ER posted this at the thread on Space/Science on engineering failures:

The natural engineering response to a problem is to build in a fix. But the more fixes you add, as you point out, the more complex things get, and the complex things get, the more likely things will go wrong, if not by an outright failure, then by confusion of the operator.

In any system, the optimum level of complexity is the one that is least likely to fail, not the one that can handle the most problems.

During the Vietnam war, fighter- bomber pilots were provided with so many defensive systems, alarms, warnings, indicators, sensors, “options” and other gadgets that studies showed the most successful pilots simply turned them all off and flew “by the seat of their pants”.

It’s not that these systems didn’t work, they were actually quite effective. It was just that it complicated the pilot’s decision-making process, confused him, and in times of stress he often misinterpreted signals or made mistakes. To a certain extent this can be overcome by training, but there is a law of diminishing returns.

Everyone knows about KISS, or to use the sailor’s version: “If you haven’t got one aboard, it can’t break when you most need it.”

There is little to argue with here.

Now think about government regulations, each of them a “fix” for a real or perceived problem, most of them added on top of countless regulations that were there before.

Nobody ever said that “fixes” are never needed. The problem is the attitude that 500 kludged fixes are better than 5, and if you want to make it even better, add 5 more.

At some point in engineering you often have to go back to square one and take a different route, particularly when the current solutions are many years old.

And as always, the fundamental touchstone: “Does the solution work?”