A. 2. 1 micro-second is 10^-6 seconds (not 10^-9). G. a) The proof that I gave in the original solutions just proves that there exists at least one configuration of chips where #bad > n/2 where it is impossible to tell if there is a good chip. Here is my original (and correct) answer. (Sorry about the seperate file for the Bug Report, but the original solutions file is no longer with us). Assume there are g < n/2 good chips then any subset of g bad chips can conspire by claiming that every chip in this subset of size g is good. We do not care what the rest of the bad chips do. If we assume that we can do every possible pairwise test (this is the best we can possibly do) then we will end up with two subsets of size g with no way to tell them apart.