A classification system doesn’t passively observe the world — it reshapes it. Credit scores change how people use credit. Auditing algorithms change how people report income. The instrument and the thing being measured are not independent, and any analysis that treats them as independent will overestimate accuracy and underestimate behavioral effects. This tag follows the formal framework developed in joint work with Maggie Penn, and its real-world applications.
Roosevelt Elementary started Year 1 of Elevate with 100 students. It ended the year with 400. The other two schools held roughly steady. The district grew from 600 students to 900, and the composition of that denominator shifted decisively toward the lowest-scoring school. That one fact explains the entire dashboard. Elevate worked. Every school’s average … Read more
Every home and office has one. You know which drawer I mean — the one with the takeout menus and the dead batteries and the rubber band that used to hold something together and the key that opens something, somewhere, you just can’t remember what. It is not a disorganized drawer. It is a perfectly … Read more
This is the third of three posts this week on prediction markets. Monday was about information — the signal that travels before the message. Wednesday was about jurisdiction — the classification question that reproduces itself at every level you try to answer it. Today is about something prior to both: whether the classification of a … Read more
The Financial Times ran a piece last week reporting that Senegal — yes, that Senegal — had borrowed €650 million through instruments it had not disclosed to the IMF, to its existing bondholders, or apparently to anyone who might have objected. Which, it turns out, is everyone. I clicked on it because it was about … Read more
Let me tell you something you already know: ChatGPT, Claude, Grok — whatever flavor you’ve adopted — is very, very nice to you. Suspiciously nice. “Your presentation looks great, here are a few minor suggestions” nice. “That’s a fascinating question” nice. “I can see why you’d approach it that way” nice. You know this. You’ve … Read more
It’s been almost ten years since I’ve written here. The last time I posted, Donald Trump had just clinched the GOP nomination, his Banzhaf power index had hit 1.0, and I was calculating the proportion of his campaign contributions that were unitemized.1 That was June 2016. I stopped writing because the general election demanded a … Read more
We now live in a world of seemingly never-ending “data” and, relatedly, one of ever-cheaper computational resources. This has led to lots of really cool topics being (re)discovered. Text analysis, genetics, fMRI brain scans, (social and anti-social) networks, campaign finance data… these are all areas of analysis that, practically speaking were “doubly impossible” ten years ago: neither … Read more