starting a notebook

This is the first piece in what I hope becomes a real notebook. To me, a notebook is a place for thinking that’s developed enough to write down but not finished enough for anywhere else.

I’m planning to write about three kinds of things mostly. The first is something I’ve been trying quite hard to make sense of. It’s the way actuarial work seems to be splitting into a classical track that codifies what the profession has always done and an integrated track that meets modern technology head-on. I don’t have a settled view on which path makes more sense, or what the integrated path actually looks like in practice. Working through it on paper might help.

The second is emerging risk, particularly the methodology questions around AI agent liability insurance, where the data is sparse, the failure modes are novel, and the existing actuarial toolkit only half-applies. There seems to be more written about whether AI risk is insurable than about how you’d actually price it, and I’d like to try to think through some of the “how” without much confidence that I’ll land somewhere useful.

The third is just whatever I happen to be reading.

A few things on the runway:

• What “uninsurable” actually means. Trying to separate genuine blockers from things the market just hasn’t figured out how to price yet.
• Modern MGAs. What the model lets newer entrants do that traditional carriers can’t, and where the limits of that show up.

If this has been sitting untouched for months by the time you found it, oops. The rest of the site is perhaps a better record of what I’m up to.