[EM] Smith, FPP fails Minimal Defense and Clone-Winner

[Kristofer Munsterhjelm]

Juho wrote:
> On Mar 10, 2010, at 7:08 AM, robert bristow-johnson wrote:

>> so, keeping RP, Schulze in mind for later, what would be a "good" 
>> scheme for resolving cycles by use of elimination of candidates?  what 
>> would be a "good" (that is resistant to more anomalies) and simple 
>> method to identify the "weakest candidate" (in the Smith set) to 
>> eliminate and run the beats-all tabulation again? i'm not saying 
>> elimination is a good way to do it, but it might be easier to sell to 
>> neanderthal voters.

> All methods have some weaknesses that can be used against them in the 
> marketing battle. So maybe the challenge is to build a solid campaign 
> that is convincing enough to neutralize all emerging fears and failure 
> examples.
> 
> Schulze method is quite convincing since it is already in use in many 
> places. (I'm not sure it is objectively better than ranked pairs and 
> others although it is certainly a good method.) Ranked pairs can be 
> easily described as a method that sequentially confirms pairwise wins, 
> so it is almost as natural and straight forward sequential algorithm 
> based as IRV (although it is a bit more complex in the sense that it 
> compares pairs instead of single candidates when making the sequential 
> decisions). And there are also simpler Condorcet methods like 
> minmax(margins) that simply counts the number of additional votes each 
> candidate would need to beat all others.

For Schulze, I think the strategy should make use of its current 
popularity (relative to other Condorcet methods, at least). Maybe make 
Condorcet programs and patches to widen its use among those who know 
computers, and encourage organizations to use Schulze for their internal 
decisions (like the Pirate Party did with its primaries, although that's 
not quite the ideal usage of the system).

> I'm not aware of any sequential candidate elimination based method that 
> I'd be happy to recommend. One can however describe e.g. minmax(margins) 
> in that way. Eliminate the candidate that is worst in the sense that it 
> would need most additional votes to win others, then the next etc. In 
> the elimination process one would consider also losses to candidates 
> that have already been eliminated (I wonder if this approach makes it 
> less "natural looking" than the elimination process of IRV).

To my knowledge, Schulze-elimination is the same as basic Schulze. In 
other words, if you run Schulze, eliminate the loser, run it again, etc, 
you end up with the original result. That's not very useful, but still...

It might also be that any "full-blown candidate elimination method" (you 
run the election as if the one that was eliminated never stood) with a 
weighted positional base method (Borda, Plurality, ...) is nonmonotonic. 
I can't prove it though!

> But as said, maybe the key is to arrange a solid campaign. You can 
> surely e.g. find lots of election method experts that are happy to agree 
> that Condorcet methods are the best for some city for some need and best 
> single winner methods (for competitive elections) in general. But quite 
> certainly there will be also experts that think otherwise. And there is 
> certainly a risk that all the Condorcet friendly experts will use 
> different argumentation, may disagree on details and as a result will 
> confuse the audience. Maybe one should start from Scientific American 
> etc. to first firmly establish the idea that Condorcet methods indeed 
> are the de facto state of art methods (and practical too). IRV campaigns 
> have been successful, so I wonder why Condorcet campaigns could not 
> follow (maybe Burlington needs a timeout now but not much more, and 
> other cities could take steps forward already now and support Burlington 
> that way).

The first step would be to say: okay, there are many Condorcet methods, 
but they differ in fine tuning. For public elections, good enough is 
good enough, and let's pick one that's good, then unify around it. That 
method could be RP or it could be Schulze, or something else (I think 
independence from Pareto-dominated alternatives would be nice, but at 
some point we'll just have to say "good enough").

I mention RP and Schulze in particular because RP is easy to explain 
(relatively, given the criteria it passes), and Schulze has some record 
of use.
So does, say, Copeland, but in my opinion, it isn't "good enough". The 
iterated Copeland(2,1) version may be better, but probably isn't 
cloneproof.. and so on.

> (I have to add that if people want to keep the USA as it mostly is, a 
> two party based system, then I must recommend FPTP :-). And if not, then 
> maybe also some additional (maybe proportionality related) reforms are 
> needed.)

Wouldn't something like Condorcet multiwinner districts be better? Pick 
a good Condorcet method and send the 5 first ranked on its social 
ordering to the legislature. That would pick a bunch of centrists (thus 
have "stability"), but it would pick the centrists people actually wanted.

Hm, that might not provide a true two-party system, though. One could 
also have a "PR" system where the number of votes is weighted so that 
parties with broad support gain superproportional power, but then the 
question becomes why one should bother with the PR at all.