NEJM suggests it's rude to use an author's own data to prove them wrong. Um, that's not how science works.
An anonymous response to dangerous FOSS Codes of Conduct
A quantitative analysis of free speech.
An unconditional increase in teacher salary has no beneficial effect.
Fools Gold: An Illustrated Critique of Differential Privacy and On Syntactic Anonymity and Differential Privacy are both good critiques of the practical usefulness of differential privacy.
Class Systems in America, and relatedly Castes of the United States and The 3-Ladder System of Social Class in the US. Michael O. Church and Mencius Moldbug are mostly agreeing on something.
Sometime back Paul Graham wrote about a way to detect bias, namely by measuring the relative ability of people who are selected. I.e., if women are discriminated against, they should outperform men. I did a little more math on that topic. Now Jonathan Adler is using this idea to investigate whether conservative law professors are discriminated against. Apparently conservative law professors are significantly more productive than their liberal counterparts, suggesting that there may be possible discrimination.
Systems, not Sith: How Inter-service Rivalries Doomed the Galactic Empire
Hedonic Adjustments over long periods of time. This article argues a point I've long held, that inflation is non-transitive. I.e., the rate of inflation from 1970->2016 is not equal to inflation 1970->1990 + 1990->2016. The reason for this is ultimately changes in the basket of goods - changing the basket of goods gets you a very good inflation rate in any given year, but a completely nonsensical one over time.
Nice introductory article on covariance/contravariance.
Why Costs Matter - an article explaining why transit activists need to focus on out of control construction costs. My favorite point of comparison is that the NYC 7-line extension (1 mile of subway, 1 station) cost $2.4 billion, while the Delhi metro (132 miles of track, 160 stations) cost $5 billion.
Silicon Valley Has an Asian-people Problem - the first news report on ethnic diversity in the tech industry which accurately describes the numbers.
Github: The Quiet Death of One Man's Dream.
Interview with a Quant, Pt 1. Discussion of how to find/test trading strategies. Looking forward to part 2.
How to become a Bayesian in 8 easy steps. Good listing of papers to read.
Apparently the SAT, including SAT math, will become more reading heavy. It's supposed to be harder for people who are good at math but might not speak English at home. I wonder what group of people that is?
Scott Alexander takes down the github male/female pull request article. He also has a great comment.