« Democratic Peace | Main | Ah Media Bias »

Redistricting

Gerrymandering is, of course, bad. Mark Schmitt writes about some ways one could go about curbing it. I find that these discussions are often unduly constrained. The other day Will Wilkinson mentioned an intriguing idea. Lets say we just didn't worry so much about where the district boundaries were places. Just leave the current lines in place. Then on Election Day everyone votes for their favorite candidate. But instead of counting the votes and giving the seat to whoever wins the plurality, you just count up the percentages. 58 percent Candidate A, 42 percent Candidate B in one typically not-so-competitive district. Then you spin the computer equivalent of a roulette wheel in which 42 percent of the outcomes have been assigned to the Candidate B and 58 percent of the outcomes to Candidate A. Whichever candidate's lucky number gets called gets the seat.

This method has some drawbacks, which I'll leave to commenters to point out. But it also has some virtues. For one thing, it makes partisan gerrymanders basically pointless. Shifting Democrats from one district to another will have no impact on the expected outcome -- the reduction in Democratic odds in one district will be exactly the same as the increase in their odds in other districts. For another thing, it would eliminate tactical voting considerations. Now this is perhaps unconstitutional -- I'd have to go read the thing -- but it's certainly something that should be out there in the space of conceptual possibility.

May 10, 2005 | Permalink

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8345160fd69e200d83443a87753ef

Listed below are links to weblogs that reference Redistricting:

Comments

Good idea, Matt. I understand that Diebold makes really good roulette wheels.

Posted by: LowLife | May 10, 2005 2:54:45 PM

Darn, That LowLife beat me to it -- Kudos!

Posted by: Dave | May 10, 2005 2:58:39 PM

You have been reading too much Philip K. Dick lately.

Such schemes were proposed in the heyday of game theory. Mathematically they work, and can get closer to "fairness" than straight voting. But they only can be implemented in Cambridge MA, because transparency and understanding are key parts of the voting public accepting the legitimacy of the system. And unless you have an MS in statistics you won't understand why these systems work.

Cranky

Posted by: Cranky Observer | May 10, 2005 2:58:55 PM

It's like a pure democracy with a slim, but existant, chance that the "Zebra on Rollerskate Party" can win!

No thanks - pure Democracy maybe, but leave chance out of it.

Posted by: mozillauser | May 10, 2005 3:01:21 PM

It's not unconstitutional - states can apportion Reps. however they please. Can't remember where I read it (the miasma of the blogosphere, somewheres), but someone a while back was suggesting that, say, California do away with districts entirely (at least as far as vote-counting goes) and just have statewide winner-take-all party slates.

Posted by: jkd | May 10, 2005 3:02:16 PM

Actually it wouldn't be unconstitutional, since Congress gets to decide the time/place/manner of elections, and since this method isn't so arbitrary as to deny due process. But yeah, it has huge problems with inducing acquiescence in its results by the ones who lose.

Posted by: phil | May 10, 2005 3:02:36 PM

It's not unconstitutional - states can apportion Reps. however they please. Can't remember where I read it (the miasma of the blogosphere, somewheres), but someone a while back was suggesting that, say, California do away with districts entirely (at least as far as vote-counting goes) and just have statewide winner-take-all party slates.

Posted by: jkd | May 10, 2005 3:03:07 PM

I have a better idea: they should play russian roulette with the odds opposite of the percent of the vote they received. This could be interesting. Politicians always talk about courage.

Posted by: abb1 | May 10, 2005 3:09:37 PM

Yes, it would make gerrymandering pretty pointless. But it could also make voting seem pretty pointless to a lot of people. What happens when the public realizes that the vote counts don't necessarily decide the winner anymore? I think there would be less voter participation. Now, you could argue that it wouldn't decrease participation any more than gerrymandering already does, but it seems to me like there must be better ways of neutralizing the effects of gerrymandering. Besides, democracy by roulette just sounds bad- not the type of thing you build a movement around.

Posted by: Ricky Barnhart | May 10, 2005 3:10:46 PM

Wilkinson gives credit for this idea where it's due, to my Con Law prof Akhil Amar. Class act, that one.

Posted by: The42ndGuy | May 10, 2005 3:11:53 PM

I would think this would reduce voter turnout even more. It's hard enough convincing people their votes matter when they directly influence the outcome.

Posted by: Rebecca | May 10, 2005 3:19:26 PM

Fascinating! Totally unrealistic, but fascinating.

Posted by: Al | May 10, 2005 3:19:36 PM

the42ndGuy,
Over at Hit and Run a week or so ago I mention him as who I would most like to see Bush nominate for Supreme Court, with the caveat that I did not really know that much about him. Would he be a good one?

Posted by: theCoach | May 10, 2005 3:20:24 PM

If you make the simplifying assumption that all members of a particular party are ideologically equivalent, then gerrymanders would be rendered pointless.

In reality, given that the ideologies of current representatives are known, there would still be motivation for e.g. Republicans to gerrymander away a far-left Democrat representative while making it easy for a centrist Democrat to win reelection.

Posted by: Foo Bar | May 10, 2005 3:25:34 PM

Matthew, you should move to British Columbia. They're looking for ideas like this out there.

Posted by: Finny | May 10, 2005 3:26:56 PM

"I may lose the Preisdency, but I will not raffle for it.

Samuel J. Tilden (from Gore Vidal's 1876)

Posted by: Dantheman | May 10, 2005 3:27:52 PM

the Coach,
You can find out more about him here:

http://islandia.law.yale.edu/amar/

Posted by: BenW | May 10, 2005 3:30:14 PM

the Coach,
You can find out more about him here:

http://islandia.law.yale.edu/amar/

Posted by: BenW | May 10, 2005 3:30:14 PM

jkd: Mark Kleiman on the Democratic nuclear option.

My gripe with "wheel-of-fortune" voting, among others, is that it would eliminate a lot of institutional memory within legislatures. Also, in the days of seniority, this would have drastic consequences for districts which depend on the clout of their Representative to bring home the bacon (sure, it's pork, but for districts where federal employees are a third of the workforce, that pork is a big deal).

Posted by: Electoral Math | May 10, 2005 3:31:12 PM

Maybe I'm nuts, but I think this is pretty much what they do in Europe. You vote for a party, which has a set slate of candidates. If you are candidate number 39 on the list of 100, for a party that wins 38 seats, you're out of luck -- maybe next time, fella. Based on what Matt said, I see no difference. It's the same thing -- non-location-based proportional representation.

Then the permutations of relative party strength go into deciding who is in the coalition etc.

Not saying this is a reason pro or con, but the idea is far from radical.

Posted by: Martin | May 10, 2005 3:44:53 PM

No, Martin, you're wrong. There is no place on earth, I'm afraid, where the winner is decided by rolling the dice. Not even in Monaco.

Posted by: abb1 | May 10, 2005 4:03:16 PM

>Whichever candidate's lucky number gets called gets the seat.

"Nine nine nine nine nine nine nine"
"Who's that?"
"Our random number generator"
"Doesn't sound very random."
"That's the problem with random, you're never really sure."
-a Dilbert cartoon

Posted by: bartkid | May 10, 2005 4:11:49 PM

The expected result is the same as PR, but the actual results of any given election might not be. And I fail to see the advantage of this over PR anyway.

Posted by: fling93 | May 10, 2005 4:12:00 PM

And I think there's still a tactical advantage from shifting voters from a 100% district (either for you or against you) to a 50-50 district. Your likely number of district wins will increase.

Posted by: fling93 | May 10, 2005 4:16:18 PM

Just make it prop-rep with the added requirement that each party list have a member from each district.

Or, more impishly, individual district votes for candidates like now, but a prop-rep count of votes to determine party representation, with the highest vote total individuals for the party being highest on the 'list'. Yes, that means 'safe' seats are at an advantage at getting on the list, that happens now. It also means that high-turnout marginal districts may get two Representatives, as both candidates outpoll their opposite numbers in low-turnout areas. However, party breakdown won't be affected by gerrymandering, and so redistricting becomes non-partisan, if still political.

Low turnout areas aren't being disenfranchised - they still put their vote toward party representation, their candidate just lost the intramural contest. They are no more disenfranchised than the voters for losing candidates in safe seats are, anyway.

Posted by: rvman | May 10, 2005 4:16:42 PM

The comments to this entry are closed.