If you think the moral of the story is that $8M were wasted due to bad management or whatever you'd be wrong. The scales were a success because they enabled the desk fan solution. Manufacturing line operators are very clever but they don't necessarily have the right incentive. The loss to the company wasn't visible but the horns and flashing lights were.
I used to build inspection systems for manufacturing lines and we'd frequently build expensive (not $8M, though) vision systems, E-Testers, etc. just to uninstall them a year or two later when the process or automation guys had found a way to prevent problems from happening in the first place.
Not just incentives, but also cognition - when you change the presentation of the problem, you change the way people think about it.
I think this is the reason to build prototypes that you can play with - you really will see things differently. And (hold the hate) obviousness in patent law takes this into account, by evaluating it with respect to the publicly known state of the art - but if you have iterated a few prototypes then you have private information. You got data that no way else has; in addition, new ideas came to you for the next prototype. Each step may have been pretty obvious to you - but from the perspective of someone who had not yet even seen the first prototype in chain, you can seem like a magical genius.
Ok, good point. A better way to incent is to offer bounties for yield improvements. Most manufacturing outfits I've seen have those but they are probably too low.
Here's another problem mentioned in the article, in fact it's maybe the real problem: the engineers "who were already stretched too thin" probably didn't talk to their operators and technicians often enough. When I was a manufacturing engineer I noticed how much I learned about the stuff I was building by simply hanging around the operators for significant amounts of time. Not only did I get a lot of information about the subtle ways my systems were failing, I also learned that the operators generally felt engineering was ignoring their concerns. Just listening and trying to make the operators' life better earned me so much cooperation from the floor I later had no problems getting them to try out new things for me and providing me feedback on "improvements".
I used to build inspection systems for manufacturing lines and we'd frequently build expensive (not $8M, though) vision systems, E-Testers, etc. just to uninstall them a year or two later when the process or automation guys had found a way to prevent problems from happening in the first place.