Control Freaks and Widgets
Back about 15 years, I spent some time being a guinea pig for some industrial psychologists at the University of Illinois. They were willing to pay me $18 an hour and I was willing to accept their money. It later turned out that the parking tickets that I received while participating in said test cost me far more than the money I earned, but that's not important right now.
I won't go into the details of some of the grueling psychological testing that was done on me, except to reminisce about a certain experiment that was performed on me and which left an indelible impression on my outlook on life. A certain computer program was presented to me which modeled a widget factory with multiple conveyor belts moving widgets through the factory. At various stations, robot arms would move the partially assembled widgets from belt to belt, until finally, completed widgets would drop into a bin and a tally would be incremented.
The factory ran itself. What was asked of me was to improve the production output of this setup, which I could do by manipulating various dials to control the speed of the conveyors, the robots, etc. I also had the option of leaving it alone as well.
Of course, as the astute readers have probably already surmised, the factory ran best whenever I left it alone. No matter how hard I tried, adjusting the various dials would always have a compensating negative effect elsewhere in the system, either neutralizing any improvement, or even worse diminishing it.
Here endeth the lesson.