition" to IRV consists of. (Above, I said
that you should oppose IRV if you aren't sure that you'll
be in a voted mutual majority.)


IRV's big advantages result from meeting the following criteria:

1. Mutual Majority Criterion (MMC)
2. Later-No-Harm (LNHa)
3. Clone-Independence
4. Later-No-Help (LNHe)

The first two of those criteria are particularly powerful in
combination. They ensure that there'd be no need for other than
sincere ranking, in a mutual majority. Therefore, the winner would
come from that mutual majority's preferred-set. IRV's freedom to rank
sincerely, for people in a mutual majority, is unmatched. Well, I
shouldn't say that no other method achieves that--it's just that I'm not
aware of one.

IRV's disadvantages:

Though IRV strongly rewards and favors cohesiveness and mutuality, IRV
achieves that by strongly penalizing and disfavoring un-cohesiveness
and non-mutuality.

The freedom from strategy need, for a mutual majority, is achieved by
dumping favorite-burial need on those not in a mutual majority.

That's due to IRV's FBC failure, and its flagrant Condorcet Criterion failure.

That's a good reason to oppose IRV if you might not be in a voted
mutual majority. But how wrong, unfair or undemocratic is it?

A group that has a majority against it can't expect to win in any
method. Then maybe it isn't too unfair if non-mutual-majority voters
also don't have the same strategy advantages.

Isn't a majority who like eachother's candidates the most important
kind? There's nothing wrong or unfair about government by a cohesive
majority.

Objectively, it sounds fine. As an individual or a faction, whether
you should support or oppose it depends on being sure you're in that
cohesive majority. But, then, wouldn't most people expect that they'd
be in a mutual majority?


[quote]
the best indicator is actual evidence taken from voters actually ranking
or rating candidates
[endquote]

...But not meaningfully from organizational elections, where the
situation is nothing like official public political elections.

[quote]
 -- and possibly from counterfactual reasoning of
the form "if there was a lot of strategy going on here, then results
would be like X, but we know results are Y, so there wasn't".
[endquote]

American voters are explicit about that. They engage in strategic
favorite-burial, as their media advise them to. As I said, given their
expressed beliefs and assumptions, favorite burial is their optimal
strategy in Beatpath, Ranked-Pairs, etc.

[quote]
I think that the evidence shows there's not an undue amount of strategy
by the voters
[endquote]

How regrettable that Kristofer forgot to specify the evidence :-)

American voters make it quite clear that they use favorite-burial
strategy, as their media advise them to do (as I said above). And, as
I also said above, given their expressed beliefs and assumptions,
favorite burial would likewise be the optimal strategy in Beatpath,
Ranked-Pairs, Kemeny, etc.

I've told why, many times. I discussed it in a recent posting,
probably in the "Acronym use" thread.

[quote]
, while there may be a greater amount of strategy by
(comparatively more organized) parties. Thus I think that the method
should deter party strategy and be accurate with respect to voters'
wishes.
[endquote]

Accuracy with respect to voters' wishes requires a voting system that
doesn't give them strategy needs that will drastically distort their
preferences. That disqualifies Beatpath, Ranked-Pairs, etc.

Symmetrical ICT would do much better, under current conditions, which
is what it's intended for...if a Condorcet method could be enactable
:-)

As mentioned above, Approval, by counting likedness and acceptability,
accurately gives what voters collectively most want, like or find
acceptable.

But, if it's sincere rankings you want, then you want IRV in the Green
scenario. IRV doesn't must meet MMC. Due to compliance with LNHa, and
therefore CD, it also allows sincere rankings among a mutual majority.
That can't be said of Beatpath or Ranked-Pairs.


[quote]
- Majority criterion: a majority's wishes on which candidate is elected
is respected
[endquote]

That's vague. Is Kristofer referring to voted majority or preference majority?

Voted Majority Criterion:

If a majority of the voters vote X over everyone else, then X should win.

[end of Voted Majority Criterion]

Preference Majority Criterion:

If a majority of the voters prefer X to everyone else, and vote
sincerely, then X should win.

[eend of preference majority criterion definition]]

Voted Majority is obviously a special case of MMC.

[quote]
- Mutual majority criterion: a majority's wishes on which set of
candidates the winner comes from is respected
[endquote]

Likewise vague:

MMC:

If a certain majority of the voters all prefer a certain same set of
candidates, S, to all of the other candidates, then the winner should
come from S.

[end of MMC definition]


[quote]
- Condorcet criterion: if there's a candidate that would win a runoff
against every other candidate, ballots unchanged, then that candidate
should win
[endquote]

I won't keep commenting on inadequate definitions. But I'll point out
that Beatpath's Condorcet Criterion compliance, under current
conditions, is made meaningless by its failure of FBC, for the reasons
that I discussed in more detail in the "Acronym use' thread, and in
many previous postings. Likewise, Beatpath's MMC compliance is
compromised by its CD failure, under current conditions, and in the
Green scenario too.

Evidently, not everyone understands that a method's compliance with
one criterion can be made meaningless by its failure of another
criterion.


[quote]
- Smith criterion: if there's a small set of candidates, any of which
would win a runoff against any candidate outside the set, with ballots
unchanged, then the winner should come from that set.
[endquote]

Again, terrible definition. But what makes Smith compelling & necessary?

Anyone can list a bunch of criteria. It would be quite another thing
to actually tell _why_ that person thinks that their listed criteria
are important.

[quote]
- Independence of clones criterion: turning a candidate or party into a
bunch of identical candidates or parties (that all voters rank next to
each other) shouldn't alter the outcome.
[endquote]

Desirable, but not necessary, for reasons that I've given here, and at
Democracy Chronicles. Beatpath passes, IRV passes, Symmetrical ICT
fails.  But,, as I described then,  failure only results in an
Approval-like strategy situation. And Approval strategy is the best
that can be achieved by rank methods anyway, under current conditions,
especially in a u/a election.

As is often pointed out, different people advocate different criteria.
At EM, of course we've always heard that going on. What's missing is
justifications for that advocacy.

I've told why FBC and CD are important.

I've told the big advantage of the combination of MMC and LNHa, under
Green scenario conditions.

Michael Ossipoff