Skynet Championship Series: Have the Computers Been Compromised?

Let's first recap what happened in the actual BCS standings this past week. Because of the incompetence of the Big 12 in NOT providing enough tiebreakers, the BCS Standings are determining who goes to the Big 12 Championship game. Of course, drawing lines between Texas and Oklahoma is probably arbitrary and the difference is slight.

The human voters were a virtual deadlock with the Coaches giving the edge to Oklahoma 1397-1396, and the Harris poll voters giving the edge to Texas 2575 to 2569. The SIX computer polls still count for a third of the votes, and the computers gave the edge to Oklahoma.

Now, for those long-time readers who have been following all along, you know that I only use four of the SIX computer polls to compute the "Skynet Championship Series"... because the other two (Billingsley & Massey) start with the bias of rankings before the season begins.

Normally, this would not be a big deal. Until I made the most startling discovery when computing the "Skynet" rankings for this week - the Skynet rankings choose TEXAS (as you can see by scrolling to the bottom), NOT Oklahoma.

This then begs the question: do the other two polls favor Oklahoma because of the original rankings?
Answer: We don't know exactly. There's no way for mere mortals like myself to determine if the these two rankings chose Oklahoma over Texas because of the "human pre-rank" before the season... or if it's because of some other more appropriate reason in the formula (i.e. strength of schedule, quality victories, quality losses, etc.) These "computer" polls have varying length descriptions. Let's take a look at both of the polls and see what we can gather:

Billingsley
What I can present is this: Billingsley states that starting position only makes a "slight difference in my rankings." He also states that "Each year the #1 team starts at 270 points; #2, 269 points; #3, 268 points and so forth all the way to #120 starting at 151 points." In Week 1, Oklahoma was ranked #10, Texas ranked #11, meaning that Billingsley only has given Oklahoma a 1-point advantage. Texas is currently trailing by 9 points in the Billingsley poll, so I believe that Billingsley has not been "compromised."
Conclusion: Billingsley NOT compromised

Massey
Massey states that "Preseason ratings are typically derived as a weighted average of previous years' final ratings. As the current season progresses, their effect gets damped out completely. The only purpose preseason ratings serve is to provide a reasonable starting point for the computer. Mathematically, they guarantee a unique solution to the equations early in the season when not enough data is available yet." Unlike Billingsley, this is not enough of a description to accurately determine if they have Oklahoma two spots ahead of Texas for valid reasons. Currently, Massey has Oklahoma at 2.593, Texas Tech at 2.508, and Texas at 2.505.
Conclusion: Unclear, though probably NOT compromised, assuming Massey's statement is true that the effect "gets damped out" as the season progresses.

Now that that mystery has been addressed, a second question is appropriate:

How do my "Skynet" rankings differ from the BCS computer rankings???
Answer: In simple terms, the BCS formula only merely averages the standing of each team. For example, in the Sagarin ranking this week, Oklahoma is ahead of Texas by a very thin margin, but this only gets valued as a "1" in the BCS formula, versus a "2" for Texas. Because I'm normalizing each of the 4 rankings and then averaging them, I'm actually accounting for the actual difference between each team in each ranking, just like the Harris poll is counted by the exact number of votes that each team gets (as opposed to just being awarded all the points for being ranked #2).
Proposed BCS Adjustment: The BCS should not just award teams whole numbers for their standings in the BCS computers. It does not take into account the actual difference between teams and is an additional error on the system. Of course, one could argue that the differences in computer rankings is arbitrary, but it seems as if that's a losing argument when teams are closely bunched together. It seems more appropriate to me to count the actual worth each team is earning in each computer ranking.

This leads to a third question:
Would Texas leapfrog Oklahoma if the "Proposed BCS Adjustment" were adopted?
Answer: No. The BCS formula involves throwing out the highest and lowest computer ranking. I didn't feel like doing this for every team, but here's the top 5 in that scenario (using all 6 polls, so that only 4 polls are left and awarding specific points to each team based on specific ranking):
1. Oklahoma-1.103
2.Texas-1.315
3.Alabama-2.068
4.Texas Tech-2.281
5.Utah-2.625

And finally...

Skynet Championship Series
2008 -week of 12/1
..............................SCS....BCS...Sched Rnk
1.Texas...............1.19....(3)...(4)
2.Oklahoma.......1.33....(2)...(13)
3.Alabama.........2.07....(1)...(82)
4.Utah................2.53....(6)...(86)
5.Tex Tech........2.67....(7)...(56)
6.Florida............5.36....(4)...(31)
7.Boise St...........6.96....(9)...(114)
8.Penn St...........8.01....(8)...(72)
9.USC................8.22....(5)...(52)
10.Ohio St.......11.21....(10)...(41)
11.TCU............12.65....(11)...(81)
12.Ball St.........13.45....(12)...(126)
13.Cincy..........14.49....(13)...(66)
14.Ok St..........15.02....(14)...(47)
15.Georgia......15.12....(16)...(16)
16.Geo Tech...15.29....(15)...(37)
17.Bos Coll......16.09....(17)...(38)
18.BYU...........16.65....(18)...(101)
19.Missouri....17.65....(20)...(54)
20.Oregon......18.68....(19)...(69)

and as a fun bonus... I computed Skynet rankings if ALL 6 polls were used!
USING ALL 6 POLLS
..............................SCS....BCS...Sched Rnk
1.Oklahoma.......1.18....(2)...(13)
2.Texas..............1.47....(3)...(4)
3.Alabama.........2.00....(1)...(82)
4.Tex Tech.......2.52....(7)...(56)
5.Utah...............2.72....(6)...(86)
6.Florida............5.03....(4)...(31)
7.Boise St..........7.52....(9)...(114)
8.USC................7.53....(5)...(52)
9.Penn St..........8.88....(8)...(72)

DATA SOURCES
The 4 polls I use for "Skynet" are:
Sagarin -
1-Oklahoma-93.94
2-Texas-93.50
3-Texas Tech-92.90
4-Alabama-90.68

Anderson
1-Alabama-.811
2-Texas-.800
3-Oklahoma-.798

Colley's Matrix
1-Texas-.940002
2-Oklahoma-.921550
3-Alabama-.912339

Wolfe
1-Oklahoma-10.215
2-Texas-10.186
3-Texas Tech-10.008
4-Alabama-9.904

The polls I don't use are:

Billingsley
1. Alabama-325.926
2.Oklahoma-322.918
3.Florida-314.792
4.Texas-313.293

Massey
1.Oklahoma-2.593
2.Texas Tech-2.508
3.Texas-2.505
4.Alabama-2.424