Irrelevant? No not at all! Maybe it doesn't change the answer, but it is still interesting (well to me anyway). In fact it seems to me that the non-existence of the expected value, and the fact that you can't use the central limit theorem and everything that goes with it, is the only thing that makes this an interesting question at all.If I was going to answer the original question I would first try to calculate the expected value, which is infinite. If I did a simulation to find the expected value I would get a meaningless result (unless I actually did something wrong in my previous posts), even with a really good r.n. generator, because the sample average will always exist but the expected value doesn't. So then I would try to find the most likely value of the sample average of 100,000 matches. This I think is what you (Zerdna) did, although maybe I misunderstood that as well. I dont know if using simulation to find this maximum is affected by the same problem. Would it matter if I was looking at the sample of 100,000,000 matches? Or 10? What if I had a really bad random number generator?. I was hoping for some comment on this and maybe some more general comments about this problem with random number generators. For what it is worth, I did a simulation in Matlab, and I get more or less the same graph that you did with the most likely value at about 25. My results give me an expected value for my simulated distribution of about 40ish. Since this is wrong, how can I trust the rest of my distribution?Am I the only person here who thinks this is interesting?Maybe I should have started a new thread.
Last edited by spacemonkey
on July 31st, 2004, 10:00 pm, edited 1 time in total.