Practically, iff browsers still relied upon SHA-1 to fingerprint and pin and verify certificates instead of the actual chain, and there were no file size limits on x.509 certificates, some fields in a cert (e.g. CommonName and SAN) would be chosen and other fields would then potentially be nonce.

In context to finding a valid cert with a known good hash fingerprint, how many prime keypairs could there be to precompute and cache/memoize when brute forcing.

"SHA-1 > Cryptanalysis and validation " does list chosen prefix collision as one of many weaknesses now identified in SHA-1: https://en.wikipedia.org/wiki/SHA-1#Cryptanalysis_and_valida...

This from 2008 re: the 200 PS3s it took to generate a rogue CA cert with a considered-valid MD5 hash: https://hackaday.com/2008/12/30/25c3-hackers-completely-brea...

... Was just discussing e.g. frankencerts the other day: https://news.ycombinator.com/item?id=26605647