At the Milwaukee School of Engineering (MSOE), where I worked for 28 years, we had a lot of projects along the lines of what Bob Borrelli did at Harvey Mudd. We probably did somewhere in the neighborhood of 1000 projects: a large number, and during my first two years at Super Steel, we probably did 20. One thing that I would encourage people interested in these projects to do is to identify up front the project variables. Hearing a little of Bob's horror story brought to mind the importance of defining, very clearly in writing, what the variables are and also what the milestones in the solution process might be. One of the things that we want to know is where we're at, right? Avoid the kinds of projects where the client asks you to develop a new process to test such and such a thing. Those are really very, very difficult to do.

At Super Steel we have been successful about 75% of the time, but the 25% failures contain the seeds of success if one learns from them. A very good friend who worked for many years at research had an attitude about failure. Every time his group failed in a research project, they'd have a party to celebrate the failure. Now this had to be an honest-to-god failure, not just giving up on something, but a genuine failure and the reason was that, on average, one out of every 500 of their projects would fail, so that when they failed, they took the attitude that their failure was behind them and they could move on to a string of successes.