The evolution of a manifold under Ricci flow is deterministic as mentioned, and it is also irreversible, ie many manifolds are smoothed to the same manifold. It's also a diffeomorphism, meaning that it preserves all of the topological information of the manifold that it acts on. Furthermore it can act on manifolds which have no normal form, in at least some sense. Because smooth manifolds with no triangulation do indeed exist.

The manifold cannot, in whole, as a result of this, get "simpler" or even maintain the same amount of simplicity, it must somehow get more complex, which would require information to be generated. The primary problem with deterministic models, at least as I understand it, is that there's no obvious way for deterministic processes to get more complex.

Cranks to the left of me, jokers to the right, here I am, super determined with you...

Even the comments on the blog post seem exceedingly cranky. I can imagine it's very hard to work in such an area, with such a noisy peanut gallery.

Because Perelman.

I mean he wrote Perelman’s name so he must be onto something.