Nice research, Doc.
A word on my (admittedly eccentric) propensity to view and/or discuss CAREER minor league lines. Actually, many, many words - as I am many things - but brief is generally not one of them.
I like to view minor league career lines because it is the LARGEST data pool for the prospects. I completely understand it doesn't tell the entire story - but for me, it provides some context that examining level-by-level progression doesn't give me. (It's also readily available). One of the big plusses is that it gives me a comparison for a given player between a specific year and his entire minor league lifetime WITHOUT having to concern myself with order.
This doesn't mean order is completely irrelevant. But, I have a sore spot in regards to prospects, because the group mindset of prospect watching is a foundation belief that development is LINEAR. This is because all the research into large pools of minor league populations ends up MAKING them linear. This leads to the perception, (and belief), that players improve "because they get older."
I have railed more than once that players do not improve by aging -- they improve by fixing SOMETHING that was broken, or by adding a new skill that they previously lacked. It's been my experience that when spec-talking, the "hottest" specs are the ones who follow the development actuarial charts EXACTLY, (at least up until the point they implode). I think for many prospects, the perception is polluted, because of the very human trait of wanting to "be right." When Prospect A follows the "expected" progression, excitement builds. When Prospect B goes "off script", he is quickly dismissed, (if it is off-script bad) -- OR, enthusiasm explodes exponentially if he is off script good, (A.J.)
Prospects are "supposed" to dominate in A ball and AA ball -- maybe struggle a tiny bit in AAA, and then transition smoothly to the majors. If it is a "quick twitch" type, then the need to dominate in A ball is removed, but said player has maybe two years of learning baseball before prospectdom abandons him and moves on to the next pheenom.
I think most people simply don't LIKE messy, erratic minor league career paths, (even though these are more common than most people realize). The human mind WANTS order - so we're naturally more comfortable with order than chaos. And frankly speaking, learning how to hit a round ball, with a round stick, square -- well that is a chaotic process.
You know what is amazing to ME? That a *MAJORITY* of major leaguers actually outproduce their minor league career lines. For most, it's not a big number. But, just as a sample set, I went and looked up the best team of all time, the 116-win Seattle club.
OPS comparison
Player ------ minors - majors
Dan Wilson - .688 -- .691
Brett Boone -.831 -- .767 **
David Bell -- .697 -- .716
C. Guillen -- .748 -- .808
Al Martin --- .745 -- .783
M. Cameron .774 -- .788
E. Martinez - .851 -- .933
M. McLemore.714 -- .690 **
Olerud had no minor league line to speak of and I just don't even want to visit the Japanese conversion discussion in this instance. TWO of the guys on the list had superior minor league lines. Most of the differences were pretty small. But, the minor-to-major conversion charts ALL are focused on the AAA numbers translating to *WORSE* major league numbers upon arrival. This is common. There is a LEARNING process. Clearly, Boone's peak years in Seattle beat the snot out of his minor league line -- but it was a brief peak in a long career.
What is interesting here is that I bet in 9 of 10 cases, fans get excited mostly about the prospect that puts up a .950 line in AAA, (the Adam Joneses of the world). They ignore the CAREER minor league line, (typically a better 'tell' for career path), and concentrate on the SMALLER data pool -- the one great season as the foundation block for believing in the impending greatness.
MOST players exceed their career minor league lines. Most do NOT exceed their minor league PEAK seasons. Typically, somewhere along the line, the player puts it all together (in that environment), and puts up great numbers. But this is sometimes in A ball, sometimes in AA, and sometimes in AAA. Of course, when it happens in AAA, we naturally assign greater value to it. But, Boone's .895 in AAA didn't define his major league career and more than Carlos Guillen's .896 in A ball.
My problem is that while I understand that expectation of linear development -- the basic reality of smaller data pool seems to get tossed out completely when talking prospects. I like looking at career numbers because it's the biggest dataset. I don't HAVE to worry about whether a given year is a career year -- it is MUCH easier (and less dangerous) to assume that being the overall picture, that the streaks and slumps are likely to be as even as I can ever expect them to be.
AFTER I set my foundation -- THEN I can start looking at whether I think there is progression, regression, signs of stagnation, etc. My spidey-sense gets uncomfortable when discussions of prospects seem to begin and end with only what is the latest tidbit of data. "Joe Schmoe crushed AAA (for 95 ABs), therefore he's going to be the next Bonds!"

Good points around, Sandy. As if we've come to expect anything else ;)
My sole point to make is that AAA numbers are *generally* going to provide the most accurate picture of a player's ability (at least when you split out minor league numbers) because of attrition, and because of the proximity of the data. If it's the most recent data set, that suggests that the majority of the improvements and shortcomings have been observed statistically, at least in some fashion. But your assertion about people going bananas over a James Loney posting a .380 BA in AAA for a season (comPLETEly out of line with the rest of his minor league career) and predicting annual batting title contention as a result is absolutely correct.
The other thing to remember, which has been discussed in previous threads, is that the most recent data can indicate a fundamental shift in some aspect of a player's skillset, whether it be a decrease in overall speed which necessitates a slightly different approach offensively, or a mechanical change such as a stance or delivery alteration. But outliers should be regarded as such until proven otherwise, at least in my opinion. Not disregarded entirely, but you don't project forward that Beltre is going to hit .320+ with 40+ HR's for the rest of his career simply because he did it for one season. You *can* assert (with relative confidence) that he will have reached a new plateau offensively, and project modest/significant improvements in performance accordingly.
Funny that you should mention MLB career lines outperforming MiLB career lines. This was something I'd noticed while scanning through thebaseballcube.com's profiles looking for a comp for Carp. How much of this is simple attrition, and how much of it is actual continued improvement? I have no idea, and I'm not educated enough to go about finding the answer. But it is an interesting observation, nonetheless.

mj,
My theory on why major leaguers tend to outperform their minor league lines --
POINT OF DEVELOPMENT -- players at a given minor league level "tend" to be the same age and have roughly the same experience, (until you begin seeing the journeymen) - but the point being that (aside from talent), most of the other variables for minor leaguers are fixed - or at least pretty close. So, you are competing in the minors with players on roughly the same skill level *AND* at roughly the same EDUCATIONAL level.
BASEBALL IS HARD -- While fans and pundits focus on the physical, baseball seems to have the highest LEARNING curve of all sports. You can have 16-year old NBA players. You can have 19-year-olds capable of excelling in the NFL, (though SIZE is a major limiting factor there), and 21 is the standard professional starting age for both of these sports. Let's not even think about the dominance of teens in Tennis.
But VERY FEW 21-year-olds even get a chance at the majors. You've got a "gem", if a kid is deemed ready to STRUGGLE in the majors at age 23. Golf is about the only sport I know where the learning curve seems equally daunting.
It is my opinion that the failures to reach the majors do NOT fail PHYSICALLY. They fail mentally. They cannot adapt to the complexities of the pros. Most have thrived by simply relying on physical reactions. When that fails -- *MOST* stagnate or regress. The ones that are capable of playing in the pros are the very ones who have the ability to adapt their abilities to the changing conditions.
In the minors, most are getting their first taste of personal evolution. (I think this is why college players are safer bets - because the ones who are unable to adapt weed themselves out). It is completely reasonable that as you take the first steps into adaptive athletics that most players WILL struggle. This means for some chunk of development time, the individual will be playing UNDER their raw skill level. When they are experiementing with a new stance - or new swing - or new pitch -- they'll flounder.
But, getting to the majors means that (for most), they have ALREADY proven their adaptability, and they've increased their skill sets from where they began. It is NOT the ability to beat up on inferior athletes that defines them as major leaguers -- but their ability to improve themselves and better leverage whatever athletic skills they are blessed with that allows them to make it out of the minors.
Of course, they're all superb athletes, (compared to us schmoes in the stands), because you cannot hit a 95 mph FB just by being bright and coachable. And many players fail because they cannot keep up athletically.
I would suspect that the guys who tend to underperform their minor league lines were probably already leveraging their skills in the minors. I would suspect that, as a group, they were always viewed as the "smart" players, but not necessarily the elite athletes. So, they didn't have the same learning trough that others had. But, much of this is conjecture, and my theories could be wildly inaccurate. What isn't in dispute is that the AAA to MLB conversion factors that are popular with prospect watchers for year-one prognostications do NOT map well at all to AAA-career to MLB-career conversions.