Quote Originally Posted by jemidiah View Post
These are definitely math questions; I'm not sure why this got moved, even if it happens to relate to current events. The probability of at least one rig failing is 1-P(no rig fails). There are 1000*20 places to fail, where each place has a 49,999/50,000 chance of not failing. So, the chance that no rig fails is (49,999/50,000)^(1000*20) ~= 0.6703. The chance that at least one fails is then 1-0.6703 ~= 0.32968 ~= 33%.

The second question is similar, though with 10 instead of 20, and without subtracting the value from 1.
Thanks, Jemediah. I have no idea why Hack moved this thread to world events either. I believe the correct model to use is the Bernoulli process. The probability of failure is surprisingly high, given the assumption that there is only one chance in 50,000 that an existing rig fails in a given year. So, there was about one chance in three that one of the rigs was going to fail during the past 20 years. I'm rather surprised it didn't happen earlier than 2010.

Even if we had built a whole bunch of them recently with a P(Failure) much less that one in 50,000, most of the rest were already out there. And, as time goes by, the probability of failure is likely to be increasing on any of the rigs and that makes things even worse.

As for part (2) of the problem. This probability is surprisingly low if anyone wants to do the math. I predict we will see at least one more major failure prior to 2020.