Sunday, March 27, 2011

Dan Gardner: Future Babble

Dan Gardner's latest book, Future Babble, was just released in the U.S.: click to buy via amazon.com.

It's a whole book about the self-described experts who predict the future and who are almost always wrong - and about the irrationalities and biases that allow these "experts" to maintain their self-confidence and influence despite their repeated mistakes.

The author divides the experts to "hedgehogs" and "foxes": the former always feel absolutely certain and they love to present their predictions in an unambiguous way; the latter partially realize the complexity of the questions but their predictions are found boring by the public. Consequently, the hedgehogs are always wrong while the foxes are just almost always wrong but the hedgehogs, because of their higher attractiveness, have a greater impact on the mankind. ;-)

The book offers quite some selection of particular nonsensical predictions about the future that were proved wrong - as well as particular people who have always been wrong, with a repeatability that reaches comic proportions, but people haven't yet managed to deduce the consequences. Gardner also tries to explain why the people keep on believing experts even though they must know that the predictions will be wrong.

In 1911, experts would predict that there would be no more wars in Europe. In 1967, experts predicted the USSR to be the fastest-growing economy of 2000. In 2008, when oil surged above USD 140, experts would say that the oil price would soon hit USD 200. Mr Alexander Ač of Czech Globe, who has been collecting and believing every single prediction of this kind, had a version of this prediction on steroids: he followed an expert Robert Hirsch to claim that the oil price would reach USD 500 per barrel in 2011-2013.

Face it, "experts" are as accurate as dart-throwing monkeys. The book may also offer some answer why some people such as Mr Alexander Ač keep on believing so self-evidently irrelevant random farts by so random "expert" crackpots such as Mr Hirsch.

Some biases are behind the repeated errors. For example, the "hindsight bias" leads most people to reinterpret the past according to the outcomes that only became known much later - to retroactively "create a story". Such a reinterpretation makes the people incorrectly believe that the events could have been predicted more reliably than they actually could.

Psychologist Steven Pinker of Harvard University has nicely and accurately :-) summarized how the book will change the humanity if people start to buy it and read it en masse:
It’s rare for a book on public affairs to say something genuinely new, but Future Babble is genuinely arresting, and should be required reading for journalists, politicians, academics, and anyone who listens to them. Mark my words: if Future Babble is widely read, then within 3.7 years the number of overconfident predictions by self-anointed experts talking through their hats will decline by 46.2%, and the world will become no less than 32.1% wiser.
There's no reason to doubt Pinker's figures. After all, he's the world's top expert in psychology.

See Introduction, an excerpt via the New York Times, a review in The Boston Globe (extended), interview with the author for U.S. News, and a review in NYT. The lady who wrote the latter review is unfair; Gardner is surely not an anti-intellectual who claims that predictions are always impossible. But, as he says in the interview, if it comes to really complicated events, e.g. those that depend on the weather if not the human psychology, the task to sensibly predict is just hopeless.

Radiation increased 10 million times?

Today, we saw another example of the mechanism that the more preposterous a statement is, the more quickly it is spreading. Pretty much all global media have "informed" us that the radiation near a Fukushima reactor was 10 million times higher than the normal. That would be a lot, indeed. Death within an hour or minutes.

Of course, TEPCO has revealed that the information was a "mistake". How did it happen? They confused the readings for cobalt and iodine. Well, that's a pretty bad mistake for a report that is guaranteed to be reprinted at thousands of places and read or watched by a billion of people.

But the error could have been found at other layers, too. No one has done so, however: this news is so "hot" that it doesn't really matter whether it's true, does it? My blog has gotten immediately bombed by anti-nuclear zealots who spread the wrong number, too.

By the way, if you care about the details, it's very likely that they just measured the concentration of iodine-134 and cobalt-56 whose half-lives are 50 minutes and 77 days, respectively. They may have interpreted the concentration of cobalt as that of iodine, but because the cobalt is decaying 2,000 times more slowly, the actual radioactivity caused by it is approximately 2,000 smaller - so 10 million gets shrunk to 5 thousand. ;-) That allows you months of life over there without cancer.

Obviously, the people who have actively propagated the bullshit about 10 million will prefer to say to 5 thousand and 10 million is essentially the same number and nothing changes - instead of admitting that they have qualitatively screwed their job and they just generally suck.