I liked this line from this piece at Farnam Street:

A lot of us are on the treadmill of consuming expiring information.

Expiring information is not the same as trivia, which can be useful and interesting. It is instead knowledge which has a short shelf-life.

A lot of talk about Brexit has been expiring information. We are no nearer today to knowing what type of Brexit we'll get (or even if or when we'll get it) than we were on June 24^th 2016. All the effort expended over the last 32 months on wondering what will happen has therefore been largely wasted. News which looked useful at the time turned out not to be. It was more noise than signal.

Much the same is true in stock markets. It's very easy to spend your life accumulating lots of detailed factoids about individual companies which give the impression that you are an expert but which are otherwise useless. Either they have no ability (pdf) to predict future returns, or if they had it once they might not have it now or in future. Such factoids are expiring information: in fact, insofar as the efficient market hypothesis is correct, the expiry date has passed.

What can we do about this? I have two responses.

First, I ask of any claim or fact: so what? Why should I care? In stock markets, we have a handful of factors which have (at least in the past) been good lead indicators of returns on the aggregate market. These include: the dividend yield; foreign buying of US equities (a great indicator of investor sentiment) and - yes - the time of year. The rest is noise.

Similarly, if I want to know whether the US will experience a recession soon, I look at the yield curve and consumption-wealth ratios (pdf) and discount heavily most other information.

This point probably generalizes to fields other than finance.

Warren Buffet, I suspect, does much the same thing. Despite Farnam Street's portrayal of him as a gatherer of useful facts, his success has rested largely upon exploiting two big things: an ability to raise very cheap finance; and the buying of defensive stocks, a category of share we've known since 1972 to be under-priced.

Secondly, I try to collect examples and counter-examples of general purpose knowledge. For me, there are two big sets here.

One set is of cognitive biases - mistakes we make and which we should avoid in future decisions.

The other set is of mechanisms. As Jon Elster said, the social sciences are essentially a collection of mechanisms. Sometimes, for example, there are mechanisms through which markets work well to allocate resources and sometimes there are ones (pdf) through which they work badly.

What I'm doing here is blurring the distinction Ernest Rutherford (probably) made when he said that all sciences were either physics or stamp-collecting. It's true that the social sciences don't have the handful of powerful predictive laws that physics does. But this doesn't mean we must content ourselves merely with gathering disparate unrelated facts. Facts can be organized into examples and counter-examples of mechanisms - although we might never have a complete inventory of these.

I'm also doubting the relevance of the ancient distinction between foxes (which know many things) and hedgehogs (which know one big thing). The important thing isn't whether you know one thing or many things, but whether you what what's relevant and what's not.