When should you assume I died?

If I go for a while sans new posts online — as I did just before today — it could mean that I was just distracted from what I had to post by other matters, or that I ran out of things to post. It could also mean I'm dead, imprisoned, or otherwise greatly harmed or restricted. How are you to know which happened?

"Interval X sans posts" is a piece of partial evidence like any other. To process it requires a prior. If you know me personally, you might encounter me directly, or get strong evidence that I'm alive and free, by other channels. If you only know me from my posts, your priors are formed more vaguely.

Either way, an interval sans posts (E) suggests I'm incapacitated (H) insofar as I post less often when incapacitated (P(E|H) > P(E|¬H). If I died, I'd never post, and so in those cases you'd see arbitrarily long intervals sans posts, so P(E|H) ≈ 1. The complicated part is the rate P(E|¬H) at which you see a given interval sans posts when I'm alive. Such intervals arise as suggested by my past behaviour, i.e. you should expect my future posts, when I can still make them, are spaced roughly like my past posts.

From 2024-016, I published adjacent essays at various intervals, among which the 90th percentile was 13 days. My last gap (24 days) was much longer than that, which casual observers could take to mean I was plausibly incapacitated, at least until I closed the gap.

On the one hand, that kind of conclusion may be the best information you can get from the limited channel of essay dates. On the other hand, that suggests "compare to high quantile of gaps between adjacent posts" is too naive. Notably, my essay "schedule", when capacitated, fluctuates over months, such that more recent gaps could follow a different trend than the long run. Shortly before the 24-day gap was a 26-day gap, after which I proved to still be around. When modelling from my past behaviour, take the more recent past as a more important indicator.