[EM] Approval & MMC, contd.

[Abd ul-Rahman Lomax]

At 02:50 PM 6/6/2013, Michael Ossipoff wrote:
>Mr. Lomax says:
>
>[quote]
>Michael has a bit of misunderstanding floating in his head about this.
>[/quote]
>
>When someone is as vague as Mr. Lomax, no one can be blamed for not
>knowing what he meant.

Ossipoff confuses dicta with substance. What is this "blame" thing, 
anyway? What does blame have to do with whether or not Ossipoff 
understood something, or didn't?

>[quote]
>  I pointed out that definitions have been manipulated with certain
>voting systems criterion, in general, in order to generate "failures."
>And, of course, they can be manipulated in the other direction. This
>was on the Center for Election Science mailing list, which Michael
>just unsubscribed from, saying there were no open issues.
>So he brings one here? Ah, well, to each his own.
>[/quote]
>
>Yeah i decided to answer Mr. Lomax's confusion about Approval & MMC,
>though of course I didn't expect to thereby help his confusion.

While simultaneously proclaiming (on the CES list) that there were no 
issues remaining involving him. Ossipoff has earned the FOS award, 
many times over. Ah, well, has nothing to do with voting systems, but 
Ossipoff is probably the world's foremost proponent, standing, of 
using voting systems criteria, as arbitrarily defined by him, to 
issue definitive judgments on voting systems, so his probity and 
clarity do become issues.

>[quote]
>  He did clearly misunderstand the issue.
>[/quote]
>
>Forgive me if I misunderstood what Mr. Lomax meant :-)

Forgiven. Cause unspecified.

>[quote]
>  First of all, the definition of "sincere vote" was not the issue.
>[/quote]
>
>Sure it was. Mr. Lomax said that I contrived to define sincere voting
>so as to make Approval fail MMC.

No. I did not say that. Denied. Notice that Ossipoff does not quote 
me. And I made the issue clear, I submit, in the post to which 
Ossipoff is here replying, so his concern is what was said before, 
he's defending his error, even though it's really moot who said what when.

Now, looking for a statement like that can be tedious. However, I did 
search and found this:

>>Here's what you're missing. MMC's premise says that a majoriity of 
>>the voters prefer the set S candidates to all the other, and vote sincerelly.
>>In Approval, though you prefer all the Set S candidates to all the 
>>other candidates, you can vote a sincere ballot that doesn't vote 
>>all of the set S candidates over all the other candidates.
>
>Michael just redefined the Mutual Majority Criterion. Essentially, 
>he defined "votes sincerely" in a way that causes the method to fail 
>the Criterion. His definition of MMC differs from that of Woodall's 
>Majority. Watch the ball under the shells:

I explained my meaning in detail. It's possible to interpret the raw 
statement the way that Ossipoff is apparently doing, but the issue 
was not over the definition of "sincere vote," but what *kind* of 
sincere vote is cast. Michael crafted the definition *as used with 
the MMC,* to allow an unexpressed but necessary preference, even 
though the voting system allows the expression of the preference. If 
a preference is not expressed, the system cannot use it. Therefore it 
must fail. Space Alien Failure.

"We preferred A and B over C, but we voted for A and B and C, all 
sincere votes, but that darn system allowed C to win even though we 
preferred A or B."

Big surprise. The original definition of Majority, Woodall, referred 
only to "preference listings," and it is *obvious* that the 
preference must be voted, when we translate the criteria to voting 
systems, or no system could use it.

http://www.votingmatters.org.uk/ISSUE3/P5.HTM

>Majority. If more than half the voters put the same set of 
>candidates (not necessarily in the same order) at the top of their 
>preference listings, then at least one of those candidates should be elected.

Approval allows a preference listing, setwise, a primitive one. Range 
allows deeper preference listing, also setwise. Approval passes 
Majority, Range does not.

What is a "preference listing"? It is a *list*, not some mental preference.

The voter "votes sincerely." *What does the voter vote?* We are 
concerned with a preference for a set, so, I claim, the simplest 
meaning is that the voter votes the set. That is the obvious meaning 
of "prefer."

http://www.merriam-webster.com/dictionary/prefer
>1
>: to promote or advance to a rank or position
>2
>: to like better or best <prefers sports to reading> <prefers to watch TV>

At first glance, it seems like I am insisting on the first 
definition, and, indeed, I'm suggesting that, since no system can use 
an unexpressed preference. By adding the vote for C, the voter 
*unexpressed* the preference for B>C, choosing to do that. The 
expressed preference is no longer exclusive.

However, the second meaning, if we look closely, also requires an 
*action* to be clear. How do we know what someone prefers? If they 
always do something else, how can we say that they prefer it. Such a 
saying would be *hypocritical,* i.e, insincere. They may have a 
"reason" for this. "I preferred B to C, but I also voted for C 
because I wanted D not to win." That is, they *thought* that they 
preferred C, but they actually did not.

And surely they know that by voting for C in addition to A and B, 
they were not preferring A and B over C, by their voting.

>But now Mr. Lomax apparently wants to change what he said, after
>finding that it didn't work.

If I say something, and it's misunderstood, I'm supposed to just say, 
"I said it"?

I *will* repeat this: "The definition of sincere vote is not the issue."

That is because Michael and I agree on what "sincere vote," as a 
general concept, means, see below. There is not just one sincere 
vote. However, what does it mean to "prefer" a candidate set?


>[quote]
>In Approval Voting, there is a large set of possible sincere votes.
>[/quote]
>
>That's right. If some voters prefer A>B>C, and others prefer B>A>C,
>then we all agree that it would be sincere in Approval for the A
>voters to approve only A, and for the B voters to approve only B.
>Therefore with all of the A voters and B voters voting sincerely, C
>can win, even though the A voters and B voters comprise a majority.
>That's why and how Approval fails MMC.

By Michael's definition and application. But Michael does not 
understand that his definition was a *choice* that he made. He then 
takes *his choice* and uses it to make an absolute statement, 
"Approval fails MMC." He added the "sincerely," because it 
legitimizes the failure to express the preference. Without that term, 
would Approval fail MMC? Without it, we'd have the normal avoidance 
of Space Alien Failure as a cause of criterion failure. I find it 
fascinating that by specifying "sincerely," failure becomes *more 
defensible.* Because it then makes it relativelyi explicit that 
failure to express, even though expression is permitted by the 
method, is allowed in setting up a failure.

We didn't need "voting sincerely" in defining the Majority Criterion. 
It's assumed that the vote is sincere and that the preference for the 
candidate is expressed by the majority. (And, by the way, when I 
first had this discussion, it was with James Armytage-Green, and he 
was claiming that Approval failed the ordinary Majority Criterion, by 
an argument very similar to Michael's. Essentially, there can be a 
sincere vote that suppresses the majority preference necessary to 
judge the application of the Majority Criterion.

I'm suggesting that the Mutual Majority Criterion is best 
interpreted, as a baseline, as being a set preference, and as 
requiring expression of that preference, if allowed by the voting 
system. (If not, it can be confusing, I'm not addressing that issue.)

It is always possible in Approval to express the approval of a set, 
fully and exclusively.

>In Approval, those voters have a strategy decision that they need to
>make.

Again, this discussion took place with James. Strategy is not part of 
the definition of the criterion. The motivation of the voters is 
irrelevant. We grant that the vote must be sincere, but I am claiming 
that the test of the criterion -- except as Michael is now defining 
it -- requires more than "sincere," it requires the expression of the 
preference being tested.

Michael is not the only analyst confused about this. Or, to put it 
more charitably, not the only one whose position does not recognize 
Space Alien Failure. No unexpressed preference can be used by a 
voting system. If voters pairwise prefer a candidate over all others, 
a system cannot detect this unless the voters express the preference 
in their votes. If the system allows that expression, and that 
candidate can fail to win, then the system violates the Condorcet 
Criterion. Now, this is interesting: these criteria were developed to 
handle full rankings.

Does Plurality fail the Condorcet Criterion? I'm aware that if I say, 
No, that I can expect incoming Rotten Tomatoes. However, Majority 
passes the criterion, if voters vote sincerely. Majority is one of 
the most common voting systems in the world, rather thoroughly 
neglected by voting systems analysts. Under Majority, if no majority 
is found, the election *fails* and is repeated until a majority is 
found. To avoid the minor problem of Voting Effing Forever (VFFF), 
new nominations and withdrawals are allowed, and the electorate may shift.

>'Should I approve A, in addition to B, to reliably defeat C, or
>should I just approve B, so that I'm doing the best I can in the
>contest between B and A?"

They can worry themselves dizzy, but what does this have to do with 
whether or not the system passes MMC? It's *imagined* that it does, 
because somehow, in the minds of some, a "legitimate reason" to 
suppress a necessary preference makes it "reasonable." And, of course it is.

But the preference was not expressed, and the reason does not matter, 
it could be Space Aliens. They did it.

Michael went on with explanations that completely ignore the issue.

He is arguing that there is some value to defining MMC the way he 
does. I'm not arguing with that. What I was pointing out was that 
voting systems criteria were being manipulated, consciously or 
unconsciously, through shifting definitions.

The strongest point was about "strategic voting," the definition of 
which radically shifted after Approval was proposed by Brams. 
Originally, it was proposed as "strategy-free," and by the 
definitions of strategic voting in usage at the time, involving 
preference reversal, it was. It is never beneficial, in Approval 
Voting, to reverse preference.

>As I said, MMC measures for a genuine tangible advantage, in spite of
>Mr. Lomax's long, unintelligible, and thoroughly-confused ramblings.

Occasionally, by sheer accident, like a monkey typing, I say 
something of value.

>[quote]
>Rather, the issue is about the question of whether or not a vote that
>does not express a preference that is at the foundation of a voting
>system criterion satisfies the precondition for applying the test of
>the criterion.
>Michael, here, did not give the relevant definitions.
>[/quote]
>
>I gave a very thorough definition of MMC, and I've told why MMC
>measures for something of practical importance, offered by Beapath,
>IRV, Woodall, Benham and Schwartz Woodall, but not by Approval.

He didn't give it in the present discussion. So nobody would know 
what we were talking about. I don't write just for myself, I write 
for two purposes: to learn, and to allow others to follow this process.

>That is irritatingly typical. There were two definitions examined:
> From Wikipedia:
>
>[quote]
>  The mutual majority criterion is a criterion used to compare voting
>systems. It is also known as the majority criterion for solid
>coalitions and the generalized majority criterion.
>[/quote]
>
>Incorrect. Wikipedia is mistaken.

According to the Reliable Source, Michael Ossipoff. My point.

>  I introduced MMC. I wrote it, and
>named it. Its definition is as I specified. the Majority Criiterion
>For Solid Coalitions (MFSC) is defined quite differentlyl from MMC.

He didn't add "I copyrighted it, I patented it, I own it, so go away."

But he does have a point.

>[quote]
>The criterion states that if there is a subset S of the candidates,
>such that more than half of the voters strictly prefer every member of
>S to every candidate outside of S, this majority voting sincerely, the
>winner must come from S.
>[/quote]
>
>Here "prefer...to..."  is being used to mean "vote...over...".  Fine,
>just so we agree on what is meant.

Which is weird. To vote every member of the set over every 
non-member, the voters must *exclusively approve only of set 
members." This is the opposite of what Michael has claimed.

I get what Michael means. He wants MMC to specifically detect and 
fail a method because it allows the voter to suppress a preference. I 
am, in general, suspicious of any "failure" that results from an 
increased freedom of the voter. It's a semantic issue.

>[quote]
>This is similar to but stricter than the majority criterion, where the
>requirement applies only to the case that S contains a single
>candidate.
>The Schulze method, ranked pairs, instant-runoff voting, Nanson's
>method, and Bucklin voting pass this criterion.
>[/quote]
>
>So do Approval and Plurality. MFSC doesn't measure for the automatic
>majority-rule advantage that I discussed above. No problem, just so we
>all understand the difference between MFSC and MMC.
>
>[quote]
>The plurality vote, approval voting, range voting, the Borda count,
>and minimax fail this criterion.
>[/quote]
>
>Incorrect. Approval and Plurality pass MFSC.

Michael is coming down on my side of the issue in this case. I get 
this: he is defining a *different criterion*, but the names are so 
similar, and the definitions so similar, that confusion is inevitable.

>For an organization that prides itself on reliability, the wikipedia
>can be quite sloppy and incorrect.

Wikipedia has no conscience, so it can't pride itself on reliability. 
Editors do take pride, and the poliicies have been written based on 
ideals, but editors have no power to ensure reliability, and the 
structure simply does not allow any such ensurance. There is no there, there.

However, where a page has been stable for a time, there is some kind 
of indication that what is on it is generally accepted. No guarantee. 
At one time I did do what I could to clean up the voting systems 
pages, but it's like trying to hold back a landslide with a shovel. A 
small shovel.


>[quote]
>The point I have made is that "strictly prefer" must require that the
>voter actually vote the preference.
>[/quote]
>
>Mr. Lomax wants to make an issue of that, implying that it's something
>that I've missed, or that it's the crux of his disagreemnt with me.
>Actually, I've acknowledged from the start that that definition of
>MFSC is using "prefer...to..." to mean "vote...over...".

Yes. Glad he's making that clear. And he's making clear that he 
insists on a different definition in order to make a point about 
certain other systems. Yet this tends to make *my point*, which is 
that definitions are manipulated to make points.... I'm not saying 
his point is wrong....


>[quote]
>Otherwise we have what I call Space Alien Failure. Under Space Alien
>Failure, Plurality fails the Majority criterion, because the voter,
>being informed by Space Aliens that their favorite cannot win the
>election, vote for someone else. No, the voter must *actually vote the
>preference* to set up the condition for the test.
>Otherwise we are led to a series of preposterous conclusions.
>[/quote]
>
>I won't pretend to know what that means. If Mr. Lomax means that
>"prefer...to..." is being used to mean "vote...over...", in that
>wikipedia definition of MFSC, then he's correct. As for the part about
>Space Aliens, Mr. Lomax isn't being entirely clear with us about what
>he's trying to say.

I wonder if anyone else gets it. Obviously I'm failing to communicate 
with Michael.

Space Alien Failure refers to the failure of a voting system 
criterion because of something hidden, not manifest in the votes, 
based on a strategic or other decision by the voters, which cannot be 
distinguished from "Space Aliens told me to do it." It can be 
difficult to distinguish Space Alien Failure from "strategic 
forcing," and, really, it's a matter of degree.

>[quote]
>The term "stricly prefer" is ambiguous, unless we interpret "prefer"
>as "act to prefer."
>[/quote]
>
>Yes, we've established that the wikipedia definition of MFSC uses
>"prefer...to..." to mean "vote...over...".   :-)

Thanks, Michael. Many have argued otherwise.


>[quote]
>With Approval voting, a voter acts to prefer A to B by voting for A
>and not for B. The voter acts to prefer a set by voting for every
>member of the set and not for every non-member.
>[/quote]
>
>Yes, I suggest that we've established that that's what wikipedia means
>in their MFSC definition, when they refer to preferring A over B, or
>preferring one set to another.
>By "prefer...to....", they mean "vote...over...".   Haven't we already
>established that?  :-)

It was not only established, it's now been repeated. A few times.

>I'd said:
>
>One way to answer his objection is to ask him to compare Approval with
>methods that meet MMC, and ask himself if he notices a difference.
>
>[quote]
>  That's not an answer, it's a socratic question, and it requires me to
>study matters where I have little interest.
>[/quote]
>
>It was a demonstration of why it's better to meet MMC than to fail it.
>It was a demonstration of what it means to meet or fail MMC.
>
>That was the topic on which Mr. Lomax was posting. He shouldn't post
>on what he isn't interested in.

Actually, I was posting on a somewhat different topic. I was posting 
on the topic of the shifting definitions used in applying voting 
systems criteria and concepts originally designed for ranked systems, 
to cardinal systems, or what Arrow calls systems that rank in categories.

>[quote]
>I see no sign that Michael has understood the objection.
>[/quote]
>
>Have I not understood what Mr. Lomax was trying to say? Maybe only Mr.
>Lomax knows what he was trying to say.
>
>But what he "meant" seems to change, when what he previously meant has
>been answered or refuted. That's the nice thing about meanings that
>are vague or variable.

Indeed. I went to school to learn how to do this :-)

The School of How to Be Always Right Even if You Don't Know your Ass 
 From a Hole in the Ground.

I'm always pleased to meet other Graduates. (That part is true. The 
part about Always Being Right is boring. There is no way to learn if 
one is Always Right.)

And I'm stopping here, because I sense declining value.

[Michael went on.]