Earlier quoted context omitted.
That would be true if you were programming your team, but what you're really doing is setting the magnificent machine known as the brain on the problem, which does all sorts of things we can't explain. I'm all for figuring out explicitly how to deliver software better, but pretending that implicit knowledge doesn't exist seems like folly.
In a sense, you are programming your team, in all the things you emphasize and spend time talking about at meetings. Also, there are obvious things that make developers more able to deliver quality software, and there are obvious things that make gamers more able to win games, just as there are with famous athletes. Thinking that there aren't specific predictors of success here is as silly as saying there aren't spec…
I'm saying that the analogy is just an analogy, and doesn't apply strongly enough. What you're doing has strong components of feeding an ML algorithm training data. Yes, people have intentions and can understand processes, but that's as far as it does. We're certainly not writing neuroassembly.
I don't mean to imply that there are no predictors for better performance, but I bet there are fewer than one might think. A lot of the factors interact in discontinuous or nonlinear fashion, making it very hard to just draw correlations and call it a day.