[EM] The weak-preferring clause, of Conditional MMV

[Michael Ossipoff]

 In my definition of Conditional MMV, I added a clause to the end of
paragraph 3, which isn't in Unconditional MMV. I'll call that clause
the "weak-preferrng clause".

I can't really justify having that clause in paragraph 3 if it isn't
in all 3 paragraphs. If it's right in paragraph 3, then it could be
expected to be right in paragraphs 1 and 2. If it's wrong in
paragraphs1 and 2, then it could be expected to be wrong in paragraph
3. Its rightness or wrongness seems the same in all 3 paragraphs.

So I'd like to replace my previous Conditional MMV definition with two
other Conditional MMV definitions:

Conditional MMV:

The weak-preferring clause isn't present in any of the paragraphs.

Conditional MMV, weak preferring:

The weak-prefering clause is present in all 3 of the pararaphs.

So, then, I define 3 versions of MMV:

1. Unconditional MMV
2. Conditional MMV
3. Conditional MMV, weak-preferring

It was probably the weak-preferring clause that led Markus to comment
on similarity to CSSD. It's true that Conditional MMV, weak-preferring
loses much of the character of RP, and acts somewhat more CSSD-like.
It would be nice, or at least interesting, to find that Conditional
MMV, weak-preferring is equivalent to CSSD, but I don't suppose that
there's any reason to expect that, since there's no mention of the
current Schwartz set.

I don't know what Conditional MMV, weak-preferring's properties would
be like, but it seems to lose much of its RP character. Maybe losing
RP advantagers, an maybe not gaining any advantages in trade. Not
knowing anything about what Conditional MMV, weak-preferring would be
like, I don't advocate it.

For deterministic MMV, I suggest Conditional MMV and Unconditional
MMV, without any suggestion about which is better than the other.

I expect that MAM is probably better than MMV, because, despite its
randomness, it seems to act more like RP would if there were no equal
defeats.

As for Conditional MMV vs Unconditional MMV, it's difficult to say.

You could say that, in my A = (B>C>D>B) example, there's dynamism in
solving the cycle and returning a 2-alternative tie, because each of
{BCD} is better than one of the others in some way. A is equal to all
of them. Safe vs dynamic?

But if the electorate aren't stuck in a conservative rut, is that
method-caused dynamism necessary?

You could say that all of {B,C,D} are doubteful, because each is worse
than another in some way.

Conditional MMV responds only to the more reliable, consistent and
simple message from the voters, in comparison to Unconditional MMV.
Conditional MMV is also more decisive.

The same can be said for SSD vs CSSD.

I've written to Prabhakar, to propose MMV as I define it. If Prabhakar
wants to keep the MMV definition that is currently at his website,
then of courseI can't use that term for the method (in 3 versions)
that I define.

If that turns out to be so, then I'll have to re-name the method that
I've been calling my MMV. If that turns out to be so, then, at least
as a first naming, I'll call my deterministic RP method Hierarchial
Deterministic Ranked-Pairs (HDRP).

...in 3 versions: Unconditional, Conditional, and Conditional, weak-preferring.

I'd say that I prefer Conditional to Unconditional, even thugh it
fails Clone-Independence and has more words in its definition. As I
said, I don't propose Conditional, weak-preferring, because I don't
know anything about its properties, except that it departs a lot from
what we expect RP to be like. Who knows, it might be really good, but
I don't know.

Michael Ossipoff