Quantum Algorithms 30 Years After Shor

I'll reflect on our knowledge of quantum computing speedups, nearly 30 years after the discovery of Shor's factoring algorithm. Some people expected Shor's algorithm to be followed by a profusion of other quantum algorithms achieving exponential speedups, for example for NP-complete problems, and were disappointed when that didn't happen. Was there a deep reason for this? Discounting hype, and insisting on a fair comparison against the best known classical algorithms, what can we say in 2023 about the structure of problems most amenable to large quantum speedup?