The Equal Weights Theorem proves that provided you know the direction of causality, assigning equal weights to each predictor variable is almost as good as linear regression. I suspect this is strongly related to the success of the top 10 predictor.

How to get Dreadful SQL performance

10 things statistics taught us about big data

The Fed's new risk retention rules require lenders to take less risk than they did before the crisis. It's a very interesting read that gets into details and busts a few myths.

Bayesian Bodybuilding?

Alternatives to GADTs in Scala

Interesting critique of Piketty's empirics rather than his theory. Very useful read. And Debraj Ray on Piketty's Theory. See also Tyler Cowen comparing Piketty's model to the standard Solow model. And Arnold Kling reviews a book studying inequality with longitudinal data rather than cross sectional. Very different results. And of course Tino Sanandaji on the wealth distribution in Sweden.

After discovering Debraj Ray, I also noticed her article encouraging cricketers and movie stars to explicitly denounce rapists. (This is probably less interesting to my non-Indian readers.) Her blog is hit or miss (lots of boring personal stuff), but her hits are interesting. Definitely worth checking out.

Scott Alexander asks naughty questions about Ta-Nehisi Coates' The Case for Reparations. I'd also ask another naughty question. Let us take as a given that Person A did bad things to Person B. What moral axioms and reasoning lead to the conclusion that Person C owes reparations to Person D? Sadly, Coates basically ignores this question.

Just discovered Colossal, which is a wonderful art blog.

It's surprisingly difficult to make a moral argument for open borders from a liberal perspective. Crooked Timber has quite a few posts on this topic, but nothing satisfying. In contrast, Bryan Caplan's moral reasoning is quite simple and satisfying, but would of course lead to conclusions that Crooked Timber would find completely unacceptable.

Details on normal approximations to other distributions. Handy.

For all my non-American readers, here are a couple of posts about bullying in the US.

#yesallcats because I can get myself out of a tree.


Subscribe to the mailing list


Comments

comments powered by Disqus