Jedi Day

[gfish]

May the Fourth be with you!

I get the stupidest grin whenever I say that. And I've been saying it a lot today.

Also, watching a room filled with AI profs and grads discuss the Single Transferable Vote referendum here in BC was much fun. Encouragingly, they were almost unanimously in favor of it, they just had a lot of interesting implications and boundary conditions to pick apart. I don't think anyone with BC voting privileges who reads this journal isn't already a supporter, but in case I'm missing someone, please let me proselytize at you in person! It's killing me that I can't vote for it myself.

Comments

[ashley-y.livejournal.com]

Meh. STV is only good for eliminating the spoiler effect of a minor candidate. Once you get three or more major candidates, STV is not so good. Condorcet methods are better.

Also STV has to be tabulated centrally, while Condorcet methods typically can be tabulated at precincts.

[gfish.livejournal.com]

Are you sure you're thinking of STV? It's a multiple-winner proportional system, with more than one representative being chosen from each riding.

(Anyway the gap between first-past-the-post and just about ANY modern alternative is so vast that arguing about the distinctions seems kind of trivial to me. I'll gladly take the 95% gain and let some future generation tweak out that last 5%. Letting the perfect be the enemy of the good when perfection is mathematically impossible seems kind of silly.)

True about it having to be tabulated centrally, though. I guess that just doesn't bother me much -- having taken part in precinct tabulation, I can promise you there isn't anything magical about it.

[ashley-y.livejournal.com]

Oh yeah, I was assuming you meant a single-winner system. Single-winner STV is the same as IRV.

I don't see a big improvement of IRV over first past the post. I think Approval would be better. IRV tends towards candidates with intense support among the few, whereas Condorcet and Approval favour candidates with broad support, the "everybody's second choice" candidates (that are the first to be eliminated by IRV).

[anansi133.livejournal.com]

The ballots for all of these methods are the same, right? I have to wonder if the results were tabulated, and *all* these methods were applied to the results, if the winners would change any... Or how hard would Murphy have to work in order to generate a controversial result?

(yes, I know we'd all have to agree on the same method before trying it, just the differences aren't all that obvious to me.)

[ashley-y.livejournal.com]

No, Plurality (the usual system) is to pick one candidate, Approval is yes or no on each candidate, while for IRV and Condorcet methods, the voter ranks the candidates.

Part of the appeal of Approval is that it's easy to adapt from Plurality. It's just Plurality except that you can vote for more than one candidate.

[ljtourist.livejournal.com]

I don't think that tending towards centrist candidates is necessarily a strength. But that's just my opinion.

I used to like IRV. At the time, I did not believe that its Monotonicity and Participation failures would happen in real-world scenarios. It has since been demonstrated to me that this is not true.

Condorcet, OTOH, is simply awful. It fails later-no-harm (adding preferences can hurt you), which is bad enough; it also fails Participation (voting can hurt you), and it does so spectacularly, in a way that IRV doesn't (called failing mono-add-top). Let's say that your top choice (candidate A) is winning in an election that you haven't yet voted in. Then you vote, with A as your top choice. Whoops! Now A might lose!

Approval is... not bad. It also fails later-no-harm, which is why I suspect that if it were adopted in the US then within a few elections pretty much everyone would end up bullet-voting rendering the system useless(just like what happened with Bucklin voting in the early 1900s).

I've done a lot of looking at ranked vote systems, and if I had to pick one for the US, it'd be Descending Solid Coalitions. Its two major weaknesses are that 1) it has a weird edge-case failure mode when one small group of people vote a candidate in first place and every other voter puts that same candidate in last place, and 2) it can be difficult to explain to people how it works. The best short description comes from Antonio Oneala, who said: "[it] assigns you to a team if you prefer every individual in it to every individual outside of it, and then it takes the biggest team within that, then the biggest team within that, etc... until it gets to one person".

I have not done nearly as extensive a survey of rated voting systems, so you're spared that analysis.

Probably more information than you were interested in, but I was Focussed on this problem a while back. :-)

[ashley-y.livejournal.com]

It's not so much tending towards centrist candidates as broad candidates that I consider a strength.

If I understand correctly, if there's a Condorcet winner, then no-one regrets their ballot (and I suspect this is appropriately generalisable to the Smith set). Here's the thing: if there's a Condorcet winner for everyone's sincere votes, I cannot see why that should not be the winner. I am not bothered by failing later-no-harm and Participation provided tactical voting is impractical.

[cow.livejournal.com]

> having taken part in precinct tabulation, I can promise you there isn't anything magical about it.

I really want to argue and say it's awesome, but I just can't bring myself to do so. It's not like I'm signing up to volunteer for the election this month...

[sistawendy.livejournal.com]

vixyish is a rare lady indeed for putting up with this sort of thing. Don't forget that.

[vixyish.livejournal.com]

...what?

[vixyish.livejournal.com]

I'll rephrase that:

This sort of thing is exactly why I married him.

[sistawendy.livejournal.com]

Even better for gfish.