I agree with you to a fair extent. I dont agree with the degree to which some ADCOMs adhere to this (we must average no matter what) policy and I personally dont think you an ADCOM is interpreting the "evidence" from the AAMC studies the best way if this is your conclusion(nor is the AAMC if that's there conclusion).
The problem is you never know these "factors" behind a score. You can never really say if someone goes from 30 to 37 what caused the jump? Poor preparation the first time? Will that poor preparation/mentalty to such a big test carry over to Step 1(many people take the MCAT only 3 years before Step 1 so it's not like it is a huge jump in time)? Does the 30 reveal something about their ability; ie did they get lucky to some extent on the 37 the second time? You cant answer these questions really. But they exist and are more prevalant and likely to be relevant than somebody who takes the MCAT once and gets a 37 flat.
What I will say is by and large this "averaging " policy in theory shouldnt affect that many people(again in theory). The data shows only 30% of people who retake any score between 21-35(give or take 5%) show a 3+ point improvement. So even the person who goes from 29 to 32 and is averaged at 31 is applying with a 30.5 vs a 32. That 1.5 point difference really isnt going to decide someones fate. Now if there are schools where a 30.5 vs 32 is looked at as rather different, I find that an issue with the school and what they are valuing more than the average the MCATs policy. But that's a separate discussion and like I said "in theory" if admission is a reasonable process where we dont look upon a 30.5 vs 32 as significantly different at all, it wont have much bearing.
How would making the MCAT P/F make it a better predictor of anything? Im assuming that's what you mean by saying "passing". As it is we barely have any way of standardizing applicants.
"Succeed"= graduation by that logic. But that's not what it is. Graduation is the absolute bare minimum. In medicine, bare minimum is the worst possible standard imaginable. That doesnt include the number of people who fail a board exam along the way, remediate a year, fail a clinical rotation, get a bad review from at least one attending, fail a shelf exam etc. Bascially graduating doesnt remotely mean that you didnt have problems meeting the bare minimum competencies required along the way. And even if you graduate, having problems meeting that training is a cause for very real concern. And much more than that, many residents/attendings who struggle or who you could say are lacking in key areas of the job didnt fail any of these things along the way or often didnt come close to failing.
You can read this link to see the data between academics and graduation(start at page 44). Basically if you just want to define "graduating" as success there is no point in even using academic metrics. The solid majority of 2.7/21 people who enter medical school will graduate.
https://www.aamc.org/download/434596/data/usingmcatdata2016.pdf