Science of the Times

ScienceBlogs is well stocked with people who will comment at length on the latest news from biology or climate science, which is nice, because I don’t usually feel moved to remark on those subjects. The large amount of quality commentary on those subjects does make me fell like I ought to make a point of commenting on physics-related news stories. Happily, there aren’t nearly as many of those, so I can easily accomodate this with my relatively light posting schedule (light compared to some of my every-hour-on-the-hour colleagues– don’t you people have day jobs?).

There were a few science-related stories in the New York Times this week that caught my eye. One of them is yesterday’s NASA release about liquid water on Enceladus, one of Saturn’s moons. PZ has already commented on this (update: So has Alex Palazzo), and he’s got it about right: it’s interesting, but not earth-shaking. They’ve got some reasonably good evidence (and cool pictures) of liquid water geysering up from one of the moon’s poles, which is unexpected, and will cause many a planetary scientist to do the Happy Dance, but the man-on-the-street reaction isn’ likely to be much beyond “Hey, cool pictures,” despite some hype about the possibility of life.

So why the hype? The same reason as always– money. (More below the fold.)

As extensively documented here and elsewhere, NASA’s funding for science missions is being squeezed, and some popular programs are likely to be cut. The only hope for those missions is to get Congress to add funding for them in an appropriations bill, and the only way to get Congress interested is to get the public excited. Hence the “possible life on moon of Saturn” angle to these stories– if that’s what it takes to get the story in the New York Times, that’s what they’re going to do.

It’s a little cynical, but honestly, I hope it works.

The other foolproof way of getting yourself into the Times is scientific misconduct, whih brings us to the other story from this week (well, two stories, on the same topic), regarding the investigation into claims of fusion ina sonoluminescence experiment (follow-up here). The story here is that a scientist at Purdue claimed to have detected signatures of nuclear fusion taking place in a table-top apparatus– specifially, neutrons released in the process (the absence of any neutrons was one of the key data points that sunk the original Pons and Fleischman cold fusion paper). Other scientists have been unable to reproduce the results, and at least one person claims that the neutrons detected are much more likely to have come from improperly stored Californium samples in the same lab. The university is opening an investigation, the scientist responsible stands by his results.

On the surface, this might look like the physics equivalent of the whole Hwang Woo-Suk mess in biology (though he wasn’t a complete fraud). As it is, though, it’s pretty much a non-story for one simple reason: nobody really believed the original claim, so it didn’t have much of an effect on physics in general. A few people tried to replicate the results (and at least one got significant funding to do so), but most physicists said “Cold fusion again,” shrugged, and went back to trying to make a quantum computer.

The final Times that I marked this week isn’t actually a science story, because it’s about economics. This one concerns a claim of massive point-shaving in basketball, detected by “forensic economists” who analyzed how often teams beat the point spread set by sports bookies. They found that narrowly favored teams tended to beat the spread about 50% of the time, more or less what you would expect from random chance, while heavily favored teams were much more likely to narrowly miss covering the spread.

I haven’t read the research in detail (and likely wouldn’t get much out of it), but the article in the Times leaves me extremely skeptical. It sounds very much like the researchers have fallen for the classic sports fallacy of the point spread. Contrary to what they seem to think, the point spread is not the best expert estimation of what the final score will be. Instead, it’s the number that bookies need to use in order to get half of the people placing bets to bet on each team.

Those aren’t necessarily the same thing. When you’re dealing with evenly matched or well-known teams, they can be almost the same, because the people betting are likely to have a pretty good idea of what’s going on. In big mis-matches, though, or games involving one marquee team and one relative unknown, the spread doesn’t have much reality– bettors are much more likely to be swayed by emotion or prior knowledge, and bet less rationally. The bookies are in the business of predicting bettors, not sporting events, so the point spreads will reflect this.

This gets rediscovered by the sports media about four times a year, most recently when Pittsburgh opened as a big favorite to beat Seattle in the Super Bowl. That wasn’t so much a result of the fact that the Steelers were rationally expected to beat the Seahawks by a huge margin, as it was a result of the fact that the Steelers have a wider fan base and better name recognition than the Seahawks. People know more about the Steelers, and are more inclined to bet on the Steelers, so the line had to be pretty large to get people to bet on the Seahawks. And bookies make money by getting half of the people betting to bet on each team, so the spread goes up.

There are other factors involved here having to do with psychological aspects of the game of basketball, but just from the nature of point spreads, you’d expect about the results reported in the article: point spreads should be a pretty good predictor of the actual score in games that are expected to be close, and should be systematically too high in big mis-matches. There’s no need to invoke the specter of point-shaving here.