Thursday, February 17, 2005

Authorities

From a story on a vaccine treatment for cancer: Treatment for Prostate Cancer Shows Promise (washingtonpost.com):
Although hope had dimmed for the approach, some scientists have quietly continued to pursue it.

“It seemed really good, and then the pendulum swung back because we didn't see promising results, but now this is clearly some of the strongest evidence we've seen of a clinical benefit,” Gulley said.
This is a good, quick example of why scientists can't be talked about as “scientists” in some at-large sense. Lots of scientists were interested in this technique for curing cancer, but when it didn't pan out, many went on to other things. But some scientists stuck with it, and are getting results.

Scientists are united not by their interests, but their intellectual methods. Scientists are united by the scientific method, a means of asking questions. Secondarily, and only secondarily, scientists are linked by the answers to those questions. But I feel a connection to geologists and biochemists even though I couldn't tell you anything about either.

It's hard to fool a bunch of people who see themselves as questioners, and only secondarily answerers. That's a mindset that has instant skepticism of absolute answers. Use absolutes (“all,” “must,” “did”) around scientists, and you'll get jumped on. Use qualifiers (“some,” “may,” “might have”) and you'll have an interesting conversation. You might wind up getting jumped on anyway, but politely.

Some people think scientists are like bankers for facts, and that we hoard facts, doling them out to the public in little spurts. Scientists want attention, and the only path to attention in science is to talk about your results.

This isn't the same situation as journalists. The Gannon-Guckert flap, like the credulous WMD reporting and a thousand other errors, show the flaws in the modern journalistic enterprise. The collapse of journalistic infallibility, like Watergate's opening of politics' smoke filled rooms, has lowered the institution as a whole, and made it easy to criticize journalists as partisan, flawed people. But good journalists are also dedicated to asking good questions, and have a skeptical nature. Why should we give scientists more credit than journalists?

One reason is time. Journalists are on deadline, so it's easy to spin them. Give them a half written story, and they'll be happy to polish it up and print it. A journalist might not care why the Kerry campaign gave them registration data from particular areas only beginning in 2004. If the Bush campaign wanted to, they could have given out the same data for the previous two years, but reporters are busy, so they might not dig that up on their own.

The other reason is that journalists aren't (usually) known for the claims they make in their reporting. Dan Rather and Mary Mapes will not be forever stained for their sloppiness in putting together the story of Bush's National Guard service, and neither will any individual covering the run-up to the invasion. Why? Because everyone was covering the rush to war, and everyone was trying to be first. There wasn't time to distinguish the good reporters. Knight-Ridder papers got good reporting over all, but I don't know who wrote the stories.

Scientists are about carving a new direction. Even if 1,000 scientists are studying the origins of HIV, each is doing something different. They are asking different questions, trying to get good answers. Journalists see verification when every paper runs an identical story; scientists see signs of fraud. Scientists want crazy ideas to test, it's a shot at an open field.

My advisor likes to say (in a statistical context) that truth is the intersection of multiple, independent lies. That's what scientists see. Each paper is flawed, each experiment is imperfect. But those errors are random, and they cut out some falsehoods. Eventually, you carve out a nugget of almost pure truth. Scientists want to be right in ten years. Journalists want to be right now, and have to cut some corners to get there.

In reading Theodore Rex by Edmund Morris, there were Democratic papers and Republican papers, and you knew who was who. Now, there aren't any Democratic papers. The Washington Times and New York Post are Republican, and their reporting and editorial pages bear that out. But the New York Times and Washington Post aren't Democratic, they just want the truth. That leaves a gap.

Scientists are adversarial. The easiest way to get a paper published is to re-analyze some other guy's data and try to tear his conclusions down. Journalists don't do that, at least not from both sides. They should. Bob Novak should have had to worry that he and his sources would be exposed by another journalist in the Plame case. Judith Miller should have worried that her connections to Ahmed Chalabi would be exposed by another journalist. That would keep the enterprise honest.

That's why scientists are less unified and more likely to be right when they agree. If scientists are all on the same page, it means every possible angle of dispute has been explored and rejected.