illustrated cross section

An illustrated cross section of Kowloon Walled City

Probable points and credible intervals. A great article about things to do with a posterior once you compute it.

How misaligning data can increase performance 12X by reducing cache misses

Scalazy streams - assorted tricks to make your Scala 2.11.x collections faster.

Why protocols are a messy concept. tl;dr; binary protocols are super awesome and we should use them more. Fuck json-over-http.

Sketching as a tool for numerical linear algebra. Very interesting tool - for solving many problems like A x=b, you can often solve a problem SAx = Sb, where S is a random matrix that reduces the dimensionality of the problem.

The introduction to reactive programming that you've been missing

How to desugar Haskell code

The Binary Problem and The Continuous Problem in A/B testing.

The Counter-intuitive Non-informative Prior for the Bernoulli Family.

How do you write your model definitions. Very nice article about how defining models via a calculation procedure hides the salami.

The flaw lurking in every deep neural net is a pop science article, and the original paper is Intriguing properties of neural networks. The gist is this: neural networks are not "continuous" functions, in the sense that an adversary can take almost any image and produce a slightly perturbed variation that will be misclassified. Moreover, the adversarial result generalizes across a variety of different networks.

Low Rank Matrix Approximation in Linear Time. Very clever result.

Convex.jl - disciplined convex optimization in Julia. See also John Boyd's free book on the subject.

Why I support statisticians and their resistance to hype

Suddenly, the opposite happened. This article is a great takedown of the anti-intellectualism that various folks are pushing on computing under "won't someone think of the women/minorities/children" banner.

First go read this article on Japanese Zoning Regulations. It's a very interesting look at how Japanese zoning regulations are far more rational than American ones. The basic idea is that Japanese zoning rules restrict the amount of nuisance allowed in a region - you are permitted to build a low nuisance home in a region zoned for medium nuisance (light commercial). In contrast, American zoning rules restrict areas to an exclusive use (i.e., shop or home, not both). After you read this article, read this article on Social Technology and Anarcho Tyranny.

Hollaback and Why Everyone Needs Better Research Methods, And Why All Data Needs Theory. I didn't watch enough of the video to catch the racism, but yeah. WTF.

A radical solution to global inequality - tl;dr; temporary economic migration combined with a caste system. Intuitively I want to oppose this, but their arguments seem sound.

This Time Isn't Different - Robin Hanson on the AI Winter.

everything personal can’t be political and incentives for argument, continued. Freddie Deboer is one of the few remaining intellectuals in the academic left, and is a strong opponent of the anti-intellectualism that is currently being pushed.

Apparently they've found many of the genes influencing intelligence. It's more boring and incremental than many people would have thought - there isn't one intelligence gene, there are a lot of "slightly more intelligent" genes.

Sixth Circuit on the ‘rational basis’ for opposite-sex-only marriage laws. Interesting article on the intellectual bankruptcy of many recent court decisions about same sex marriage. Note: I favor marriage equality by ending state sanctioned marriage for all people.

The Rise of the Gray Tribe, by Pax Dickinson. Good to see him speaking publicly again. Also related: Gamergate, 3 stages to orbit by popehat.

Scott Alexander on the Motte & Bailey Doctrine, and related thoughts by the Practical Ethics Blog.

Apparently women are not significantly discriminated against in academic science. In fact, if you assume male and female candidates are equal on average, then it would appear that discrimination significantly favors women. The article covers their bases, and presents evidence that male candidates tend to be superior, suggesting even more discrimination in favor of women. Overall, this is a very thorough and detailed analysis - it should not fall down the memory hole simply because it's results are politically incorrect. Here is a widely linked counterpoint, which mostly seems to miss the point (but has the correct mood affiliation).

Subscribe to the mailing list