Originally Posted by Madory

This makes sense, thanks.

Perhaps it is a case of 6 and one-half-dozen. I got my explanation from "Attacks On RC4 and WEP" by FMS:

"The probability that three locations will not be pointed to by a pseudo random index during the
remaining N - 1 - x rounds is better than ((1-1/N)^N)^3 ~ e^-3 ~ 5%."

((1-1/N)^N)^3
can be reduced to
(e^-1)^3
and finally
e^-3

-OR-

(1-3/N)^N
reduced directly to
e^-3

Anyway, thanks for the general formula - crystal clear now.