There is a fixed (decreasing) budget for research.

What should be the balance between repeating things that have already been done, and expanding knowledge?