> but this doesn't mean induction is unformalizable,
> it just means that the formalization of cognitive-science
> induction in terms of algorithmic information theory
> (rather than experience-grounded semantics) is
Imo, induction only works when the complexity
of the data is not larger than the complexity
of the theorem (or the model, or the theory,
etc.) we wish to prove. In other words, if those
data are special, induction does not work.
Isn't this another parameter for that formalization?