Wednesday, November 21, 2012

Three Thoughts on Silver's Book

I've indulged myself with a quick red of Nate Silver's The Signal and the Noise and yes, it's wonderful.  The mature autumnal wisdom of a 34-year-old, dressed out with a near limitless fund of good stories about predictions gone wrong and otherwise.  I don't suppose I can add a lot of value with a complete review just now--maybe later--but let me offer a few disjointed passing thoughts.
  • It's really all in the "case study" chapters--on baseball, sports gambling, climate change and so forth. Such as there is by way of abstract lesson-drawing is good enough but hardly necessary if you have read the specific chapters, pretty much worthless if you haven't.  I do wish he hadn't hit us over the head so often with "Bayesian,"  when he could just as well have repeat Brad DeLong's catchphrase that we must mark our beliefs to market.  On the other hand, I much enjoyed his survey of possible innocent explanations for the presence of a pair of panties in your husband's underwear drawer.
  • Here's one proposition that intrigues me.  Silver makes the point that sometimes has the incentive to exaggerate, even at the expense of being wrong.  Quite persuasive as he presents it: if you are a TV pundit, nobody will notice if you go along with the crowd; the only way to get air time is to say something weird or off the wall.  Quite plausible as set forth here but do you notice how it contradicts what you were taught in school.   That is: we used to say--don't predict; if you are right, no one will notice and if you are wrong, everyone may notice.  Silver seems to be suggesting that (as with so many proverbs) at least part of the truth might be just the other way round.
  • Silver obviously did a lot of interviews for the book. And he goes out of the way to show that he did them in person: he specifies where they met, what the guy looked like, and so forth (Donald Rumsfeld is old, and short).  A little mental arithmetic told me that he must have been doing all interviewing at the same time as he was riding the wave of success for his 538 Blog.  Which led me to wonder--who's minding the store?   How do you get out a world-class blog if you are gallivanting all over hell's half acre?  The immediate answer seems to be: the blog is all in his laptop, and he can do that from anywhere--the departure lounge, the hotel lobby, even the great man's waiting room.   But wait a minute: if we live in an electronic universe, why do all these interviews have to be in person?  Left as an exercise for the reader.


Jimbo said...

I've met Donald Rumsfeld. He's old and he's certainly an arrogant, bad person in my estimation but he's not short. He's about average height for American males, around 5'9 or 10. In Holland or Rwanda he would be short but not in our multi-ethinic country.

Jimbo said...

I give thanks for MST 3000 as was and will ever be and for Scott and Mary's wonderful site and for their always educational movie reviews, which I ripped off to get a PhD from the University of Bismarck Online. Yay!