And a single coin may roll under a dresser, fall in a drain, be flattened by a train. This makes little difference in the difficulty of computation.
I don't think it's a very good example as it doesn't generalize well. For example, it's nearly impossible to predict the weather for next July 7, but the weather averaged over all July is fairly accurate. Likewise, I cannot predict a single roll of a roulette wheel; but I can predict the outcome of many throws.
The point in the original post is that the process of solving a problem is easier with fewer factors to consider: