Assuming it works, I'm interested in its limitations. At various complexity levels, there are subsets of problems that yield easily, like unsafe primes in RSA, or trivially, how I can instantly get one factor of an even composite.

Schnorr's main novel claim here seems to be a speedup in finding the SVP and CVP in some cases (he explicitly acknowledges limitations.) A Proof Of Concept seems like it would be great to test for edge cases, and that's where I think the interesting bits are likely to be. Disclaimer: Not a mathematician or complexity theorist. Just here to learn, and glad to be corrected any time.