Not sure if there are books specific on that. I picked it up as part of Six Sigma training. The concept is you list all the possible failure modes in one colu

Not sure if there are books specific on that. I picked it up as part of Six Sigma training. The concept is you list all the possible failure modes in one column, then score the severity (1-10, 10 most severe), occurrence probability (1-10, 10 highest probability), and detection probability (1-10, 10=lowest probability) in other columns. Multiply those to rank the risks and then list out ways to improve or fix the risk, etc. A software angle would be listing ways your application would fail, or go down to lower level components inside the app. A tricky concept with the scoring is the detection piece. If you don't have a good way to detect the failure then it ends up being really high ranking. I typically used 1,5,10 for the scores since ending up in a debate about is it a 6 or 7 or 8 is a waste of time. Once you work on improvements for a failure mode, like additional logging for an NullPointerException, then adjust the scores. If you are doing TDD, this would be a good way to identify areas to build testing coverage.