There should be value in increasing the class size if (a) the number of applicants increases and (b) the average quality of the top applicants remains at least constant.
During YC's history so far, we know (a) is true and (b) seems anecdotally to be true.
If there's only one winner in a class, at best, does it matter if the class is 10 or 50? 50 isn't going to give 5 winners, is it?
Is the idea that 50 gives a 5 times higher chance of finding the 1? Is that really true, though?
Would focusing on a smaller number of potentials actually be better for finding the 1? Thinking through the scenarios more thoroughly?