[EM] wv autodeterence obtains with bigger faction-size differences too.

[Michael Ossipoff]

On Sat, May 4, 2024 at 02:41 Chris Benham <cbenhamau at yahoo.com.au> wrote:

> Mike O.,
>
> One or two examples might be interesting.
>

Of course, & In the morning (or early afternoon) I’ll post some examples
from the 48-case test.

It’s late ⏰ in my timezone. The only reason I’m up now is because I
stayed-up inadvisably-late. But tomorrow I’ll post those examples.

>
> Say there is a defensive strategy that those who prefer the sincere CW
> to the favourite of would-be buriers can employ and it will be effective
> if all of them use it.


Yes. There is, because the wv Condorcet methods meet Minimal-Defense.

Don’t rank anyone you don’t like. If the CW’s voters don’t, then no one
that they don’t like will be able to take-away the CW’s win via burial (…&
offensive truncation by one faction won’t work anyway.) …& if anyone
attempts such a burial, it will backfire.

I think it would be interesting to know how well this works in different
> methods when those voters don't all use it. In other words, for some
> given method, how high does the proportion of voters who prefer the
> sincere CW to the BF using the defensive strategy have to be for it to
> work?


I don’t know the details, but yes, it’s definitely a problem….or would be,
without autodeterence.

It’s the reason why the wv Condorcet methods’ autodeterence is so important.

Without autodeterence, if I can’t count on enough CW voters using
defensive-truncation, then they might let an Unacceptable win by burial.

I consider the merit-difference between Acceptables to be incomparably
greater than any merit differences *among* the Acceptables or among the
Unacceptables. Therefore not electing an Unacceptable is, for me, is what
matters in the election.

That means *fully* & *maximally* protecting the Acceptables.

So, just in case the CW voters don’t protect their candidate, I must try my
best to.

To do my best to not help buriers make an Acceptable pairbeat the CW, I
must not rank any of them over the likely CW. I must rank all of the
Acceptables together in 1st place.

But what if insufficiently many people do that? It will be ineffective. So,
again, I can’t count on that.

So I must, instead, rank the Acceptables in order of how likely they are to
be the CW. …so that I’m doing my best to try to *actively-oppose* any
burial-attempt.

That likely will mean burying my favorite.

(In fact, of course it’s even worse with a Condorcet method that doesn’t
even meet Minimal-Deterrence at all—which is why I say that such methods
share IRV’s serious FBC-failure…for very similar reason.)

But, with autodeterence, even without defensive-strategy use, burial is
well-deterred.

…& so no defensive strategy is really needed. We have no need to do other
than rank sincerely. That fulfills the promise of Condorcet.

 …& the ideal of rank-methods.

…&, in fact, the very ideal of perfection for voting-systems.

I’m not saying that I wouldn’t do defensive-truncation anyway—as part of
*maximal* defense of the acceptables.

But, if I really like some Acceptables better than others, I wouldn’t
consider it necessary to not rank them in order of preference. Due to the
autodeterence, it would be safe to benefit from the full Condorcet promise.

…if I feel sufficiently strongly about the merit-differences among the
Acceptables.

I should have mentioned that, not only might the CW-voters not protect hir,
but, also, their idea of what they like or is acceptable might be different
from mine—like when Jill’s other preferrers vote for Joe. :-(





>
> Chris B.
>
> On 4/05/2024 7:04 am, Michael Ossipoff wrote:
> > Chris—
> >
> > As you suggested, I tested wv Condorcet with greater faction-size
> > disparity.
> >
> > Instead of placing the faction-sizes as closely as possible, this time
> > I spaced them from eachother by a factor of 3/2.
> >
> > That’s about as much disparity as possible without giving someone a
> > majority.
> >
> > The relative faction-sizes were 4, 6, & 9.
> >
> > Of course that could mean 40, 60, & 90…etc.
> >
> > The result was the same: Burial’ backfire was 10 times more likely
> > than its success.
> >
> > Nothing like that has been said for any other Condorcet-complying method.
> >
> > ----
> > Election-Methods mailing list - see https://electorama.com/em for list
> info
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20240504/27302dff/attachment.htm>