vetta project

After all this time, I’m finally getting to the stage of knowing enough to know how much I don’t know.  Or in Rumsfeld’s lexicon: unknown unknowns are slowly becoming known unknowns.  Equations, theorems and whole sub-fields of study that I’d previous heard mentioned, or that I’d studied and largely forgotten as they didn’t seem useful, are turning out to be essential for problems I have to solve: optimal control theory, measure theory, stochastic calculus, variational calculus, martingale theory, mean field theory, perturbation theory, kernel density estimation…  if I’d fully understood all these things to start with, I would have been able to knock off a month of work in just a few days.  :-/

As a child in school my intelligence and knowledge was my greatest asset, but as a researcher I’m finding that it’s the main thing holding me back.  If I could double my knowledge and intelligence, I could easily quadruple my productivity.

Anyway, our proto-AGI is now able to learn and abstract spatial patterns, and it should soon start forming generative temporal abstractions.  A backend for securities trading is also starting to take shape.