At my previous work unit, I had a number of PhDs working for me. I have a BS degree, so they have a whole lot more schooling than I. We were all very close, and worked extremely well as a team. Part of our closeness was a jabbing back and forth. One of my favorite jabs if/when one of them got to being a little 'particular' for my taste about a procedure was that all this 'only change one thing at a time, write down EVERY POSSIBLE aspect of EVERY MINUTE detail of what you're doing, and ISO-9000-style codify it' mumbo-jumbo isn't because doing things that way is 'better,' but because that way even the monkeys could do it. Now, I certainly didn't think they nor any of our people were monkeys, and they knew that, but frankly the general point held up. We needed to make things so specific and so detailed that someone brand-new and utterly inexperienced with what we were doing could at least come in and not screw things up too badly.
And when I wanted to make the eyes roll of the Brainy Smurfs who worked for me, I'd say "Don't tell me what you think, tell me what you know."
Pharmaceuticals and medical devices are not made by monkeys, but would you rather have a hip implant or take a drug made by a scientifically validated method, or by a guy who thought he knew a better way? I recall a number of knee implants that had to be removed from patients because a raw material supplier made an unvalidated change that turned out not so smart. Would you want to fly in a 747 assembled by standard methods, or one where 2,000 assembly guys with their first names on their shirts all knew they were smarter than "the ties"? At what point would you be satisfied that the pilot of your 747 had flown enough that he did not need the checklist any more?
It is true that standard methods can "dumb down" a procedure to a point where less educated (read: cheaper) people can perform it, but that is by far not the only reason to employ them. Though many ways might "work" (or not), there truly is "one best way". In the current discussion, if actual experience shows that the Honda recommended torque of 22 ft.-lbs. is wrong, they can change (and improve) the guidance in an update. For right now, it is the only number with testing behind it that has been shown to work. Will other numbers work? Probably, if they are close. Will guessing work? Maybe. But why would you? How high are the odds that your guess or feel will be better than the tested value?
It is not a 747 or a drug, but it could be a stripped engine case or an oily rear tire. Rack 'em up.