mondrian cake

Modern Art Desserts - a very cool book which I take some inspiration from.

Where Superforecasters Start - tl;dr; the base rate.

Statistics and Computer Science

JavaScript and fast data structures: some initial experiments

Why do we make statistics so hard for our students? Discusses how basically all hypothesis tests have the same structure, and our overreliance on formulas makes that hard to see.

An investigation of the false discovery rate and the misinterpretation of p-values. This article is about plausible estimates for false discovery rates, based on the assumption that researchers use a p=0.05 cutoff value. (Hint: it's high.)

Bayesian Reasoning and Deep Learning. Very interesting, from one of the DeepMind guys.

Why Do High-Frequency Traders Cancel So Many Orders? By Matt Levine, so more or less self recommending.

A Simple Differentially Private Procedure. Describes how running the bootstrap, one can make a statistical procedure differentially private. Also, relatedly, Differential privacy in a Bayesian setting through posterior sampling describes how computing a posterior and then sampling it can result in a differentially private result. This paper is very readable, and actually generalizes differential privacy in an interesting way.

Everything I Needed to Know (About Publication Bias), I Learned In (Pre-) Kindergarten

Measuring Polymorphism in Python Programs

Economics and social science

Sanity for Sociality: A Theory of Religion. This is a reflection on religion, portraying it as social benefits (look how awesome Mormon life is) at the cost of some sanity. The unanswered question: "...for those of us sitting outside wishing we could be inside, but unwilling or otherwise unable to pay the sanity cost, the question arises: why do all religious, especially the good ones, require a sanity ante?"

Closing the Black-White IQ Gap Debate. This is a very nice article which digs up data on black American academic performance, comparing it to African immigrant academic performance. Genetic theories of IQ predict that black Americans (genetically 10-50% European) should outperform their African counterparts. Turns out that reality behaves the opposite way. Anecdotally, this makes sense - my ex-girlfriend's (Nigerian) dad sounded like the sort to say "Hepatitus B? Why not Hepatitis A?"

Anatomy of a ‘liberal’: Indian liberals do not offer solutions, they use the term as a class marker. Some very spot on observations, which also apply in modified form to the US. I'm amazed that I'm linking to Chetan Bhagat. Related thoughts about the US b Freddie Deboer.

Living the Good Life? Mortality and Hospital Utilization Patterns in the Old Order Amish This study compares lifespans among Old Order Amish (the anti-technology folks) and regular white people, and discovers that life expectancy increases are unlikely to be caused by large amounts of medical system usage. The Amish live longer than others.

If I was devising a Panics and Bubbles course... A nice reading list about panics, bubbles, and other irrational market behavior.

Why Education - Tyler Cowen comes up with a novel theory why education is productive, but which seems almost indistinguishable from the signalling theory. I still believe the Sheepskin Effect is better predicted by the signalling theory than Cowen's theory, but it is nice to see a new idea out there.


A Modest Proposal for Renewed Imperialism. Big idea: Nigerians think Finland is awesome and want to move there. It's not because of the beautiful weather or scenery - it's because of the governance and social technology. So rather than moving the people, why not move the borders?

Scary Halloween costumes. Apparently a college administrator wrote an email to students saying that if a halloween costume is too scary you can just close your eyes, and that she isn't the costume police. Now the students are trying to get her fired. I am not making this up. Freddie Deboer discusses in It’s my job to take college students seriously

A discussion of the costs of migration, specifically Syrian and fake Syrian economic migrants to Europe. I generally lean in an open borders direction, but I don't think we should ignore the costs of such an approach. We also shouldn't ignore the fact that a large chunk of the "Syrian refugees" are probably neither Syrian nor refugees. This article also suggests that a big chunk of the problem is an unwillingness to enforce non-migratory laws (e.g., anti-rape laws) against migrants, and the fact that many migrants are seeking welfare. My preferred model of migration would be to provide 30 days of bland food and a bed for migrants, anything beyond that requires getting a job and paying for it themselves. And if anyone causes trouble (e.g. running a brothel, or preying on other migrants), the result is immediate and permanent deportation.

Servants without Masters - interesting discussion on the difference between having impersonal and institutional servants (Handy/Uber/etc) and a single human personal servant.

Subscribe to the mailing list