[EM] Scoring (was Re: OpenSTV 2.1.0 released)

[Juho Laatu]

On 21.9.2012, at 22.52, Michael Ossipoff wrote:

>> Just "in practice". Some more weight on Duverger's law, some less on media (would happen also without media).
> 
> So you keep repeating. But, in this country, the 1-party monopoly
> _wouldn't_ happen without the media fraud that I've discussed.
> 
> You're quite vague and vacillatory about whether you're speaking of
> the U.S. or whether you're speaking in general.

As already said, U.S. meant when "U.S." mentioned. That should give you the correct context with about 95% probability. You can correct me if I make mistakes.

> You said "would happen
> also without media". Where would it happen also without media?

I meant that "parties + plurality + single member distritcs + representative body" has always a tendency to elect from two strong parties. (It is possible that different districts have different two parties, or the parties may change slowly in time.)

>>>> It is hard to find methods that have no weaknesses. Luckily we can often use methods whose weaknesses are weak enough.
>>> 
>>> We can do better. We can avoid certain strategy needs. For instance
>>> there are now a wide variety of FBC-complying methods. They have
>>> absolutely no favorite-burial incentive.
>> 
>> I think it makes often sense to trade one full compatibility to numerous "well enough" compatibilities. As in security, the system is as strong as its weakest link. One should thus focus on making the weakest points stronger, not on making strong points even stronger.
> 
> Your fallacy is your implication that there are other necessary
> properties, lacking in Approval , Score, and Symmetrical ICT, but
> possessed by unimproved Condorcet  ...if unimproved Condorcet is what
> you're suggesting. And if unimproved Condorcet isn't what you're
> suggesting,then what is it that you're suggesting?

I was talking about general rules concerning all methods. No intention to refer to any particular methods.

If you want my opinion on Condorcet methods in general, I think they are remarkably well balanced methods, for compromise seeking, competitive, majority style elections. They thus have quite well balanced "well enoughs" / vulnerabilities.

> If there's something that they don't do well enough, and that
> unimproved Condorcet does well enough then you forgot to tell what it
> is.

What kind of comparisons would you like me to make? If we are talking about the "well enougs" of other than Condorcet methods, then we should focus on the most probable vulnerabilities of each one of them (in the given environment). You already addressed the chicken dilemma in Approval. I think we agree that in real life that may well be the most problematic one. Same with Score.

> You keep making authoritatively-worded
> statements to the effect that unimproved Condorcet would work fine in
> the U.S. Sometimes you insist that you're only speaking in general,
> but then you go back to specifically making that statement about the
> U.S.

I'm trying to avoid anything U.S. specific, except when answering to U.S. specific points that you want to present to me.

Concerning Condorcet in the U.S., I believe the strong two-party and plurality tradition would cause some problems in transition, but I do believe that eventually people would vote pretty much the same way as in other countries (no panic burials etc.).

The only U.S. specific "proof" I have is that in the Burlington IRV elections I didn't notice any burial like or other strategic activity that could be harmful in Condorcet. The votes seemed in general quite sincere to me, and I don't recall any reports on strategic voting. I made some ballot content analysis myself, but I couldn't find any meaningful traces of strategies. I'd expect voters to behave pretty much the same way also with other ranking based methods.

I don't expect e.g. Australian style party guided voting to appear in the U.S. since people have a strong tendency to make their own independent decisions "my way" (probably not in Australia either unless their strangest IRV rules would be copied to Condorcet too).

> You said:
> 
> I may comment your comments on the U.S. system, but when I can, I
> prefer taking about election methods at a general level, not about the
> specifics of individual countries.
> 
> [endquote]
> 
> ...then you often violate your own preference  :-)

The alternative would be to leave your questions / comments unanswered :-).

> I'm merely pointing out that your claims about how Americans would
> vote are lacking in authority.

None of my comments are based on complete understanding of what I'm talking about :-).

Juho