It does seem safe to say that technically they really dropped the ball. The inadequate estimation of server load combined with the (apparently difficult) task of adjusting to the actual load suggests at least a few groups of people didn't do their jobs properly (or the company denied them the resources to do so).