Quote:To reply with the same type of snark you have engaged -- gosh, reading comprehension is hard - for you, those that refuse to understand the problems with how the league office made this determination, and the league office.
Those of us who know what words mean and know how math works haven’t had any difficulty parsing through the CUSA release. Cipherin’ is hard work!
Quote:Your link - did you read page 2? It is an almost verbatim copy of the tie-breaker provisions adopted by the league in 2005 and posted on the website well before the Nov. 24 release by the league. Here is why this is important:
* the 2005 tie-breaker provision identifies the highest BCS ranking (without the parenthetical explanation "i.e. average computer ranking added in the Nov. 24 press release) as the third tie-breaker;
BCS ranking is an integer number between 1 and 25 inclusive. Teams outside the top 25 do not have a ranking. They do, however, have a raw score. Nowhere does it say that BCS raw score is in the tiebreaker, though Marshall fans have repeatly tried to misapply it. I asked you for proof that the raw scores would be considered for teams without BCS rankings (i.e. Marshall and Rice) – yet you provide none.
Quote: * the "highest BCS ranking" per the BCS ranking formula is:
Coaches' Poll Component + Harris Poll Component + Average Computer Rating Component / 3;
Actually, it’s (Coaches' Poll Component + Harris Poll Component + Average Computer Rating Component) / 3;
That formula – the essential BCS formula – did not change throughout this entire endeavor. That formula exists so that small changes in poll rankings don’t overwhelm large differences in the average computer rating, as was the case here.
Quote:* the Nov. 24 release states that the BCS ranking is the same as the average computer ranking -- it is not; when there are human votes to consider, the average computer ranking is but 1/3 of the consideration; only when there are no human votes in either poll is the average computer ranking equal to the BCS ranking (since the human poll components would be zero and zero).
Not quite the case, but you are having problems with the simple math so I won’t go into details. BCS ranking is still an integer between 1 and 25 inclusive.
Quote:What the link that you provided proves is that the league office was inept and sloppy, and was probably presumptuous. If ECU wins, under either formula (the BCS formula or the average computer ranking in isolation), the conversation is moot because ECU would have the higher BCS formula value and the higher average computer ranking value (regardless of the scale used). The league (probably) expected ECU to win, and assumed even if ECU lost to MU that MU would not pick up votes in the human polls (which was a stupidly presumptuous attitude to have). Under that scenario - MU and Rice wins, and neither have human votes in either poll, the average computer ranking becomes the deciding factor.
If neither program has a BCS ranking and no poll votes, then the average computer ranking is the only thing to go with.
It was unlikely that any cusa school was going to end up ranked in the BCS, and the difference in computer rankings suggested between the 3 schools was so large, and not likely to change much since 11 games had already been played, that when the BCS formula (1/3 of each from two polls and computer ranking) was applied that any team would be able to make up the difference.
I’ll agree that they were too brief in their explanation of all the possibilities and could have saved a considerable amount of confusion.
Quote:Except that didn't happen. Even excluding the now reverse-conspiracy theory that Doc alone voted for Marshall in the Coaches' Poll, and placing a zero value for that component, Marshall received 10 votes in the Harris Poll. That alone would have been enough to determine the highest BCS ranking according to the actual BCS formula. Instead, because certain expectations were not realized (no votes for MU and Rice), CUSA cobbled together a formula that went against its approved tie-breaker method for reasons truly known only to the league office.
Can you show me in writing where CUSA or BCS says this is a way to calculate BCS ranking? No? Didn’t think so.
My guess is that Doc did indeed vote for his team to attempt to game the system. Bailiff obviously did not. It would not surprise me if CUSA caught wind of this potential conflict of interest.
At any rate, 13 and 10 are miniscule numbers in those polls, and could result from 2 wayward voters. It is not the intent of the system to allow the opinions of 2 individual voters (or 23, about 15%) to circumvent the will of the other 85% of voters and 83% of computers.
Since
neither team received a BCS ranking (again, an integer number between 1 and 25, just like all the rankings), the conference was forced to decide an alternate method.
One method they could have chosen, which would be less labor intensive, but inconsistent with the BCS ranking methodology and subject to wider corruption and general incompetence, is the raw score method that Marshall fans would have liked. Unfortunately, you get a score of (x/62 coaches consider + x/105 Harris voters considered + 0/0 polls considered). 0/0 is not a defined number and l’Hopital’s rule doesn’t help us here. This is not a suitable method for determining who is ahead per the BCS formula (again the 1/3 of each of the available datasets)
Another method would have been to use simply the computer results. Whereas this would provide the same end result, it unfairly neglects Marshall’s (relatively miniscule) contributions from the polls.
Thus, they had to use a formula which took all the features into account. Looking at the BCS raw score data, a Harris poll vote is worth 1/(105*25)= 3.81e-4 and a Coach’s poll vote is worth 1/(62*25) = 6.45e-4 points. Place in the average computer rankings, on the other hand, is worth 0.04 points. The conference could have certainly extended the computer rankings into negative numbers – those things do exist for non-junior Math majors at Marshall. Instead, they chose to scale the ranking (making each position in the poll worth only 0.008 points) which undervalued Rice’s relative strength in the polls compared to the traditional means.
Let’s look at a thought experiment for two cases:
Case 1: Team A is the 26th best team in the country (10-2, 6-2 in conference). Every computer poll and every pollster thinks they are #26. Team B is less so (6-6, 6-2 in conference). Every computer poll has them ranked 100, because they have some losses to some really bad teams. One voter decided to have pity on them (or as a joke, or because he is an alone) ranks them #25.
Using the raw score method, that .000381 points from the raw score would trump everything, and team B, although far less deserving would host.
Case 2: Teams A and B are really close in the computer polls, separated by 0.5 standing place (B ahead of A) when averaged out, but neither value being in the top 25. Team A has, however, been far more impressive in their wins, and gotten dozens of poll votes, but not enough to crack the top 25. The purely computer poll method would neglect this, and a less deserving team B would host.
The CUSA method of applying the BCS rankings methodology would get the right answer in both cases. Unfortunately, in this case, applying an appropriate methodology disadvantaged Hillbilly U while rewarding Egghead U, hence the outrage.
In short, this entire thread can be summed up by the following three statements:
(1) Marshall fans are upset because they lack reading comprehension skills
(2) Marshall fans are upset because they lack basic math skills
(3) Marshall fans will believe anything to be true, as long as it is what they want to hear