You don't understand me - slow-motion and camera angles aren't quite good enough, because the problem is that people are physically using their hands to start and stop the timer. People just don't time that accurately enough to distinguish between the speeds of different people - you can't know whether you stopped the timer at the right time, or 0.05 too early, or 0.05 too late. Since we'd need to know every time to an accuracy of about 0.01 or less to tell which of various fast people is the fastest, the only way to actually find the UWR would be to have every fast person submit a high-fps recording of a PLL, and then to have someone count the frames of each one.