He tried many fixes. He did everything short of a redesign and short of replacing components that he didn't test with tested ones. Those untested components had come from dear and loyal friends, and Charlie didn't want to disappoint or insult them by suggesting that they shipped inferior components. He even tried to use more lenient testers.

And now Charlie saw the solution. The idea came from none less than the U.S. Department of Defense, which had been encountering similar problems. The DoD's experimental missile shield kept missing. It failed to intercept incoming targets. Sometimes it located the target, but the firing mechanism didn't work. It was often just a trivial problem that would be easy to overcome.

To the dismay of the contractors that profited from its development, the missile shield did not shield very much. It failed just about every test.

The failures were not absolutely consistent. If an "enemy" was sporting enough to alert the "defenders" as to the location and timing of a "missile," and if that enemy was considerate enough to avoid sending distracting duds and chaff along with the real missile, the missile shield just might hit an incoming target.

Then along came this wonderful solution, from Donald Rumsfeld, the Secretary of Defense: Build now; test later. "I don't know a single advanced research and development project in the history of mankind that didn't suffer a series of failures," he said. "You end up learning something by trying."

What a terrific idea, Charlie thought. Now if only he could get American taxpayers to put up a few billion dollars to build his product, then a few billion more if the first try failed. After all, his friends could use the business.