The key of falsifiability

When I hear people talking about teaching intelligent design, "proofs" of economics, and touting home remedies, I think about what makes something a scientific theory.

It needs to make some kind of predictions that can be tested. A scientific theory has to be falsifiable; it simply has to be the case that measurements could be made that would make it false. When they fail to do so, we proclaim the theory good.

An object falling does not prove gravity. I could just as easily say that things want to be close together, that celestial bodies naturally move in circles while terrestrial bodies want to be stationary on a low-lying surface, and explain everything that way. Data always underdetermines theory, points underdetermine functions, facts never tell the whole story of a case.

So why? The best theory is the one that walks the knife-edge between falsifiable and false. The slightest changes in the data could invalidate it - but somehow, they haven't. Arguments about government policy illustrate the point perfectly; is the problem too much protectionism, or too much free trade? Too much subsidy, or not enough? We can stretch and contrive a complex explanation justifying ourselves, and in fact, pundits seem to do so all the time.

What would it take for you to disbelieve this idea? If you can answer that question, you know whether you're relying on science or faith to justify your belief.

Median voter theorem

Why politicians talk in code

The median voter theorem is an interesting result in the study of elections. Suppose you have some political spectrum, and two candidates. Doesn't matter how many dimensions there are to that spectrum, even, with two candidates; the politician who is closer to the center (median) of voter positions will be closer to more voters. Voters - if acting rationally - will vote for the closest candidate to them.

Voila, so suddenly we have centrist candidates - or do we? There's another problem: Turnout. The further away a voter is from a candidate, the less likely a voter is to turn out. In fact, if they feel far away on the fringe, they may feel the difference between the two centrist candidates is negligible. Perhaps a third party candidate will look attractive - it's time to make a statement!

But what if you can occupy more than one position at once? Suddenly, you can secure much more of the political spectrum. This is why politicians talk in code, and try to position their opponents as extremists; why we see different messages sent through different media. Coded language is understood differently by different segments of the political spectrum, and by segmenting your audience into different groups, it allows you to try to position yourself near the median of each group rather than the whole population.

And this is how a politician masters the median voter theorem; not by moving carefully to the center of the population, but by seeming to be in different places when looked at from different perspectives.

Measuring a libertarian

I've been thinking about this one for a while. There are social libertarians; there are economic libertarians; the idea is less government. And at the end of the day, I think that this might be the best way of all to measure whether or not someone really is a libertarian:

What is out there, that you think is wrong, but nevertheless believe should be legal?

For example, as I mentioned the other day, I think prostitution should be legal - carefully regulated in the public interest, but legal; however, I do think there's something terribly wrong with selling sex services. I'm even bothered by the overly mercantile nature of much dating, by mothers who tell their daughters they should judge a man by the price on the ring he brings them, by the high class escort services that carefully step around prostitution laws, by gold diggers, and by "Who wants to marry a millionaire?"

I am at least a little bit of a libertarian in that way. I want the government to step in because there is a compelling public interest - not because my personal sense of right and wrong is affronted.