For a fixed graph H, what is the probability a random graph G(n,p)
contains more than twice the expected number of copies of H. In this
talk I will sketch the answer to this question in the case when H is a
clique. A conjecture for general H will also be presented. (Joint work
with Jeff Kahn)