[EM] Independence from clones, and Condorcet//FPP criterion failures

[Kristofer Munsterhjelm]

On 10/27/2012 07:25 AM, Michael Ossipoff wrote:
> Kristofer:
>
> You said:
>
> On to some criterion failures:
>
> If there is no truncation or no equal-rank, then (I've been told) ICT
> is equal to Condorcet//FPP. Hence, finding an example where
> Condorcet//FPP fails independence from clones with no equal-rank or
> truncation suffices:
>
> [endquote]
>
> Correct.

I was going to write a point-by-point response to all of this, but then
I reasoned that it would lead to one of those really long threads that
I've no interest in starting. Therefore, let me be brief.

> The Condorcet Criterion, Condorcet Loser, MMC, Smith, and Schwartz
> lose their meaning and value when voters experience the
> preference-distorting strategy needs of methods that fail FBC and CD

This is the crux of our disagreement. I disagree that a method that 
fails FBC and CD leads to a garbage in, garbage out situation where 
every voter is so afraid of having the greater evil win that he votes 
for the lesser evil.

If one expects the voters to be mostly honest, then FBC and CD is no big 
deal. The criteria above then have meaning, and support the methods that 
pass them over ICT.

What evidence do I have that the voters are mostly honest? They were in 
Burlington in both elections, even though the honesty didn't pay. Had 
there been lesser-evil voting of the scale you seem to imagine, I doubt 
we could have detected the close three-way race there in 2009. There are 
other examples, too, like organizations using Condorcet methods and not 
repealing them, MJ exit polls, and multiwinner methods results.

To be more specific about the latter, in countries with party list, it's 
tactically better to vote for a party close to the quota than to vote 
for your favorite, but voters in those countries rarely (if ever) 
complain about it. STV in New York also provided local multipartyism 
right after the switch, no acclimation period needed, even though STV 
fails FBC.

All considered, I think that evidence is of a little greater weight than 
your "voter voting Democratic above third-party in a web poll" sample.

(I can define "mostly honest" if you'd like.)

Furthermore, let me say that an objection to a criterion failure 
claiming the failure can be bypassed if only the voters keep it in 
mind... it does little to convince me. By having the voters act as 
manual DSV systems to patch up problems with the method, you burden the 
voters, whereas a method that passes independence from clones outright 
doesn't make such demands. Also, by expecting the voters to patch the 
problem by equal-ranking, you introduce something quite similar to the 
Nader, Bush, Gore problem in Approval. The A-voters in my example have 
to judge whether A is secure enough that they can vote their conscience 
and say A > A2, or whether the other candidates are dangerous enough 
that they have to vote A = A2.

If you want the voters to act as manual DSV systems, I have a better 
suggestion. Actual DSV. "Never send a human to do a machine's job", if I 
may be cute. But of course, then the inherent limitations of or risk for 
misjudgment in the strategies or patches used may show up in the form of 
failure of yet other criteria of that DSV method, instead of being 
concealed between the voter's intent and the actual method to which the 
patches are applied.

And by extending the concept of patches just a little, to outside the 
system itself, even Plurality can be patched to pass independence from 
clones. It's called an exhaustive-runoff primary.

If you want a method that passes independence from clones, why not look 
for one? I never claimed CD and independence from clones to be 
inherently incompatible.

And yes, there does exist a Condorcet method that passes Mono-Add-Top. 
It's even advocated by a member of this list. So please do your research 
before claiming others are fitting their criterion selections to the 
method rather than vice versa.

Speaking of which:

> TUC advocates forgive failure of FBC, CD, LNHe, 0-info LNHe,
> Participation, Consistency and IIAC, but they draw the line at
> Mono-Add-Plump and Plurality (complied with by Approval, Score, ICT
> and Symmetrical ICT, along with TUC), and Reversal Symmetry and
> Clone Independence (at least if their favorite TUC version passes
> them). Could it be that we're being a bit self-servingly selective?

If we're talking about self-servingly selective, I notice that in regard 
to your original (pre-fix) "improved, with power truncation" way of 
counting top and bottom rank, you said "That's why, in keeping with what 
the voter would prefer and wishes with hir equal top and equal bottom 
rankings, Symmetrical ICT interprets equal top and bottom ranking as it 
does". You also spoke quite warmly about (unqualified) LNHelp. Then you 
find out that keeping power truncation would lead to CD failure, and 
oops! Out it goes. So much for "the voter's wishes" of how his ballot is 
interpreted. So much for plain LNHe, which was so desirable up until 
that point. It's now okay to weaken both.

But we can keep up with these "perhaps your intentions are X" claims all 
day. I could say that you're being awfully insistent about this 
groupthink claim of yours, so perhaps you are projecting, or perhaps 
you're so set in the simplicity of Approval that you must disarm any 
possible contender to its quality because that would weaken the reform 
movement. Then you could, in turn, claim that we're all trapped by our 
groupthink and I'm just going on the defense. What good would come of 
it? We'd just both get more annoyed. Let's keep to what can actually be 
checked, as I'm getting really close to following RBJ's plonk.

-

Oh, and about the value of independence from clones: it lets voters know 
there aren't going to be any split-votes problem from similar 
candidates. You know, like what happened in South Korea, or in 2000 in 
the US. It'll also assure the voters that powerful parties can't run a 
bunch of copies to win. Kinda like FBC lets the voters know how they 
don't have to vote anyone above their favorites. And that will happen 
without the voters having to play the manual DSV roulette.

(And please don't say that slight violations of independence from clones 
don't matter while slight violations from FBC do. Or if that's your 
preference, don't expect others to share it.)