1+1: Normal Two-Dimensional Singularities, RenTech, and Forking Paths

[I’m (trying) to write a short post every day.]

Renaissance Technologies is one of the most inspiring businesses. Not the first, but arguably the most successful quant fund in history: ~40% annualized performance for 30 years, after fees. I’ve been interested in how technology, data, and AI can radically change private investment for a while. (I wrote this post Can You Build a Quant VC Fund back in 2018.) I want to update it now for the post-LLM world. I’ve been talking to ChatGPT 4o about ideas for the updated post and it is failing miserably (the worst performance of AI I’ve experienced - curious).

I have a small bookshelf in my office where I like to keep books that I often keep coming back to for reference (think Using Language by Herbert Clark, The Embodied Mind by Varela, The Wealth of Networks by Yochai Benkler). I recently added to it with Henry Laufer’s 1971 work Normal Two-Dimensional Singularities. I am quite lost (although I like the feeling of having to find my way). I’m relying heavily on ChatGPT 4o to understand the work, but that’s not really the point of it. Henry Laufer was the Chief Scientist at RenTech for 20 years joining only a couple years after Jim Simons started the fund. It’s just a cool reminder that a lot is possible.

For those curious (via 4o): A normal two-dimensional singularity is a specific type of singular point on a complex surface (a 2-dimensional complex manifold or variety) that satisfies certain technical conditions making it relatively “well-behaved” from the perspective of algebraic geometry or complex analytic geometry.

N=1: Starting at the start

I’m going to (try to) write a short post every day.

There are a number of examples of people doing this with great success: Jerry Seinfeld’s “don’t break the chain” method for writing new material, Fred Wilson on avc.com for years and years, even David Hockney has shared some fun, but seemingly deeply iterative, work in progress. (Below is my favorite from Hockney: Untitled 10 from his “iPhone Drawings”. This work was done in 2009 only 2 years (!) after the phone was introduced, pretty remarkable early adoption. Also it was presumably his 10th iteration of something. It’s quite strong motivation, things don’t need to be perfect, even if you’re a world famous artist.)

I’m also a pretty strong believer in the power of improvement via iteration. There’s a now famous study from Art & Fear that discusses the differences in quality of ceramic pieces generated by a group that was told they would be judged on how many pots they made, and another group told they would be judged on the quality of just one pot after hours of theorizing about how to make the perfect pot. Surprise, surprise: the best pots came from the group that just made as many as they got. Time to start making pots.