gwern comments on Estimating the kolmogorov complexity of the known laws of physics?

[gwern]

has the disadvantage of leading to less optimal predictions

As far as I know, any computable prior will have that disadvantage relative to full uncomputable SI.