Why the BCS ERA Sucked

The more I read about how the BCS was manipulating the computer rankings, the more convinced I am that we need to remove humans from the equation entirely.

Yes, we saw some good football during the BCS era, but off the top of my head I can think of 2 years when the obviously best team in the country was not in the championship game:
1998 Ohio State wrecked folks, and 2003 USC was clearly the best in the country.

The real problem, though, was a lack of understanding of computers and statistics. Moneyball hadn't become a thing yet, Bill James was a kook, and people just knew who was best (hint: my team was best). The internet was brand new and most people didn't have personal computers. There was zero appreciation for what even simple data analytics could do.

We also had (still have) the good ol' boys club that ran the bowl system and who were very protective of their monopoly. The first commish of the BCS was a former SEC commish instead of a neutral party.

1st Issue:
"Computer average: An average of the rankings of a team in three different computer polls were gathered (Jeff Sagarin/USA Today, Anderson & Hester/Seattle Times, and The New York Times), with a 50% adjusted maximum deviation factor. (For instance, if the computers had ranked a team third, fifth, and twelfth, the poll which ranked the team twelfth would be adjusted to rank the team sixth.)" ~Wiki

Sagarin has been held up as a gold standard for computerized rankings for a long time. Maybe they're awesome, maybe not. As I've stated in earlier posts, it is my opinion that the measure of how good a computer system is has been tied to how closely it resembles human voters for far too long. Mr. Sagarin has a BS in mathematics from MIT, so I have no business questioning his ability. My degrees are in science, not math. That said, I have yet to find anything that explains the rationale behind his ranking system. What is the goal? I expect that he explained it to the BCS folks, but his entire system is known only to him. And what did the BCS want? A system that resembles human voters would be my guess.

Anderson and Hester also do not appear to divulge their methods or reasoning. They do make a big deal about how margin of victory doesn't matter when scores are run up, but they don't say if those margins are simply capped or ignored entirely or controlled some other way.

I'm pretty sure only God and the editor in chief know what did or did not go into the NYT rankings.

Why was there a limit on the ranking differences? If the idea is to make sure there's not too much variation, why have multiple inputs? Or pick different computer rankings that more closely match each other? Why only 3? The answer to all of these questions is this: it was nothing more than lip service to the idea of a fair ranking system. They limited the differences so that one computer system didn't drop a popular team too far. They used 3 because someone said "Hey, 3 is a good number." 

Also, I've seen more than a few systems tout their retro accuracy: if you run the system on a year before computers, it spits out a top 20 that looks like the AP poll from that year. This should never be the goal of a computer/statistical system in cases like this, where human judgement is impaired at best and deliberately manipulated at worst. No, the goal should be to create a 100% impartial, objective, and transparent system that cannot be manipulated.

2nd issue:
"Poll average: Both the AP and ESPN-USA Today coaches polls were averaged to make a number which is the poll average."

Human voters. I want you to very carefully parse what I am about to say. Look around at our election process and understand that what we see wrong with political elections is what was wrong during the BCS era. Lobbying for votes; voting for self-interest instead of what is actually best, voting against someone you dislike vs having a candidate you can get behind. Do you think that maybe a coach whose mid-tier program stands to gain a substantial amount of money if the conference leaders make a BCS bowl is a "motivated voter?" How could it be otherwise? Or what about the coach who doesn't want to face a better team, why not vote that other team lower than they deserve?

As you've seen from my previous posts, human voters express bias one way or another. As bad as that is for a playoff, it is an order of magnitude worse for BCS championships because, by the neutering of the computer models, human voters had an outsized impact. This is exactly the opposite of a fair system.

3rd issue - Quality Win Component:

• In 2001, a quality win component was added. If a team beat a team which was in the top 15 in the BCS standings, a range of 1.5 to .1 points was subtracted from their total. 
• Beating the No. 1 ranked team resulted in a subtraction of 1.5-point, beating the No. 2 team resulted in a deduction of 1.4 points, and so on. Beating the No. 15 ranked team would have resulted in a deduction of .1 points. 
• A team would only be awarded for a quality win once if it beat a Top 10 team more than once (such as in the regular season and a conference championship game), and quality wins were determined using a team's current subtotal, not the ranking when the game was played. The subtotal ranks were used to determine quality win deductions to create a team's final score.
• The BCS continued to purge ranking systems which included margin of victory, causing the removal of the Matthews and Rothman ratings before the 2002 season.
• Sagarin provided a BCS-specific formula that did not include margin of victory, and The New York Times index returned in a form without margin of victory considerations. 
• In addition, a new computer ranking, the Wesley Colley Matrix, was added.
• The lowest ranking was dropped and the remaining six averaged. 
• In 2002, the quality win component was modified such that the deduction for beating the No. 1 team in the BCS would be 1.0, declining by 0.1 increments until beating the 10th ranked team at 0.1. 
• Teams on probation were not included in the BCS standings, but quality win points were given to teams who beat teams on probation as if they were ranked accordingly in the BCS.

Remember, lowest score for the best team, that's why they deduct points for doing better.

So, what do we have here?

1. A "quality" win only counts if you beat a team in the top 15, and later top 10. The spread doesn't matter: #1 beating #3 deducts 1.3 points from #1 and #25 beating #1 deducts 1.5 points from #25. Seems to me that the #25 over #1 win is of higher quality than  the other. There should be a modification for the ranking spread. Hell, even my system doesn't have that component yet, but it will for 2020/2021, and it will account for the gap in rankings at the end of the season. Why? That is a true measure of where a team stands. It's easy to look good before running the gauntlet. 
2. A quality win only counts once. Now, I don't know how many times a team has had to beat the same opponent twice in a season. I know it happens in the SEC, and this past season OSU had to beat Wisky twice. You're going to sit there and tell me that the Buckeyes' conference championship comeback wasn't a quality win? Because reasons? 
3. Margin of victory doesn't matter. #5 winning by 3 at home over #6 is the same victory as #5 going on the road and pounding #6? I don't think so, Tim.
4. You're on probation. You don't get to play for all the marbles, you don't even get to show up in our rankings, but if someone beats you we'll pretend that you matter only for the points.

Starting in 2004, the human component was given even more weight. Both the Coaches' popularity contest and the AP (and then Harris) polls were given as much weight as the average of 6 computer ranking systems. 2:1 ratio in favor of humans there, in case I wasn't clear. Why? Because people complained that there was no way a computer could figure it out. Again: I know what I know and what I know is my team is better!

Then, one magical day in 2009 the commish for the Mountain West conference proposed an eight team playoff, with the eight teams chosen via a selection process similar to the NCAA basketball tourney, and a ranking system that more heavily weighted inter-conference play. The rest of the commissioners saw the wisdom of allowing more teams more opportunities and adopted the proposal on the spot, and we all lived happily ever after!

Ok, what really happened is the Big 5 commissioners claimed that there was no support for such a thing among the BCS member conferences. In other words, "You don't belong here, little man, even though your best team just whupped Oklahoma a couple of years ago on New Year's Day in front of millions of viewers. We're gonna wait a few years an then introduce this idea as ours and make sure there's no way in hell Boise will ever get another shot at #1."

I'll give you a sneak peek at my next Ray's Retro Rankings, because it is 100% on point with how awful the BCS was. In 2011, the year of the LSU-Bama snoozefest, the SEC was ranked 11 spots higher on average than they should have been by the BCS. Eleven. Spots. The 2011 National Championship Game should have been played between LSU and Stanford. Bama wasn't even top 4.

College football is more of a business now than ever. It is entertainment more than sport. How can I say that? Look at the playoff: 4 teams, subjective selections. What does that do for the game other than generate controversy - deliberately. A fully transparent process that removed human bias would not generate nearly as much controversy, so fewer people would be interested in TV talking heads praising/excoriating it, and so fewer ad dollars would be had. This is the true legacy of the BCS, and the core reason that the BCS sucked. Every single controversy raised awareness, and just like a junkie who needs bigger and bigger hits to get high, college football needs more and greater controversy to grow that ad revenue.