Perhaps the BEST B1G Forum anywhere, here at College Football Fan Site, CFB51!!!

I think the case of the BCS forcing the nerds to eliminate MOV was a classic example of throwing the baby out with the bathwater. 

As I saw it, the BCS leaders concerns with including MOV were mostly:

• It would give NC Contending coaches a motivation to run up scores on hapless opponents. 
• It would permit teams with relatively weak SoS to overcome that by putting up Playstation #'s.
• It can be misleading. 

My proposal was to include MOV but with a cap and also with some consideration of how close the game was. 

Specifically, I would have taken the sum of the point differential at halftime, the point differential at the end of the third quarter, and the final MOV with the following limits:

• Any OT game is scored as a 1 point win.
• The differential is capped at 21 at each point so the maximum is 63.
• If the winning team has a negative total, the totals are adjusted to +1 for the winner and -1 for the loser. 

I think the case of the BCS forcing the nerds to eliminate MOV was a classic example of throwing the baby out with the bathwater.

As I saw it, the BCS leaders concerns with including MOV were mostly:

• It would give NC Contending coaches a motivation to run up scores on hapless opponents.
• It would permit teams with relatively weak SoS to overcome that by putting up Playstation #'s.
• It can be misleading.

My proposal was to include MOV but with a cap and also with some consideration of how close the game was.

Specifically, I would have taken the sum of the point differential at halftime, the point differential at the end of the third quarter, and the final MOV with the following limits:

• Any OT game is scored as a 1 point win.
• The differential is capped at 21 at each point so the maximum is 63.
• If the winning team has a negative total, the totals are adjusted to +1 for the winner and -1 for the loser.

Anything requiring any amount of nuance just wasn't an option then or now. 

Yet at the same time, they wound up constructing a Frankenstein monster of computers, polls, simple SOS, bonus wins, etc.

Ugh.

Anything requiring any amount of nuance just wasn't an option then or now.

I think the nuance would have been ok because most people don't understand the computer rankings anyway so changing from one formula they don't understand to another formula they don't understand wouldn't be a big deal.

I kinda like the square root idea posited above by @Cincydawg but I think that might be too nuanced. With a cap of 21 you could just say "It is capped at 21 to disincentivise piling points on hapless opponents." The square root thing more-or-less accomplishes the same thing but it is too complicated for a lot of people.

I don't know if the computers already do this but I think it is very important to look not just at final score but at what the committee has called "game control".

The examples I always use are two Ohio State games that I attended:
In the mid-90's the Buckeyes had a game against Iowa in which Ohio State just obliterated the Hawkeyes. At one point Ohio State led 56-0 and the game was never in doubt. However, Ohio State's backups, third stringers, waterboys, cheerleaders, and mascots got outscored 35-0 such that the final score was "only" a three-TD, 56-35 win.

About a decade later the Buckeyes beat Penn State in a tough and very competitive defensive slugfest. In the first ~55 minutes of the game the only difference was that Ohio State had done better in goal line situations such that the Buckeyes had a 14-6 lead on two scores for each team. With time running out Paterno was compelled to let his QB sling it around and Ohio State promptly scored 14 points on back-to-back pick-6's for an impressive looking final score of 28-6.

I use those two examples to illustrate the point that the final score can be misleading. In the tOSU/Iowa game the Buckeyes could have won by 100 if they had wanted to and the 21 point MOV severely understates Ohio State's level of dominance. In the tOSU/PSU game it was a hard fought game and the 22 point MOV severely overstates Ohio State's level of dominance.

tweaking the formula each season to try to fix the result from the previous season was silly

each season is different

That is what I think "we" learned when we did our own computer poll back when.  If it generated something odd, we "fixed" it, to the point it basically aligned with the human polls, more or less.  

This is a problem with climate modeling as well, they all have to align with past records (hoping they are accurate).  If they don't, they get adjusted.

Why do people refer to them as "computer polls?" 

Who are the computers polling, to determine their rankings?

Also why do people refer to the poll voters as "pollsters?"  Pollsters are the people who administer a poll, not the people who vote in a poll.  Those are just "voters."

So we have computer rankings, and voters.  Let's get this vocabulary straight, folks.

I suspect most 3 loss teams would have lost in at least one upset.  Take last season, say UGA loses to Mizzou, Tenn, and Georgia Tech, but win the SEC and is 10-3.  They'd make a 12 team playoff, they might even be a top 4 seed (depending).