This is just a minor point, but I don't think the BMI is a good example. Any statistic that classifies Tom Cruise and Matt LeBlanc as "obese" has something seriously wrong with it.

The expectation for using BMI is pretty good, though. Most of the unwanted variance is due to folks w/ lots of muscle, which weighs more than fat. So if you want more fine-grained stats, percentage body fat would do the trick.

It seems like every other week there's something in the news about research linking some condition to a germ (schizophrenia/rats not caring about getting killed by cats, prostate cancer, and now obesity). Would you say this is a research program about to take off?

agnostic,

Yeah, that's the problem with the BMI -- it has no way to differentiate muscle from fat or between different body types. Percentage body fat would be much more meaningful.

Darth Quixote: “It seems like every other week there's something in the news about research linking some condition to a germ (schizophrenia/rats not caring about getting killed by cats, prostate cancer, and now obesity).”

I believe this reflects the growing power of new biotech tools. E.g., there is a now a “viral chip” that can detect the presence of several hundred known viral strains with a single, cheap test. That allows a researcher to routinely test tissue samples for broad viral activity. Likewise, modern sequencing technology accelerates the discovery of new microbes. New molecular probe technology can detect extremely low concentrations of specific proteins. Computer modeling ties the information together.

New biotech tools are revealing details that were below the resolution of previous technology. Thus we see a flood of new discoveries.

And a flood of new misunderstandings and confusions too. An example: I was diagnosed with what looked like a routine cancer five years ago, and put through the usual preliminary tests. (Only with shiney, powerful new machines.) Uh-oh! The tests turned up shadows and other indicators that looked for all the world like the cancer had spread. It took an invasive test and a week of seriously not-fun suspense to find out that my cancer in fact hadn't spread. The docs apologized and explained that the new machines are so much more powerful than what they're used to that they're turning up shit that the docs simply had no way to see before. Is it normal to see shadows there, with this strong new machine? They simply don't know. As a consequence, a lot of cancer patients these days are being put through Omigod-I'm-a-dead-man agonies that you might label "unnecessary" ...

Darth: Part is better tech, but detection isn't so hard sometimes -- if antibiotics work against it, it's probably infectious. In the late '40s in NYC, doctors used antibiotics to cure peptic ulcers, though utterly ignorant of the infectious agent.

The problem was assigning everything to genes and lifestyle. Paul Ewald's Plague Time is an easy read and provides plenty of background. Infectious causation of X has never been popular, and is harder to show as the causal chain becomes more cryptic.

There are plenty of "normal" human behaviors that would perplex a Martian evo biologist, but a lot of them allow people to lead enjoyable lives, so if we proposed that they were due to infection, our disgust instinct would kick in and say, "Oh I see, so now you're saying X is pathological?" Say, preferential male homosexuality (Cochran's "gay germ" hypothesis). From a well-being p.o.v., not pathological. But from a fitness p.o.v. -- hell yes.