|Expert Political Judgment
||[Dec. 29th, 2015|12:08 am]
De Horror Vacui
Expert Political Judgment: How Good Is It? How Can We Know?, Philip E. Tetlock:|
There are some ideas that are too good to ignore. And this one is it. Tetlock, and I assume whoever he was working with, had the bright idea to check up on those experts in political and economic ideas whose divinations Those Who Rule consult before they act -- to invade a country or design a tax credit or some other good that is evil or evil that is good. They had these experts answer many different questions over many years based on what sort of things could possibly change in a reasonable amount of time (and all questions were time-boxed: e.g., "Will Leonid Brezhnev release a new country and western album in 1981?"). These were then run through statistical analyses.
( ReviewCollapse )
A few years ago, I was approached to do a study like this one (I don't remember who was in charge), but I dropped out after a few months. The main reason is because the questions were so far out of my expertise. I'd fill out these questionnaires and constantly have these answers like "I kind of think Idi Amin's next release will top Benito Mussolini's record sales record, but there's still a chance that he won't make it past Fidel Castro's 2nd place" and then have to tell the people in charge of the study that "I'm not terribly qualified to make judgments about despotic rock music" and "I'm only 20% sure in the prediction." Obviously, the questions weren't quite that detailed, but the extra should give you a feeling for what I'm doing.
That makes me wonder a little about the drop out rate for Tetlock's previous studies and how it affects his evidence, especially about his foxes.
However, I think that the basic results of this book are wonderful:
(1) The best prognosticators are experts that are least sure of themselves.
(2) The best prognosticators are no better than flipping a coin.
(3) A linear extrapolation is better than flipping a coin.
(4) The worst expert prognosticator is the one who is the most sure of himself.
(5) The prognosticator who is most likely to be listened to by others is the one who is most sure of himself.
(6) The worst expert prognosticator is better than a very, very smart amateur.
We are already well past the point when the machines should be in charge. If nothing else, a random number generator is cheaper to run than the government.
And it's a lot smarter.
* Interestingly, The Hedgehog, the Fox, and the Magister's Pox was the last book I bought by Gould, and the first one I couldn't get through because it was so bad. After reading this book, I tried again, with no luck -- even on a transpacific flight.
( Other Books, 2015Collapse )
I could possibly get three more (two almost certain), unless I brought a lightweight like "Nudge" with me to Houston or run off to the used book store and find a nice, old novel that reads well (current novel: too much description and talk, not enough plot). That means no chance of hitting fifty. Another year within 1 or 2.
What happened to the days when I could read three novels in a weekend?