That’s not a helpful way of looking at it I think. Ideas don’t exist in isolation, and thus old experiments have to be carefully analyzed to extract what’s still true out of it. Often times it’s just cheaper to redo the experiment.

This is true across the board for engineering. It’s hard to build something that survives long prolonged contact with engineering realities. Computer architectures change. How things are done efficiently changes. The set of trade-offs we choose to make changes as the underlying business shifts or better business models are discovered. And heck, businesses can be cyclical for no better reason than humans are naturally cyclical in trends (see fashion)