# Research runs on tokens. Where should we spend them?

If scientific knowledge is any useful, we should scale it. We display the best questions ― those which deserve the compute.