« January 2019 | Main | March 2019 »

February 2019

"Machine learning 'causing science crisis'"

I know only a thimble-full of stuff about "machine learning". It's on my ever-lengthening list of things to read more about. I intend to look for how it handles the well-known econometric problem of over-fitting. I thought, surely, it has some excellent way to handle that. But maybe not:

Dr Genevera Allen from Rice University in Houston said that the increased use of such systems was contributing to a “crisis in science”.

She warned scientists that if they didn’t improve their techniques they would be wasting both time and money. Her research was presented at the American Association for the Advancement of Science in Washington. . . .

But, according to Dr Allen, the answers they come up with are likely to be inaccurate or wrong because the software is identifying patterns that exist only in that data set and not the real world.

"The Two Movements in Economic Thought, 1700-2000: Empty Economic Boxes Revisited"

A recent paper by Deirdre McCloskey which productively combines and extends two of her recent themes: the "Great Enrichment" and empirical "oomph".

As a bonus, the reader gets a list of 104(!) possible imperfections in the market followed by Prof. McCloskey's ringing declaration that none of them have been shown to be very important empirically.

"The Resurrection of Emmitt Till"

Amazing and sad. The only good news is that I think a lot fewer people pay attention to the New York Times than they used to.

In 1980, the name “Emmett Till” did not appear in the pages of the NYT. In 1990 it showed up twice, and in 2000 four times.

From 2004 through 2012, the Times mentioned this old incident an average of nine times per year, and from 2013 to 2016 almost two dozen times per year.

Last year, “Emmett Till” appeared in 72 different Times articles.

"Sumner on teaching economics"

I agree wholeheartedly with John Cochrane:

An important point is implicit. Economics is not hard because of math. The math in even graduate level economics is no greater than in sophomore physics. [I'm not sure about this. --Craig] Classical economics is hard because it can attack social problems in a value-free, cause-and-effect way, and upends the little morality stories that most people use to think about those problems -- rents are high because landlords are greedy. "Learning to think like an economist" is indeed best learned by application. And "learning to think like a behavioral economist" requires learning to think like an economist first.


I also agree wholeheartedly with Peter Gordon. (And this is rarely addressed by climate change fanatics.)

Climate change, to the extent that it is human-caused is a global commons. The Prisoners' Dilemma indicates the futility of individual action.  The chart below (from The Economist) shows that in recent years, the U.S. has been an over-achiever reducing carbon emissions. China and India are under-achievers.