Given an error rate and a binary (n,k,d) - code I can find out the probability of making a certain number of errors but how do I work out the probability of correcting such errors if I don't know what the codewords are?