The problem^ is that the biggest predictor of knee pain/problems is previous knee pain/problems. So you could run for years (or even decades if I recall the research) with no problems but once it occurs it can be hard to recover from or avoid in the future. As someone with plenty of metal in my legs and interior surface damage to my knee joint, I'm acutely aware of my need to find non-running exercise.

^ as I understood the paper I read last year, which I now can't find but it might have been mentioned in the NYT