This is a rather general problem, and here's my general solution: Currently(0), we use the same virtual grid for visual layout as for mouse/touch interactions.

This is a rather general problem, and here's my general solution: Currently(0), we use the same virtual grid for visual layout as for mouse/touch interactions. If there's a box displayed visually, then there's a corresponding box that's available to receive clicks. If that box moves 100px down the screen, then the click-receiving box also moves 100px down the screen immediately. If you clicked the original box's location just as it moves, the computer replies with something like, "hey idiot, why did you just click empty space?" Why can't the computer be a little less of a bully and recognize the intent of the user by looking into the past a little to see what the user tried to click? Instead of keeping these grids in lock step with one another, the click/touch-receiving grid should be delayed by 10-200+ milliseconds (and tunable -- esports players on one end and elderly/disabled folks on the other). This grid from the past is actually what you intended to click. Back to the example above, if you click the box just as it moves, the click is now checked against this slightly-old grid instead of the current (suddenly changed) visual location of the box, and the computer correctly detects that the click was aimed at the box. 0: in any system I've worked on; I imagine some system has tried to make this better, though?