Friday, February 13, 2004

Random Enfranchisement

A passing comment by Brian Weatherson (on the topic of vote-counting in Florida) got me thinking – and we all know how dangerous that can be.
So we’re just going to trust the computers. Given how reliable we all know computers to be, this is about as democratic as selecting candidates by lots. (Just for the record, I think it’s an interesting theoretical question about how democratic that is. It’s how we pick juries after all, and they are often considered an important part of the democratic process.)
Set aside selecting candidates by lots, and consider a somewhat different proposal: selecting voters by lots. Instead of allowing anyone who meets the formal criteria for voting to vote, why not take a random sample of registered voters (or people in the census) and designate them as the only people allowed to vote in a given election? If the sample were sufficiently large, it would represent the views of the overall population, within some margin of error (a sample of a mere 10,000 would generate a vanishingly small margin of error, if the sample were truly random). The people selected would have a stronger incentive to become informed and vote wisely than in the status quo, because each vote would stand a greater chance of actually making a difference. The chance would still be very small, of course – it must be small in order for the law of large numbers to kick in and support my prior claim about having a representative sample. But apparently feelings of civic duty interact with very small chances of making a difference; otherwise, non-voting would be even more common than it is. I surmise, therefore, that increasing the weight of one’s vote by two or three orders of magnitude would have a substantial impact on voting choices, even if the absolute likelihood were still tiny.

Aside from the benefits of motivating qualified voters to consider their votes more carefully, the proposed system would reduce the administrative costs of voting. We could invest a great deal more per voter in safeguarding against fraud and error and still save money. We’d need a mere fraction of the ballots, voting booths, and manpower to run the system. Plus, everyone else could stay at work instead of taking time off to vote.

Yeah, yeah, this is a pipe dream. It’ll never happen, which is why I don’t have to worry much about the possible downsides of my scheme. But I’m curious – what would the disadvantages be? The obvious one is that people whose name never got randomly chosen would feel disenfranchised (which, by some definition of the term, they would be). Yet everyone would have the same chance of being a voter, so there would be no formal violation of equal treatment principles. The communitarian objection would be that participation in the voting process is desirable in and of itself, as a device for making people feel connected to each other and their government. This argument is not terribly persuasive to me, since I’d prefer that people feel attached to voluntarily chosen communities instead of the state, but I can still understand the objection. What other objections are there?

1 comment:

Anonymous said...

Issac Asimov took this idea to it's minimalist extreme (one voter) in his 1955 short story "Franchise". An interesting enough story that it sprang instantly to mind when I read this blog entry.