Sunday, September 29, 2013

Review of The Signal and the Noise: Why So Many Predictions Fail —but Some Don't by Nate Silver

Nate Silver's new(ish) The Signal and the Noise: Why So Many Predictions Fail —but Some Don't opens by crediting the increase in GDP per capita since the industrial revolution to the cumulative effects of the spread of information. And it is true that many measures of information exchange or accumulation have grown in perfect harmony with increases in production and standard of living.

However, this ignores a competing explanation, that the increase in our standard of living has been driven by an increase in available energy, and specifically energy sources that yield a big return (10-100x) on energy invested (EROEI).

In general, Silver has a superficial understanding of the topics he covers. He implies that Nassim Taleb discovered that financial variables have fat-tailed distributions (p368), when really fat tails are pretty old news to everyone in finance. He mentions that the average holding period of common stocks has been getting shorter, but apparently is not aware that this is cyclical, and the previous nadir was in 1929.

Silver misses the opportunity to develop systematic explanations for why people make bad predictions, for example the principal/agent problem. The problem is that Silver is a technocrat so he can't tell you that bureaucrats make bad predictions because of an incentive misalignment wherein they prioritize their careers over the lives and property of those whom their predictions can harm. Taleb always pounds the table about lack of skin in the game.

The worst parts are when the book succumbs to left wing dogma. For example, he preaches global warming with nary a mention that there was a Little Ice Age that ended in 1850.

Nate Silver correctly predicted one presidential election where the losing candidate took a dive and another where the losing candidate correctly predicted that he would lose because there was a winning majority already on the payroll of the incumbent.

2/5

7 comments:

Anonymous said...

Can you think of major, long-term increases in available energy divorced from increases in knowledge about same?

CP said...

No.

Anonymous said...

So then we'd perhaps agree that energy availability is just a subset of information availability.

Nate Silver rose to fame through an interesting bit of innumeracy. He was the guy who correctly predicted 49 out of 50 state results. But since somewhere around 40 of these were foregone conclusions or close to it, he's the guy who got 9 out of 10 right, which is maybe a little less exciting.

That said, the fact that there was a Little Ice Age that ended in 1850 (or perhaps even later) doesn't mean that there isn't warming now (just as the fact that there was a booming economy in the 1950s doesn't mean things are wonderful now). There's a fairly broad consensus (not 100% but not that far off) that the globe is warming and that we have something to do with it. I'm not sure dismissing this as left wing dogma is wholly intellectually honest.

It would undoubtedly be better to know much more about the mechanisms of this warming (something I am sure most scientists would agree with). But this might be an instance in which the perfect is the enemy of the good.

CP said...

Hmm, I was thinking that information ability much just be a subset of energy availability. What limits the growth of bacteria in a dish, nutrients or information?

CP said...

Isn't he famous for the 2008 presidential election, where the losing candidate took a dive?

CP said...

Anyway, the thesis of this book review is:

Silver is a technocrat so he can't tell you that bureaucrats make bad predictions because of an incentive misalignment wherein they prioritize their careers over the lives and property of those whom their predictions can harm.

Anonymous said...

Certainly nutrient availability. But if we kick up the evolutionary ladder a bit to, say, a slime mold, the information about nutrient availability in adjacent petri dishes comes into play. If we go all the way up to England in the 18th century, energy availability was profoundly increased by the Watts, both in increasing access to coal by pumping out water and in creating new energetic uses for coal above the surface.

As for Silver, he was famous for predicting correctly the presidential election, yes; the 49/50 were state results in the 2008 election. Whether the loser took a dive is subject to debate but there's certainly plenty of good argument that he did.

In general, misaligned incentives and principal-agent problems abound and are not given their due in understanding of both public and private sectors. Too much rent-seeking...