Lattices vs. Cranks

If I'm thinking of the same 'crank' that the author is, I believe that the main conteibution he claims is a novel way to compute very distant parts of the keystream without having to work through the preceding digits with Newton's method. But this author dismisses it as even slower than Newton's method, and susceptible to brute force in any case. That would be a shame, since I was drawn to the work of this 'crank' a while back in trying to approximate algebraic numbers using sublinear memory.

Hm, does this kind of approach also work for the "recover the algebraic expression that generated an approximated float" problem?

Readers who enjoyed lattices vs cranks may also enjoy:

1/2 Building Lattice Reduction (LLL) Intuition (2017 blog post): https://kel.bz/post/lll/ 2017 HN thread https://news.ycombinator.com/item?id=14847560

2/2 the cryptopals crypto challenges: https://cryptopals.com/ 2014 HN thread https://news.ycombinator.com/item?id=8166064

Cryptopals problem 62 "Key-Recovery Attacks on ECDSA with Biased Nonces" from set 8 "abstract algebra" applies the Lenstra–Lenstra–Lovász (LLL) algorithm: https://cryptopals.com/sets/8/challenges/62.txt

I think this post is about how it is easy to find "a" if you have the value alpha=frac(sqrt(a)). Someone somewhere on twitter was using this for a crypto idea but no link was provided to the post or documentation.