Tuesday, August 9, 2011

Tim Harford: Trial, error and the God complex

In the Undercover Economist column he writes for the Financial Times, Tim Harford looks at familiar situations in unfamiliar ways and explains the fundamental principles of the modern economy. He illuminates them with clear writing and a variety of examples borrowed from daily life. The Guardian says of him: "Harford's invitation to a fireside chat in No 10, if not issued already, cannot be far off..."

He gave a TED talk in July this year of 2011.

Part of his talk was about a doctor in a German prison camp named, Archie Cochrane. There was a disease in the camp and he helped to figure out how to combat it. But he found another disease, one that he also had and in the end labeled, "The God Complex". We have him to thank for recognizing this important consideration. The symptoms of the complex are, that no matter how complicated a problem, you have an absolutely overwhelming belief that you are infallibly right in your solution. This last of questioning yourself can be deadly. They say, "Question Authority" and that should include, yourself.

Tim goes on to talk about a study that is being done where 5,000 products are put into a database by the physicist Cesar Hidalgo. He's at MIT. He said, "Now you won't be able to understand a word of it, but this is what it looks like. Cesar has trolled the database of over 5,000 different products, and he's used techniques of network analysis to interrogate this database and to graph relationships between the different products." His point here is that at WalMart alone, there are 100,000 products, that the world is so complex, there is simply no way to know everything. And that as excellent a study as Cesar's is, we have to acknowledge we that the world is just too complex for us. That' doesn't mean we should give up, it means we need to include that into the calculations.

Tim continues with an explanation of how we solve problems now a days. Many times things are so complicated that we simply cannot calculate an answer, but have to calculate (if we can at all) and then simply use trial and error until we hit upon the best case scenario or fix. Which, is pretty sad.

Getting back to Archie, he found that there was a question as to whether patients after heart surgery should recuperate home, or in the hospital. Of course, Doctors said in the hospital. But was that right? The Doctors knew that hospitals were the right place for patients and that it was unethical to run any kind of trial or experiment.

"Nevertheless," Tim continues, "Archie managed to get permission to do this. He ran his trial. And after the trial had been running for a little while, he gathered together all his colleagues around his table, and he said, "Well, gentlemen, we have some preliminary results. They're not statistically significant. But we have something. And it turns out that you're right and I'm wrong. It is dangerous for patients to recover from heart attacks at home. They should be in hospital." And there's this uproar, and all the doctors start pounding the table and saying, "We always said you were unethical, Archie. You're killing people with your clinical trials. You need to shut it down now. Shut it down at once." And there's this huge hubbub. Archie lets it die down. And then he says, "Well that's very interesting, gentlemen, because when I gave you the table of results, I swapped the two columns around. It turns out your hospitals are killing people, and they should be at home. Would you like to close down the trial now, or should we wait until we have robust results?" Tumbleweed rolls through the meeting room"

In closing: "Shortly after the war, this young man, Yutaka Taniyama, developed this amazing conjecture called the Taniyama-Shimura Conjecture. It turned out to be absolutely instrumental many decades later in proving Fermat's Last Theorem. In fact, it turns out it's equivalent to proving Fermat's Last Theorum. You prove one, you prove the other. But it was always a conjecture. Taniyama tried and tried and tried and he could never prove that it was true. And shortly before his 30th birthday in 1958, Yutaka Taniyama killed himself."

"His friend, Goro Shimura -- who worked on the mathematics with him -- many decades later, reflected on Taniyama's life. He said, "He was not a very careful person as a mathematician. He made a lot of mistakes. But he made mistakes in a good direction. I tried to emulate him, but I realized it is very difficult to make good mistakes."

No comments:

Post a Comment