we've come a long way baby!


Systems Engineering


Most of the time people are so enmeshed in the details, they fail to see the bigger picture. It's the essence of bureaucracy. Systems engineering is an attempt to look at the whole as a whole. It's very hard. You don't pay attention to what systems engineers say, you pay attention to what they produce.

It's very easy to say the right words in systems engineering - it's very hard to practice it.

The first rule is: if you optimize one component, system performance degrades. It sounds strange, but it's very common.

"The whole problem is too big to think about, therefore let us do what we can do" - typical technician. If we don't know what the output is supposed to be, how are we supposed to do a good job? With difficulty.

Some systems are just patched together, like the railroads.

The telcos were in a strange position: they not only designed a huge system, they operated it. Suddenly reliability, maintainability become important and it factored into design.

Second rule: systems are constantly being adjusted and changed. Part of a system engineer's job is to prepare for changes, even at the R&D level.

The closer you hit design conditions, the worse the overload. Build in some graceful degradation.

Technicians must be given opportunity to sharpen their skills. Since system engineering is never done, there's always a crisis to solve.

The individual's idea of what they want, and what you're trying to do with the system as a whole come head on into conflict.

When management changes their rules for promotions, you'll change your behavior to fit - regardless if it's good in the long term.

The design, evaluation and modification cycle is stopped in order to produce because it has to be, but it's a very hard thing to do. The client has some knowledge of his symptoms, but no knowledge of the cause. They are likely to be quite demanding. Part of the system engineer's job is to try to uncover the problem. There's neither a definite problem, nor a definite solution.

Attractive as the idea is of a course in systems engineering, it can't be done. There are too many variables and changing conditions. A course would be a lab essentially - and what works in the lab often doesn't work in the field.

The org who uses a system has a huge impact. In a high-morale org, people work overtime without fussing. In a low-morale org, they're clockwatchers and goof offs, even in an emergency. So a system or product that works okay in a decent org, could go to hell in a crappy org.

You learn by doing, including systems engineering. It's like sex: all the books in the world don't equal real knowing.

It's better you have a poor solution to the right problem, than a good solution to the wrong one. So the goal is to figure out the problem - and it's often changing.

It's the patient who has the agony. You do have to listen to him, but you also have to put more in. You must try to find the cause.

Education is a broken system. What is the educated person? It's a changing thing in the first place, but it's difficult to pin down. Complaints are clear and easy, but what are we trying to do?

The trouble with systems engineering is you are also required to think about the human beings. Human beings are all eager to optimize their performance on their problem, and they don't care one bit about the system as a whole.

Solving the conflict between the individual and the group in favor of the group is inhumane. You have to let the people get away with some things and act humanely. It doesn't work when you expect people to act like machines.

The nature of the problem is: vague problems solved vaguely.