type theoretic sandwhiches

you learn statistically
do you define stuff by it’s effects? what you can interact with in the sandwhich/hot dog
- we can choose a model for which they are the same (both breaded, both filling)
we can train a model to recongize flowers by watching videos of bees, but the model is then trained on the visibile wavelength whereas bees are trained on the UV wavelength
category theory is constraints given form
- which is why programmers love this shit
we’ve precisely defined the base, but we haven’t done anything that is generic
math is generic though
how can we better ask “what is a thing”
- foundations of mathematics
meaning out of reference
- meaning is defined as a use of a thing
- latent spaces