Let's say you don't tie game mechanics to frame rate. How often should you update the state of the game? 50 times / second? 100 times / second? You need to pick a fixed rate if you want to keep the physics engine consistent. If you make the rate too high the game will not run on low-end machines, so you need to find the right balance.

But let's say you make it 100 times / second. Now between those updates, nothing changes. You can render at 500 FPS, but you'll be rendering the same thing five times before anything changes, so the extra frames are useless. There are ways around this. You could perform interpolation of object positions between the previous state and the new state (but this introduces input lag). You can keep things that don't affect gameplay (e.g. eye-candy animations) running at the full FPS. But none of these things are trivially obvious. So it becomes a question of ambition, competence, and the will to put time (i.e. investor's money) into it. Hence many projects simply prioritize other things.