Computer polls are as bad as opinion polls

[aufan59]

Computer polls use a formula that weigh certain components of team based on how the formula's creator thinks they should be weighed. The creator's decision on how to weigh each statistic is based off of his opinion.

For example a human pollster might evaluate the SOS of a team by how many top ten teams they have played. A computer poll might be designed to weigh top ten teams very heavily in its SOS component. Both polls end up ranking teams based on the opinion that playing top ten opponents are what make your schedule tough.

It was someone's opinion that the BCS should include the AP poll. The BCS itself is a computer poll, but opinion changed and thus the BCS doesn't include the AP poll. Despite it being an unbiased computer poll, results of the BCS are based upon opinion.

It happens to be my opinion that playing a rank #100 team is not more difficult for the best team than playing a rank #150 team.

Jeff Sagarins' opinion is different. He feels that playing a rank #125 team is more difficult for the best team than playing a rank #150 team.

It is his opinion that drives this 'opinionless' poll.

I do understand that computer polls are consistent in their rankings, however they cannot interpret data that can't be fed to them. Weather, garbage TDs, injuries, match ups, etc. The human interpretation factor is more important than complete consistancy without any understanding.

So it boils down to this: Do you want a poll that is based the opinion of people who play, coach and/or study the game, or a poll that is based on the opinion of someone who has probably not ever played or regularly watches organized football?

[Blue Hen]

I'm not a big fan of the pollsters or the computer geeks. Lots of poll voters, also, never played a down of FB in their lives ( especially AP & Harris ) and this so called 'Coaches' poll is pretty much a joke. These head coaches are always telling us that they work 60+ hours a week with very little sleep, just to prepare their teams for the next opponent.........and they have time to keep up with 118 other teams and rank them ???????. You get ADs, secretaries, office clerks and assorted other designated individuals doing much of the ranking/voting in that fraudulent poll. And any poll that mandates that its voters automatically select the winner of a certain post season game a final #1 rather than allowing its voters to vote their convictions...well, that's not much on credibility, in my book.

[aufan59]

I'm not a big fan of the pollsters or the computer geeks. Lots of poll voters, also, never played a down of FB in their lives ( especially AP & Harris ) and this so called 'Coaches' poll is pretty much a joke. These head coaches are always telling us that they work 60+ hours a week with very little sleep, just to prepare their teams for the next opponent.........and they have time to keep up with 118 other teams and rank them ???????. You get ADs, secretaries, office clerks and assorted other designated individuals doing much of the ranking/voting in that fraudulent poll. And any poll that mandates that its voters automatically select the winner of a certain post season game a final #1 rather than allowing its voters to vote their convictions...well, that's not much on credibility, in my book.

Forcing a #1 vote makes sense.

Their votes are what created the #1 vs #2 match up. The #1 vs #2 match up has the top 2 teams in it according to their votes.

Therefore the winner is #1 according to their own votes. Voting otherwise would defeat the purpose of voting to see who was #1 and #2 then having them play to see who was the best.

[Blue Hen]

It might make sense if all the Coaches poll voters ( whoever they may actually be) had the same teams ranked #1 & 2 going into the PSEG season, but that's never the case. For example, if some of these poll voters had team X ranked their personal #1 going into the post season ( yet it was # 3 or 4 or anywhere else in the final poll ) and that team performed in a dominant fashion in its PSEG, why shouldn't those voters be allowed to continue voting their true convictions and stick with their original #1 ?

[CJHawkeyes]

Computer polls use a formula that weigh certain components of team based on how the formula's creator thinks they should be weighed. The creator's decision on how to weigh each statistic is based off of his opinion.

For example a human pollster might evaluate the SOS of a team by how many top ten teams they have played. A computer poll might be designed to weigh top ten teams very heavily in its SOS component. Both polls end up ranking teams based on the opinion that playing top ten opponents are what make your schedule tough.

It was someone's opinion that the BCS should include the AP poll. The BCS itself is a computer poll, but opinion changed and thus the BCS doesn't include the AP poll. Despite it being an unbiased computer poll, results of the BCS are based upon opinion.

It happens to be my opinion that playing a rank #100 team is not more difficult for the best team than playing a rank #150 team.

Jeff Sagarins' opinion is different. He feels that playing a rank #125 team is more difficult for the best team than playing a rank #150 team.

It is his opinion that drives this 'opinionless' poll.

I do understand that computer polls are consistent in their rankings, however they cannot interpret data that can't be fed to them. Weather, garbage TDs, injuries, match ups, etc. The human interpretation factor is more important than complete consistancy without any understanding.

So it boils down to this: Do you want a poll that is based the opinion of people who play, coach and/or study the game, or a poll that is based on the opinion of someone who has probably not ever played or regularly watches organized football?


Where to begin? By your strained logic, the winner of a football game and the winner of a figure skating competiton are both the result of opinion because someone(s) decided that a touchdown is worth six points as opposed to seven or eight points. Furthermore, no objective system actually considers difficulty. Certainly, objective rules can be intuitively designed to favor greater difficulty, but none ever specifically reward it. After all, nowhere in sports is the value of a win specifically tied to difficulty. Therefore, your opinion that playing the #100 team is no more difficult than playing the #150 is irrelevant because no objective system makes any claim about difficulty. Whatever wins or resumes have more value in an objective system is based on how each system defines "success" not difficulty.

As for the data that "computers" cannot interpret. Conference standings do not interpret those things either. Does that bother you? What sports actually consider such nonsense? Those factors are not needed in order to determine the winner(s) of a competition which is the purpose served by objective rules. Human interpretation is not needed to determine the winner of a competition. Given that humans can disagree on the "winner" of a competition, what problem do humans solve that make their opinion superior to the "opinion" of computers?

The bottom line is do you want a winner based on fixed objective criteria similar to how all sensible team competitions are conducted or do you want a "whatever it is polls determine" determined by arbitrary, unaccountable and irrelevant opinions influenced by matters completely unrelated to the concept of determining a winner? The difference between "computers" and polls is that the former are actually rules that you can adopt a specific strategy for whereas with polls there is no known difference between any two results. Therefore, it is not unlike playing a game always ignorant of what specifically determines whether you win or lose.

[aufan59]

Where to begin? By your strained logic, the winner of a football game and the winner of a figure skating competiton are both the result of opinion because someone(s) decided that a touchdown is worth six points as opposed to seven or eight points. Furthermore, no objective system actually considers difficulty. Certainly, objective rules can be intuitively designed to favor greater difficulty, but none ever specifically reward it. After all, nowhere in sports is the value of a win specifically tied to difficulty. Therefore, your opinion that playing the #100 team is no more difficult than playing the #150 is irrelevant because no objective system makes any claim about difficulty. Whatever wins or resumes have more value in an objective system is based on how each system defines "success" not difficulty.
Difficulty is the word we attach to things like strength of schedule. Sagarin uses the Elo chess rating system. This determines your chance of beating a team and awards you points based upon that chance. This is a simple explanation. A chance of something happening can be defined as difficulty.

The chance of one team beating another is how the formulas define difficulty. To say they don't factor in difficulty is misinformed.

I can rephrase myself not using difficulty but retain the same meaning:

In my opinion, #1 beating #100 should not be more rewarding than #2 beating #150.

Sagarin's opinion is different.

That was a lame 'technicality' in your argument against me.


As for the data that "computers" cannot interpret. Conference standings do not interpret those things either. Does that bother you? What sports actually consider such nonsense? Those factors are not needed in order to determine the winner(s) of a competition which is the purpose served by objective rules. Human interpretation is not needed to determine the winner of a competition. Given that humans can disagree on the "winner" of a competition, what problem do humans solve that make their opinion superior to the "opinion" of computers?

Conference standings are fine because everyone plays everyone they need to to decide a winner(except big10). The whole reason we have to judge teams is because they don't play each other.

Bringing up conference standings is a moot point. Polls are there because emulating conference standings for a national standings is impossible.


The bottom line is do you want a winner based on fixed objective criteria similar to how all sensible team competitions are conducted or do you want a "whatever it is polls determine" determined by arbitrary, unaccountable and irrelevant opinions influenced by matters completely unrelated to the concept of determining a winner? The difference between "computers" and polls is that the former are actually rules that you can adopt a specific strategy for whereas with polls there is no known difference between any two results. Therefore, it is not unlike playing a game always ignorant of what specifically determines whether you win or lose.

The computer polls are just as arbitrary as the human ones. They are based on what someone thinks makes a good team. Just because the rules are preset doesn't make them better. A computer poll could rank teams by the number of letters of all their starting offensive linemen. I'd rather have adaptive intuitive poll than one that is arbitrarily based on some team stats.

[CJHawkeyes]

BTW, the BCS formula is not a computer poll. The polls are the dominate factors in it nowadays. The computers used merely serve as tiebreakers if the vote for the top two is close.

[aufan59]

BTW, the BCS formula is not a computer poll. The polls are the dominate factors in it nowadays. The computers used merely serve as tiebreakers if the vote for the top two is close.

Yes it is a computer poll.

A computer poll is simply a poll that uses a formula. The 1/3 Harris, 1/3 Coaches, 1/18th each computer poll is a formula. Very simple formula, but is a computer poll.

[CJHawkeyes]

Difficulty is the word we attach to things like strength of schedule. Sagarin uses the Elo chess rating system. This determines your chance of beating a team and awards you points based upon that chance. This is a simple explanation. A chance of something happening can be defined as difficulty.

The chance of one team beating another is how the formulas define difficulty. To say they don't factor in difficulty is misinformed.

I can rephrase myself not using difficulty but retain the same meaning:

In my opinion, #1 beating #100 should not be more rewarding than #2 beating #150.

Sagarin's opinion is different.

That was a lame 'technicality' in your argument against me.



Conference standings are fine because everyone plays everyone they need to to decide a winner(except big10). The whole reason we have to judge teams is because they don't play each other.

Bringing up conference standings is a moot point. Polls are there because emulating conference standings for a national standings is impossible.



The computer polls are just as arbitrary as the human ones. They are based on what someone thinks makes a good team. Just because the rules are preset doesn't make them better. A computer poll could rank teams by the number of letters of all their starting offensive linemen. I'd rather have adaptive intuitive poll than one that is arbitrarily based on some team stats.

As someone who has devised an objective system and argues for it, schedule strength indicates how successful a team's opponents are according to the rules within the system. My system makes no claim about difficulty. Difficulty cannot be quantified. The value of wins and losses under my system is not based on their perceived difficulty.

That said, the purpose of my system is simply to determine the winner or winners of a competition. The most successful team according to its rules is nothing more and nothing less than the winner of first place. The results make no other claim.

Furthermore, the difference in schedules does not prevent objective rules from determining a winner. As for the arbitrariness of computers, I don't even think you know the definition of the word. I certainly wouldn't defend any rules that ranked teams based on the ridiculous example you cited. Certainly, there are bad and nonsensical objective rules, but rules such as mine are not arbitrary. Under my rules, the exact same results will produce the exact same standings and there will be an internally consistent reason for each team being ranked ahead of another. Every team can specifically know the difference between any two positions in the standings. Nothing could further from the truth where polls are concerned. Under polls, A > B and B > A can both be true given the exact same results and no pollster can show that his final ballot would have been the same before a single game was played. Furthermore, no pollster can show that his reason for ranking A higher than B, B higher than C , and C higher than D is internally consistent. That is arbitrary.

An example of arbitrainess where polls are concerned. My system and the AP poll has disagreed on the highest ranked one-loss team 13 times since 1978. In each of these instances, my system favored the team with the stronger schedule just as it did the 16 times it agreed with the AP's top one-loss team. However, the AP always favored the team that lost first even though it did not always do so during the years it agreed with my system.

Under my system, teams are not specifically ranked best to worst record. However, the nature of my system is such that under ideal circumstances, if two teams share the same record, the team with the better opponents' record will place higher. Since 1978, four teams have finished first under my system but did not win at least a share of the mythical title. One of these teams matched the MNC's record while playing opponents that were 20 games better. As far as I'm concerned, I'd rather compete under rules whereby I know before a single game is played that the same record versus opponents 20 games better guarantees a higher finish rather then leave it up to an adaptive intuitive poll whose results do not show how or why I won or lost especially if I were cheering for Auburn in 1983.

[CJHawkeyes]

Yes it is a computer poll.

A computer poll is simply a poll that uses a formula. The 1/3 Harris, 1/3 Coaches, 1/18th each computer poll is a formula. Very simple formula, but is a computer poll.

Computer polls (which is a misnomer anyway) are what the various OBJECTIVE systems are called. The BCS formula is predominately two combined polls with "computers" serving as tiebreakers.

[aufan59]

Difficulty cannot be measured, but it is a word we use when describing the relative strength of an opponent. "Stronger schedule" denotes more difficulty. In our case difficulty is defined as the relative strength or weakness of a team or group of teams.

The goal in sports is to see who wins. Your system, much like the one we have, is to predict who would win without the teams playing. This is so we don't have to bother with having anyone but #1 play #2 because we predicted that these two teams could beat everyone else.

Predicting who wins is out side the rules of the game. If two teams didn't play, the decision on who would win is decided through outside means. How to decide who wins a game that isn't played can only be done by opinion. Just because it would be decided the same way every time doesn't mean its not an opinion. Your formula uses things that you think are the best ways to determine the winner of an unplayed game.

If its not decided on the field, then there is no objective way to decide who would win without an opinion somewhere in the mix. Just because a system is consistent in deciding the victor doesn't mean it didn't stem from an opinion somehow.

[CJHawkeyes]

Difficulty cannot be measured, but it is a word we use when describing the relative strength of an opponent. "Stronger schedule" denotes more difficulty. In our case difficulty is defined as the relative strength or weakness of a team or group of teams.

The goal in sports is to see who wins. Your system, much like the one we have, is to predict who would win without the teams playing. This is so we don't have to bother with having anyone but #1 play #2 because we predicted that these two teams could beat everyone else.

Predicting who wins is out side the rules of the game. If two teams didn't play, the decision on who would win is decided through outside means. How to decide who wins a game that isn't played can only be done by opinion. Just because it would be decided the same way every time doesn't mean its not an opinion. Your formula uses things that you think are the best ways to determine the winner of an unplayed game.

If its not decided on the field, then there is no objective way to decide who would win without an opinion somewhere in the mix. Just because a system is consistent in deciding the victor doesn't mean it didn't stem from an opinion somehow.

My system is not predictive and it makes no claim about who would win a future football game. Under my system, Florida placed higher than Michigan at the end of the regular season because their record earned the Gators more points than the Wolverines. The Gators were nothing more and nothing less than more successful according to the rules I designed and favor to determine the winners of college football's regular season competition. Whether or not the Gators would beat Michigan if the two teams actually played is irrelevant. Michigan would have placed higher if they had beaten Ohio State or their defeated opponents won four more games than they did. With polls, there is no specific reason for either team being ranked ahead of the other. Furthermore, we can know that bias played a major role in the polls. All SEC coaches with votes favored Florida. Only two Big Ten coaches did not favor Michigan. One of those was biased in favor Florida due to being the former coach of most of its players. The other opted out of voting because he did not want to vote against a conference rival nor vote in favor of giving that rival a second chance against his own team. Objective rules are not biased in favor of particular teams nor can they avoid "picking a side" due to a conflict of interest because no such conflict exists. Objective rules have no dog in the fight. They simply favor the most successful teams as they define success. Success isn't defined by polls and is ultimately the result of multiple mental coin flips all influenced by matters unrelated to the purpose of determining a winner. Under my system, the difference between 2nd and 3rd is determined by wins and losses and, in rare instances, points for and points against. With polls, Florida might have won an opportunity to play for the national title just because a former coach favored them no matter what. To consider those two scenarios just as bad as one another is ridiculous. Polls are inherently worthless, tell us nothing about nothing, and do not solve any problem given that results produced by them could not be promised to us before a single game was played. I could explain my rules to you, post the resumes of two teams and you could know 100% of the time which team would place higher. However, I guarantee if I posted the resumes of two teams and asked you to tell me which finished higher in the polls, you would certainly not choose correctly 100% of the time aside from being the luckiest guesser or remembering who had finished higher. You certainly couldn't point to any reason for one team finishing higher.

[ktffan]

Computer polls (which is a misnomer anyway) are what the various OBJECTIVE systems are called. The BCS formula is predominately two combined polls with "computers" serving as tiebreakers.

"Computer polls" are not objective. Don't give me that crap.

[CJHawkeyes]

"Computer polls" are not objective. Don't give me that crap.


For starters, computer polls are neither polls nor have any relation to computers outside the use of spreadsheets to guarantee accurate computation. That said, computer polls is what the various objective systems are unfortunately referred to. How are those systems not objective?

[ktffan]

For starters, computer polls are neither polls nor have any relation to computers outside the use of spreadsheets to guarantee accurate computation. That said, computer polls is what the various objective systems are unfortunately referred to. How are those systems not objective?

I've done enough computer rankings to know that you can get what you want out of them, even prior to the games playing. They include the inherent biases of the programmer and I can raise up any team I want before the games are palyed, based on the history of said team. The whole 'objective' term comes from people who run said systems in an attempt to think that their system is truely the one or whatever. It's just hot air.

[CJHawkeyes]

I've done enough computer rankings to know that you can get what you want out of them, even prior to the games playing. They include the inherent biases of the programmer and I can raise up any team I want before the games are palyed, based on the history of said team. The whole 'objective' term comes from people who run said systems in an attempt to think that their system is truely the one or whatever. It's just hot air.


Nothing could be further from the truth. Obviously, if someone's goal is simply to devise rules that inherently favor their team only to change them when their team is not favored, then the "objectiveness" of such a system is worthless. However, anyone doing so would be replacing one objective system with another in order to get a desired result. But under the rules I favor, every team would know before a single game is played what a win or loss will be worth versus a hypothetical opponent. The value doesn't change to suit the Hawkeyes because that happens to be my team. What is it you think I want to get out of my system given that it treats every team as Generic State? There is no prejudice or bias inherent my system. That is the definition of objective. Granted, my system has an inherent bias towards winners and scheduling winners, but that type of "bias" does not favor any particular team. I won't defend objective systems that do that and while I may have reasons for opposing those used by the BCS, I have no reason to believe any operates with an inherent bias for a particular team or teams.

[CJHawkeyes]

BTW, I do not mean to imply that objective rules are inherently fair because they are objective. Obviously, objective rules can be devised to favor particular teams for no reason other than someone's bias towards that team. However, if a group of us sat down to agree upon rules to determine the winner of a competition involving our teams, we wouldn't agree to rules that operated as such. Therefore, the argument that objective rules are just as bad as subjective rules is only true if we are comparing unfair objective rules to subjective rules.

[CJHawkeyes]

Another point I forgot to make is what would be my motivation for creating an objective system biased in favor of my team when that bias would be self evident to everyone familar with the rules thereby guaranteeing it would not be implemented when I could simply vote for my team without any accountability?

[aufan59]

My system is not predictive and it makes no claim about who would win a future football game. Under my system, Florida placed higher than Michigan at the end of the regular season because their record earned the Gators more points than the Wolverines. The Gators were nothing more and nothing less than more successful according to the rules I designed and favor to determine the winners of college football's regular season competition. Whether or not the Gators would beat Michigan if the two teams actually played is irrelevant. Michigan would have placed higher if they had beaten Ohio State or their defeated opponents won four more games than they did. Your poll predicted that Florida was better than Michigan. More deserving. Had better judged criteria. Whatever. You are missing the point. The entire point of polls decide who the most deserving two teams are. By placing team #2 over a team that is #3, you are saying that #2 is more deserving than #3. This is predicting that #2 would beat #3. If we can't assume that one poll's #2 will beat its #3, then what is the point of the poll? These polls only predict relative strength of teams. If one team is relatively stronger than another by virtue of a specific poll, then that poll is predicting that the stronger team would win a match up.

And yes, your poll predicts relative strength. You choose to say that it only measures accomplishments. Guess what measuring relative strength means? Its measuring performance and accomplishments of multiple teams to see how they compare to each other.

This is the third time you've tried to counter one of my points by fighting semantics instead of the actual point, which is what people do when they have no argument.



With polls, there is no specific reason for either team being ranked ahead of the other. Furthermore, we can know that bias played a major role in the polls. All SEC coaches with votes favored Florida. Only two Big Ten coaches did not favor Michigan. One of those was biased in favor Florida due to being the former coach of most of its players. The other opted out of voting because he did not want to vote against a conference rival nor vote in favor of giving that rival a second chance against his own team. Objective rules are not biased in favor of particular teams nor can they avoid "picking a side" due to a conflict of interest because no such conflict exists. Objective rules have no dog in the fight.

Objective rules can have a dog in a fight. It is just that the dog in the fight is consistently chosen for the same reason over and over. It's obvious Sagarin ratings favor a schedule with 'higher quality' rent-a-wins. The SEC's location gives them 'lower quality' rent-a-wins than other teams. They in turn get worse SOS and thus worse ratings due to this 'objective' system. The objective system doesn't recognize that the difference between two different rent-a-wins does not matter.


They simply favor the most successful teams as they define success. And the criterion for success is defined by the computer poll creator. This is where opinion comes in.


Success isn't defined by polls and is ultimately the result of multiple mental coin flips all influenced by matters unrelated to the purpose of determining a winner. Under my system, the difference between 2nd and 3rd is determined by wins and losses and, in rare instances, points for and points against. With polls, Florida might have won an opportunity to play for the national title just because a former coach favored them no matter what. To consider those two scenarios just as bad as one another is ridiculous. Polls are inherently worthless, tell us nothing about nothing, and do not solve any problem given that results produced by them could not be promised to us before a single game was played. I could explain my rules to you, post the resumes of two teams and you could know 100% of the time which team would place higher. However, I guarantee if I posted the resumes of two teams and asked you to tell me which finished higher in the polls, you would certainly not choose correctly 100% of the time aside from being the luckiest guesser or remembering who had finished higher. You certainly couldn't point to any reason for one team finishing higher.

There is no correct answer to "which team should finish higher".

Much like it is a human pollsters opinion to put certain teams in certain place by his own critereon, its your opinion to judge teams by only their record and opponent's record.

Just because a system is objective does not mean it is fair, accurate, or wasn't at one point driven by opinion. Computer poll supporters seem to think that an objective poll is synonymous with fair, accurate, and opinion free poll.

[ktffan]

Nothing could be further from the truth.


Bull.

[aufan59]

Another point I forgot to make is what would be my motivation for creating an objective system biased in favor of my team when that bias would be self evident to everyone familar with the rules thereby guaranteeing it would not be implemented when I could simply vote for my team without any accountability?

Things can be biased without the bias being intentional...

[HellYeahHokie]

Bull.

I believe kttfan fan is making the point that computer polls are based on an algorithm that has the developers bias. Margin of victory is one of those parameters that has some inherent bias.

I think what CJ is saying is that these polls are based on a fixed mathematical formula that is applied equally to every team throughout the season. And the formula doesn't change based on someone's opinion, like Florida jumping Michigan last year in the human polls.

I think both of you are smart enough to understand this without argument.

[CJHawkeyes]

Your poll predicted that Florida was better than Michigan. More deserving. Had better judged criteria. Whatever. You are missing the point. The entire point of polls decide who the most deserving two teams are. By placing team #2 over a team that is #3, you are saying that #2 is more deserving than #3. This is predicting that #2 would beat #3. If we can't assume that one poll's #2 will beat its #3, then what is the point of the poll? These polls only predict relative strength of teams. If one team is relatively stronger than another by virtue of a specific poll, then that poll is predicting that the stronger team would win a match up.

And yes, your poll predicts relative strength. You choose to say that it only measures accomplishments. Guess what measuring relative strength means? Its measuring performance and accomplishments of multiple teams to see how they compare to each other.

This is the third time you've tried to counter one of my points by fighting semantics instead of the actual point, which is what people do when they have no argument.




Objective rules can have a dog in a fight. It is just that the dog in the fight is consistently chosen for the same reason over and over. It's obvious Sagarin ratings favor a schedule with 'higher quality' rent-a-wins. The SEC's location gives them 'lower quality' rent-a-wins than other teams. They in turn get worse SOS and thus worse ratings due to this 'objective' system. The objective system doesn't recognize that the difference between two different rent-a-wins does not matter.

And the criterion for success is defined by the computer poll creator. This is where opinion comes in.



There is no correct answer to "which team should finish higher".

Much like it is a human pollsters opinion to put certain teams in certain place by his own critereon, its your opinion to judge teams by only their record and opponent's record.

Just because a system is objective does not mean it is fair, accurate, or wasn't at one point driven by opinion. Computer poll supporters seem to think that an objective poll is synonymous with fair, accurate, and opinion free poll.

My "poll" did not predict that Florida was better than Michigan any more than the rules to a football game predict the team that scores the most points is better. Florida was simply more successful than Michigan according to these particular rules. Of course different rules may have determined Michigan to be more successful. Different rules that produce different results are not bad for that reason. Certainly, whichever rules one favors is a matter of preference, but favoring particular rules is not the same as favoring particular teams. My system did not determine Florida to be the higher ranked because it is biased towards the Gators in the same way SEC coaches and Ron Zook were. The absurdity of this debate is that people think objective rules are subjective rules because a person had to decide upon the objective rules. Therefore, every single competition governed by objective rules produces a winner that is the result of opinion? That is ridiculous.

[ktffan]

I believe kttfan fan is making the point that computer polls are based on an algorithm that has the developers bias. Margin of victory is one of those parameters that has some inherent bias.


There's a middle ground to this argument, but when he starts his post, "Nothing can be further from the truth", when one of the BCS computers got it's start exactly that way, I figure that nothing intelligent is going to follow. Maybe I'm wrong, but give me a break.

[CJHawkeyes]

Things can be biased without the bias being intentional...

Hypothetically, all I-A teams are ranked best to worst record, opponents' record, and so on. Where is the bias within these rules that is comparable to the bias within polls? Bias in favor of specific objective criteria is not at all similar to bias in favor of specific teams.

[CJHawkeyes]

There's a middle ground to this argument, but when he starts his post, "Nothing can be further from the truth", when one of the BCS computers got it's start exactly that way, I figure that nothing intelligent is going to follow. Maybe I'm wrong, but give me a break.


I concede that line may have been a mistake but given that I qualified it by noting that any system purposely designed to favor specific teams is as worthless as polls makes it irrelevant anyway. If anyone is arguing on behalf of such a system, I would join you in opposing it. However, I get the impression you think all objective systems are like that. My Hawkeyes have faired worse under my system than the polls. Did I purposely design my system to cheat my favorite team?

[HellYeahHokie]

There's a middle ground to this argument, but when he starts his post, "Nothing can be further from the truth", when one of the BCS computers got it's start exactly that way, I figure that nothing intelligent is going to follow. Maybe I'm wrong, but give me a break.

I think you've been posting here long enough to know what CJ advocates, and has elucidated on innumerous occasions. To dismiss his entire point based on a casual comment is a bit petty, don't you think?

[CJHawkeyes]

I think you've been posting here long enough to know what CJ advocates, and has elucidated on innumerous occasions. To dismiss his entire point based on a casual comment is a bit petty, don't you think?

Innumerous sounds like me. :D

[ktffan]

I think you've been posting here long enough to know what CJ advocates, and has elucidated on innumerous occasions. To dismiss his entire point based on a casual comment is a bit petty, don't you think?

Of course I know what he advocates. He repeats it over and over and over and overs. This site tends to be a broken record because everybody wants somthing that will only happen in our wildest imagination and nobody is willing to drop their utopian dream and deal with reality. It gets tiring.

I've done computers polls. Even if you want to be objective, you can't. The trouble is that either you don't care enough about football to understand a sport and you come up with hack's rating like Sagarin, or you love the sport so much, what makes a good team is so ingrained (usually favoring your team), you can't be objective. I refuse to conceed that computer rankings are objective in any concept outside of utopia.

[CJHawkeyes]

Of course I know what he advocates. He repeats it over and over and over and overs. This site tends to be a broken record because everybody wants somthing that will only happen in our wildest imagination and nobody is willing to drop their utopian dream and deal with reality. It gets tiring.

I've done computers polls. Even if you want to be objective, you can't. The trouble is that either you don't care enough about football to understand a sport and you come up with hack's rating like Sagarin, or you love the sport so much, what makes a good team is so ingrained (usually favoring your team), you can't be objective. I refuse to conceed that computer rankings are objective in any concept outside of utopia.

I wouldn't have to repeat myself over and over if people would finally stop arguing against objective rules based on their misunderstanding of what objective rules are and do. Case in point, your statement that you can't be objective even if you want to. I'm waiting for you explain how the rules I devised aren't objective. Furthermore, I would like to know why a preference for determining the national champion by objective means only indicates that I don't care enough about football to understand the sport. Talk about hot air.

[CJHawkeyes]

Your poll predicted that Florida was better than Michigan. More deserving. Had better judged criteria. Whatever. You are missing the point. The entire point of polls decide who the most deserving two teams are. By placing team #2 over a team that is #3, you are saying that #2 is more deserving than #3. This is predicting that #2 would beat #3. If we can't assume that one poll's #2 will beat its #3, then what is the point of the poll? These polls only predict relative strength of teams. If one team is relatively stronger than another by virtue of a specific poll, then that poll is predicting that the stronger team would win a match up.

And yes, your poll predicts relative strength. You choose to say that it only measures accomplishments. Guess what measuring relative strength means? Its measuring performance and accomplishments of multiple teams to see how they compare to each other.

This is the third time you've tried to counter one of my points by fighting semantics instead of the actual point, which is what people do when they have no argument.




Objective rules can have a dog in a fight. It is just that the dog in the fight is consistently chosen for the same reason over and over. It's obvious Sagarin ratings favor a schedule with 'higher quality' rent-a-wins. The SEC's location gives them 'lower quality' rent-a-wins than other teams. They in turn get worse SOS and thus worse ratings due to this 'objective' system. The objective system doesn't recognize that the difference between two different rent-a-wins does not matter.

And the criterion for success is defined by the computer poll creator. This is where opinion comes in.



There is no correct answer to "which team should finish higher".

Much like it is a human pollsters opinion to put certain teams in certain place by his own critereon, its your opinion to judge teams by only their record and opponent's record.

Just because a system is objective does not mean it is fair, accurate, or wasn't at one point driven by opinion. Computer poll supporters seem to think that an objective poll is synonymous with fair, accurate, and opinion free poll.

1-There are two kinds of objectve systems: predictive and retrodictive. Mine is the latter. The standings produced by it are not an implied prediction of hypothetical matchups.

2-I don't know what the entire point of polls is. Furthermore, my system is not a poll. That said, its purpose is to determine the winner or winners of a competition. In this instance, the competition is college football's regular season and the winners (higher placed teams) would presumably be awarded playoff/postseason berths.

3-My system determined that Florida was more successful than Michigan based on the relevant rules, and therefore, more deserving of a higher ranking. This does not imply that Florida would beat Michigan. The purpose of competition is to determine a winner. It is not determine the identity of the best team. Therefore, placing #2 above #3 is not a prediction that the former will beat the latter. It only implies that #2 is more successful than #3 according to the rules. Do the standings in various pro leagues imply a predicted outcome for future games? Of course not. The same is true for my system.

4-My system does not predict relative strength. It states that X record versus Y schedule is worth Z points. A team with more points is not necessarily a better team. It is simply more accomplished based on the relevant rules.

5-Objective rules do not have a dog in the fight. A dog in the fight refers a team in the competition. Preference for different statistical criteria is not the same thing. Otherwise, the rules of football have a dog in the fight due to their assigning greater value to touchdowns than to field goals.

6-I'm not arguing semantics. I'm disputing your complete misunderstanding of what my system does and what purpose it is meant to serve.

7- An objective system "doesn't recognize that the difference between two different rent-a-wins does not matter" because the difference does matter to the system based on how it assigns value to wins. To use a basketball analogy, the rules of the game do not recognize that the degree of difficulty between making a shot from 19-8 is negligible compared to a shot from 19-9. Nevertheless, the latter shot is still worth one more point. As such, if teams know that one rent-a-win will be worth than another based on known results, than it is in each team's best interests to schedule the more valuable rent-a-win regardless of how much difference you assign to the two results.

8-Yes, opinion must decide what the objective rules will be. But that is not the same as opinion decidng what the results will be. If objective rules are just as bad as subjective rules because a person chooses objective rules "like" a person chooses who to rank higher in polls, then the rules of football are just as bad as the rules of figure skating because a person must choose to assign a value of six points to a touchdown. Apples and oranges.

9-When I said that I bet you could not correctly choose the higher ranked team in polls 100% of the time, I said nothing about picking the team that should be higher ranked. There is no one right answer to who should be ranked higher. Under my system, it is right to say that Florida should ranked higher based on the rules and known results. Under another system, Michigan may be the right answer to who should be ranked higher. My system has one right answer for all possible outcomes whereas polls have multiple potential "right" answers for the same outcome with no way to determine which is THE right answer to who should be ranked higher.

10-It is true that a system can be objective and unfair if its rules do not apply equally to all teams. As for being accurate, it is accurate and factual to say that Florida was the most successful team last year under my system. However, it is not accurate to say that Florida was the best team as a result of being my system's most successful team because my system makes no such claim.

11-My system is opinion free because facts determine the standings.

[ktffan]

I wouldn't have to repeat myself over and over if people would finally stop arguing against objective rules based on their misunderstanding of what objective rules are and do. Case in point, your statement that you can't be objective even if you want to. I'm waiting for you explain how the rules I devised aren't objective. Furthermore, I would like to know why a preference for determining the national champion by objective means only indicates that I don't care enough about football to understand the sport. Talk about hot air.

Seriously, dude. You keep saying you're objective, as if you corner the market and are as pure as the snow. If I were to bother to read your rules, I'd be able to tell you the inherent biases there, but since it'd be like telling a brick wall, why bother?

[CJHawkeyes]

Seriously, dude. You keep saying you're objective, as if you corner the market and are as pure as the snow. If I were to bother to read your rules, I'd be able to tell you the inherent biases there, but since it'd be like telling a brick wall, why bother?


Where do I claim I corner the market on objectivity? I only claim that the rules I favor are objective in the same way that the rules used to determine the standings for various sports leagues are objective. No one has to like the rules I do, but the results produced by them do not my reflect my opinion on anything. If I favored ranking teams by records, it wouldn't be my opinion that 12-0 is greater than 11-1, it would be a fact. How about you provide an example of inherent bias in another system. Are there necessarily inherent biases in all objective rules used by all games and competitions? And do you think a bias towards objective accomplishments is equal to bias towards specific teams?