even without any of the overly complicated tech you describe, some journals maintain statisticians on staff who evaluate everything in every paper and then help correct errors before publication. My stats prof, Stan Glantz did this for a major cardiology journal. He said (and I agree, having reviewed plenty of papers myself) that statistical errors that affect the conclusion of the paper were found in the majority of papers.
It solves a problem, just like any other engineering solution. There are trade-offs involved like always. Obviously you understand the problem I'm trying to solve, since you're passing judgement on my suggestion - do you have a better one? Even given all of your verbiage, I'm afraid I couldn't devine one.
