Monday, December 20, 2004

It's not instant runoff

In the comments regarding yesterday's election post, it was stated that my made-up election method was similar to instant runoff voting, and also that my idea made me sound really ignorant. I don't think that it's like instant runoff voting at all. I took a look at an article titled "The Problem with Instant Runoff Voting," and my plan does not appear to have the same fundamental problem as IRV.

The problem with IRV is that, as soon as one of the third-party candidates becomes better than one of the top two, the balance of power shifts to the "wrong" candidate, making it just as bad as a regular election. However, this wouldn't happen with my plan. In my example, there is a new party called the "Perublicans." The Perublicans are a lot like Republicans, so anyone who voted for one would be almost as happy voting for the other; certainly happier than if the Democrats won. Since the Republican and Perublican candidates were so similar, the votes between the two were split pretty evenly. In a normal election, Democrats would then win, even if the Republicans and the Perublicans combined had more votes. In my system, whichever one of those two would win instead of the Democrats, which seems much more fair. Here's how it works:

Republican: 20
Perublican: 30
Democrat: 35
Green: 10
Perot: 5
TOTAL: 100

First off is Perot. He's stubborn and doesn't like the other parties, so he just deletes his votes, so we've got the same numbers as above but with a total of 95. Now Green is the losing party. The Green party candidate likes the Democratic candidate best out of the ones left, so he transfers his votes to the Democrats. That leaves us with this:

Republican: 20
Perublican: 30
Democrat: 45
TOTAL: 95

Now the losing party is the Republicans. Clearly, the Republicans would want to give their votes to the party most friendly to their interests, the Perublicans. So, then we have:

Perublican: 50
Democrat: 45
TOTAL: 95

With just two left, the Democrats have no choice but to either forfeit their votes or give them to the Perublicans. Clearly a Democratic candidate would never vote for the enemy, so we end up with:

Perublican: 50
TOTAL: 50

Thus, the election pans out in what I feel is the most logical and fair way possible. Instead of the Democrats winning like in the current system of election, which would please 45% of the voters, the Perublicans win, which would please 50% of the voters.

Am I still missing something? I'm not saying that it's bulletproof and that we should all switch today, but I have not yet been convinced that my idea is as retarded as some of you may have me believe.

21 comments:

Travis said...

You may notice that I didn't really talk much about IRV. I decided that I didn't really have a reason to. IRV may or may not be better than "normal" voting, but it certainly seems like it would be much more cumbersome. I think that it's different enough from my idea (which I'm sure other people have already had and probably even has a name) that it wasn't really worth talking about. The point of the example is to show that given a somewhat-plausible scenario, my idea gives more sensible results than normal voting. If I'm not mistaken, it would have also worked well in the past two presidential elections.

My idea turns the candidates themselves into the electoral college in a single-elimination fashion. That seems quite different from each voter ranking the candidates.

Travis said...

It bothers me a lot that the URL for this post doesn't have an apostrophe, regardless of whether or not an apostrophe is a valid character for a URL.

stack said...

though i haven't thought about it very much, and even though IRV is not the focus of this post, i must say this: i read the article about the "problems" with IRV before reading any further in this post, and it seems like a bunch of bullshit. it's poorly written, and seems HIDEOUSLY biased in some way.

like i said, i haven't thought about it very much. now for the real part of the post.

stack said...

Condorcet voting, on the other hand, looks quite appetizing.

now, as i said before, on to the real meat of the post.

stack said...

after thinking for a bit, IRV does kinda suck.

Travis said...

Yeah, Condorcet voting looked like it might make more logical sense. But, let's be realistic. There's no way that the US would EVER switch to that. People understand regular voting. People could understand my way. Most people could understand IRV. But I think that relatively very few people would be able to wrap their heads around CV.

Oh, and sorry about the typo in my example. (I wrote it correctly on the whiteboard.) The numbers make more sense now.

stack said...

Actually, on second thought, the "flaw" they describe in IRV doesn't seem logical, for reasons I don't feel like describing here.

Grr...must go with initial instincts always.

stack said...

I think your system is exactly like IRV, except asshole candidates can swing the election against the favor of the voting public. Suppose 100% of Perot's voters voted Democratic (unlikely, yes). Assume the same for the Green party candidate. Now let's change the numbers a bit:

Republican: 20
Perublican: 29
Democrat: 35
Green: 10
Perot: 6
TOTAL: 100

Since Ross Perot is a fuckhead, the Perublicans will take the election, against the favor of 51% of the public.

I hate Ross Perot. Don't let him do this.

stack said...

By the way, when I say Perot's voters voted Democratic, I mean Democrat was their second choice. Sorry if that was unclear.

Travis said...

That's by design; I don't see that as a flaw at all. The general rule in many elections is "don't vote for asshole candidates." If you don't trust the candidate to recast your vote appropriately when they inevitably lose, then you shouldn't be voting for them in the first place.

stack said...

Ok, then your design is flawed.

Travis said...

Your mom.

Anonymous said...

While not a retarded idea, or at least not as retarded as the idea of comparing what you are advocating to IRV, there is a fundamental flaw with this idea. It seems to me that you pre-suppose that most voters vote along party lines and would still be pleased to see a "similar" party in office. If you are voting for an individual rather than a party (like my vote for Kerry actually being a vote AGAINST Bush), you loose even more voting power in such a system. Lets say that I everyone who voted republican really disliked the Perublican (as she is too conservative for even them) and black balled her in the primaries, which caused the creation of this party to begin with. These people would generally rather that a Democrat or someone else be elected, but their votes would be completely marginalized when the Republican Party gave it's votes to the other party.

It seems to me that what you are attempting to get rid of is party politics in order to allow for the creation of a viable 3rd party (a worthwhile thing), but this system simply reinforces the notion of party politicking by giving the power to transfer the votes (I assume anyway) to the parties. Solvency is only met here if you give the reassignment of votes to the general public leading to lots of election iterations - smacks way to much of man mythical months to me.

stack said...

Does Meestar Anonymous have anything to say about Condorcet voting?

stack said...
This comment has been removed by a blog administrator.
stack said...

Also, in response to the most recent anonymous comment: would it not be reasonable to assume that if we are transferring the Republican votes to the Perublican party, those who find the Perublican offensive are a minority among the Republican voters? This, of course, assumes a candidate that agrees with the mainstream voters of his/her party.

Anonymous said...

That is the problem, in such a system (short of multiple iterations for assignment of "lost votes") there is no way to tell if people favoring a Democrat over a Prepublican are majority or minority. Which I suspect is the point you wanted me to make, as a Condorcet voting system would eliminate this problem.

I'm not endorsing one over the other, I was merely pointing out a possible flaw with the proposed system based on the self contained meterics presented to judge it's usefulness.

Travis said...

I agree that that situation is a potential problem. However, I don't see it as a design flaw of my plan. I think that anything that solves the problem will be significantly more complicated than my method, which may even require LESS work on the part of the voters, lifting some of the burden of voting "strategically" (i.e. never voting for a third candidate). At the very least, it doesn't require anybody to learn a new way to vote. Moving to a ranking system would not be difficult to understand, but would probably making voting more unpleasant. Moving to a crazy Condo-matrix is almost certainly out of the question... I mean, these people couldn't even figure out butterfly ballots.

Travis said...

Well, there we go. I'm glad that there are finally comments with actual content now. :)

I can also say that, in theory, I agree with the issues that have recently been raised with my idea. My response to the Perublicans being so conservative that it turned the stomachs of even the Republicans is that it doesn't seem that likely to me. If the parties were so radically different, I don't think that the Republican candidate would transfer his votes to them. They would just throw their votes away, and the Democrats would win. It's true that my system does not tell you which secondary candidate the Republican voters prefer. My idea was that this would be decided before the election. Specifically, that the candidate would generally choose beforehand, and thus people would know who their candidate chose as a fallback. They could still follow the current logic of voting Democrat if they wanted to be REALLY sure not to elect a Perublican.

I think my idea is still worthwhile on the basis of its simplicity and the fact that it seems like an idea that the general public might, in some distant future, actually go for. I'm not sure that I could see something as "weird" as Condorcet voting ever happening anywhere, except perhaps when a race of superhuman robots form their own political system.

I do not contend that my from-the-ass idea is flawless or the most fair by any means. I do think that it's an improvement over the current system that increases fairness without increasing complexity for the voting public. Thanks to those who (finally :) provided interesting feedback.

stack said...

Condorcet voting merely requires voters to rank their choices. No wacky Condo-matrices for the voters to haive!

Travis said...

Oh, okay. I thought that the voters themselves had to set up the matrices. Clearly I did not look into the idea too clearly. That's much better, though I still don't think that too many people would choose to go into that kind of system... they'd have to be forced. But, I've already made that point.