All bow to the new BCS poll

[Sloopy45]

methomps: "On the contrary, that sounds like a decent argument for keeping the humans in the system."

I applaud you for taking the time to do this, but I don't think your analysis means too much in the big picture, and in fact, I think your conclusion is inaccurate.

In all the 'Media Darling' wins, the teams are separated by only decimal points: 0.25, 0.57, 0.33, 0.66, & 0.50. In the one 'Computer' win, the difference between USC & Iowa was more pronounced: 1.16.

And lets also take this into account: sometimes the media is forced into their ranking. Take the USC-Iowa '03 Rose Bowl, for example: the CPU's & media & fans knew USC was better, but there's no way you can justify USC being ranked ahead of the Hawkeyes in the polls when SC has 2 losses & Iowa has only 1 going into the game.

That's a perfect example of how the computers knew better than the pollsters.

And are you sure about some of these computer numbers?? An undefeated Oklahoma has a lower ranking (1.86) than a 1-loss FSU team (1.29) in 2000?? That doesn't seem right.

 

[methomps]

In all the 'Media Darling' wins, the teams are separated by only decimal points: 0.25, 0.57, 0.33, 0.66, & 0.50.

Do you know how many points separated USC and LSU in the final BCS standings in 2003? 0.16.

In the one 'Computer' win, the difference between USC & Iowa was more pronounced: 1.16.

I don't consider my 'study' to be the end of the debate, but I think the numbers are trending in an interesting direction (the small sample size and the 1.16 notwithstanding).

And lets also take this into account: sometimes the media is forced into their ranking. Take the USC-Iowa '03 Rose Bowl, for example: the CPU's & media & fans knew USC was better, but there's no way you can justify USC being ranked ahead of the Hawkeyes in the polls when SC has 2 losses & Iowa has only 1 going into the game.

That's a perfect example of how the computers knew better than the pollsters.

And currently the only example, at least in the scope of BCS bowl games.

And are you sure about some of these computer numbers?? An undefeated Oklahoma has a lower ranking (1.86) than a 1-loss FSU team (1.29) in 2000?? That doesn't seem right.

Yeah, I double-checked that one. Here it is in .pdf form (scroll to the bottom): http://bcsfootball.org/files/bcs-2000.pdf

 

[martinss01]

On the contrary, that sounds like a decent argument for keeping the humans in the system.

why not just feed the proper info into the pc?

i guess i don't understand the premise of your complaint. your stating that the bcs is crap because it is not capable of determining what team will beat what team. ITS NOT SUPPOSED TO! this isn't a piece of software designed to help you determine who you should bet on. it is designed to determine who, based upon their accomplishments to that point, have the right to play for the nc. this is a completely different statement than "who will beat who in what stadium on what saturday". lets say tOSU has the best team in the nation by far. but don't show up against miami and get beat. do they have more right to play say USC in the nc over say LSU even though both USC and LSU are both undefeated? by your definition they do because tOSU would beat USC in the nc game.

think about it this way. take the tOSU 98 team. would you have bet on them to win the big 10? yes, they were clearly the best team. would you have bet on them to win against scum? HELL NO! but why? why wouldn't you have bet on scum to win the big 10? you bet on them to win over tOSU? answer is because scum's team wasn't good enough to win out in the big 10 that year. but because of cosmic circumstance (ie the reasons are obvious) tOSU vrs. scum under coop was anything but a sure bet. so while team A might loose to team B in a head to head, team B is still the team who deserves the crown. make sense?

 

[MililaniBuckeye]

Those were the only games where the humans and computers disagreed about who was the better team.

If they kept SOS and MOV in the formulas the whole time instead of screwing around with the computers, I would bet that the 5-1 human over computer margin would even out, if not go into the computers' favor. Like it was said earlier, garbage in = garbage out.

 

[methomps]

i guess i don't understand the premise of your complaint. your stating that the bcs is crap because it is not capable of determining what team will beat what team. ITS NOT SUPPOSED TO! this isn't a piece of software designed to help you determine who you should bet on. it is designed to determine who, based upon their accomplishments to that point, have the right to play for the nc. this is a completely different statement than "who will beat who in what stadium on what saturday". lets say tOSU has the best team in the nation by far. but don't show up against miami and get beat. do they have more right to play say USC in the nc over say LSU even though both USC and LSU are both undefeated? by your definition they do because tOSU would beat USC in the nc game.

My premise is that if we are going to use the BCS, the human polls should be just as legitimate an element as the computer poll.

The path you are going down is definitely a great cause of confusion that I'm surprised doesn't get brought up more often. Are we trying to find the most accomplished teams? Or the 'best' teams. I think 'best teams' is used more often in the discussion (and there are many archived threads on this topic). For example:

Texas coach Mack Brown initially was against making the coaches' votes public, but said Thursday he will continue to participate next season if asked. He said it's appropriate that just the final regular-season votes are released.

"I've always been a firm believer that the polls start too early in the season and that we shouldn't be voting on a Top 25 before October," Brown said. "By then, we have more information to work with in determining who the best teams are."

Teaff said the coaches have unanimously supported the concept of a championship game because they do not want to go back to the old system where there was a "so-called mythical championship, which did not have the two best teams playing."

What? So some coach of some otherwise inconsequential Sun Belt team can't say who he thinks the best teams in the country are?

The only good thing I have to say is that it has been fairly reasonable with the last couple years but the championship game just seems to be lacking the two best teams in certain scenarios.

I think the vast majority of people want the 'two best teams', which they would define roughly as the most accomplished AND most viable teams. Face it, after the game, we want to be able to declare a 'best team' (as in strongest, most viable), not a 'most accomplished' team.

So far, it seems that the computers can only tell us who the most accomplished teams are. Leave the humans in. For all their shortcomings, they seem apt at identifying better teams.

 

[jwinslow]

Don't feel too upset about the BCS, b/c USC has a free ride to the title game next year based on their previous performance.

If they have an average year where they have a handful of non-convincing victories... then they deserve to be questioned. However there isn't a chance that the polls will look at 2005 objectively without allowing their 2004 success to help their chances.

If OSU and Tennessee both have a great year beating a number of teams that are better than who USC will play, one of the two will get screwed, probably Tennessee.

I'm not here to make fun of USC's scheduling (they usually bring in a good OOC opponent). But next year I would argue there isn't a single team on their schedule that compares to Texas, Michigan or Iowa. I'd put Florida, LSU, Georgia, and the SEC CCG opponent above your schedule as well (Alabama and South Carolina compare to Fresno St and a surprisingly decent 05 pac10 team).

Basically the fact that it tends to work out most of the time doesn't make it right.

 

[Buckeyeskickbuttocks]

My whole beef is about who gets snubbed by the effect of the human polls, not the match ups the BCS provides.

In each of the games Thomps listed, the matchups look to be fairly good whether you take human or puter ranking. Neither human nor computer is picking the winner merely by rating past performance, but instead making matchups.

I'm more interested in the teams like Cal last year that got screwed out of the Rose because of human voters moving them down one spot when Cal had already done all it could do. The irony is that everyone loved the Texas Meshitagain bowl game.

While I think overall humans get the actual power of teams rated pretty close to correct there are some examples of egrigiousness that make me not trust them at all. OU staying ahead of OSU in 04 is another example. Humans get sucked in to hype, where computers do not, is all I'm saying.

 

[sandgk]

My premise is that if we are going to use the BCS, the human polls should be just as legitimate an element as the computer poll.

Actually, not trying to be picky, but that is more of a wish or goal statement than a premise.

A premise fitting your thesis is -- "In the BCS polls the Human Polls are more legitimate than the computer polls."

The path you are going down is definitely a great cause of confusion that I'm surprised doesn't get brought up more often. Are we trying to find the most accomplished teams? Or the 'best' teams. I think 'best teams' is used more often in the discussion (and there are many archived threads on this topic). For example:

I'm not sure I completely follow the distinction you are attempting to draw in the above. In my mind that "better versus more acccomplished" distinction slices deeper when talking about a play-off tournament. For the BCS system are not the better teams over the season also (generally) more accomplished -- over the season?
They are the only ones getting invited to the post-season dance. They have won more games, oftentimes by a greater margin of victory. Whereas, in say the NCAA Basketball tournament, a less accomplished hoops team may in fact be the better team in March. So if the BCS goes away and a play-off is instituted, a football team may be better than all the rest in December and January, though it was merely effective for the first half of the season.
In either instance you reward a team for the timing and duration of its ability to play well. In the BCS you reward those teams that play well over the long haul - almost flawlessly. In a play-off you may reward a team that may only have to come into its own later in a season -- but plays through in stunning fashion, while, say, carrying 2 losses into the tournament.

 

[martinss01]

My premise is that if we are going to use the BCS, the human polls should be just as legitimate an element as the computer poll.

imho, both should be equally important and the human information should be the basis of the computer poll. the human poll should be used to feed information into the pc polls. humans should be setting values for things like "quality wins" and "sos". a pc has 0 ability to determine whether or not playing harvard at home is any different then playing tOSU at home at night. you use humans to input values for things a pc can't possibly understand. but you let the pc crunch the numbers. kinda like what every company on the planet does :p. if you do this, you eliminate at least 90% of the bias in college football. most importantly you create a measureable and transparent system.

a system in which a coach and sit down, and if he had all the info being fed into a pc, would be able to know exactly what the outcome of a season would be before the season began no matter who won what game at what point in the season. it would be 100% completely measureable and everyone would know the rules. this would give teams the ability to create schedules that played towards what they wanted to achieve. USC for example would know whether or not they needed to add a big name for a "sos" win to increase their shot at a nc. that way at the end of hte day, if your team doesn't make the nc. the only one you have to point the finger at is your team.

you need a trasparent unbias system. you will never get that so long as human beings are making the conclusions.

The path you are going down is definitely a great cause of confusion that I'm surprised doesn't get brought up more often. Are we trying to find the most accomplished teams? Or the 'best' teams. I think 'best teams' is used more often in the discussion (and there are many archived threads on this topic).

i think ultimately you have to be looking for the best of the most accomplished teams. if you have tremendous talent on a team and it takes 5 or 6 games to get them on the same page and they drop 2 or 3. i don't care how godlike they are, they flat out don't deserve a shot at the nc. you don't reward someone who doesn't put their all into it just because they are the most talented. you most certainly don't punish a team that leaves it all on the field and wins out because they aren't "pretty". if talent and style were the watchwords of college football tOSU vrs. miami never happens, period. even though they were the only 2 undefeated teams, the mass majority of analysts didn't believe tOSU belonged on the same field as miami. let alone had the right to play them for the nc. THAT is what computers remove from the system. at the end of the day you as a coach know exactly why you didn't make the nc game if there are 3 undefeateds and you can act on it next year. with humans there will always be the question of popularity.

 

[Zurp]

And besides, if you lose a game in CFB you lose all right to bitch about where you get placed ... Bottom line. If you want no arguments, win out. Period.

Whats the world coming to? I'm agreeing with Sloopy45. I had this conversation with a friend right after the BCS final standings in 2003. I said it was funny how the formula "screwed over" USC. He said, "No - USC screwed over USC." And I agree. Once you lose one game, you leave the fate of college football to the rest of the teams and to polls and to formulas.

The only time there's a valid argument is if there're 3 undefeated teams for the two spots. Otherwise, take your 1 loss and go cry elsewhere.

Yeah. I think Auburn has the right to bitch a little after last year.

As for computers vs. human polls, I think they're both important. The strength of the computer polls is that they don't have west coast or east coast biases, or anti-Big Ten bias. They don't care if you lost the first week or the fifth week or the last week. They don't care if the head coach is begging for votes. They don't care if you scored a touchdown on the last play of the game to win by 57 points. They only care what your record is, who you beat, who you lost to, what their records are, who they beat, etc. However, the human poll is important because people actually watch the games. People saw that Miami beat FSU because FSU missed a last-second field goal. That's different from Miami scoring 80 and FSU only scoring 3. People voting know that Ohio State's (6-4) win over Michigan (9-1) was more than just an upset, as its a huge rivalry for both teams.

I think it needs to be a nearly equal combination between computers and human voters.

 

[Vaevictis]

Heh, yeah, 2003/2004 was fun. I have USC, LSU and OU fan friends (and am an OU fan myself).

USC Friend: "My team got screwed!"
My response: "Beat Cal, you don't have this problem."
LSU Friend: "People are saying my team doesn't deserve it!"
My response: "Beat Florida, you don't have this problem."
OU Friend: "People are saying my team doesn't deserve it!"
My response: "Beat Kansas State, you don't have this problem."

2004/2005 was not fun.

Auburn fan: "My team got screwed!"
My response: "Right on."
Utah fan: "My team got screwed!"
My response: "Right on."
Cal fan: "My team got screwed!"
My response: "You got overtaken by the 2nd best team in the Big XII, then got raped by the 4th best team in the Big XII. You got screwed alright, but not how you mean."
USC fan: "Over-rated!"
My response: "STFU! *cry*"
OU fan: "That hurt."
My response: "Yes. It did. A lot."
LSU fan: "Sorry about that, that must have hurt. Screw USC!"
My response: "Right on!"

Heh :)

As far as the polls, BCS, etc goes... the BCS is broken. It will always be broken because there is no way to fairly match up undefeated third place against undefeated first place. Polls are broken too; there's obviously a bias introduced when you lose early as opposed to losing late. You want fair, you need playoffs, and that's all there is to it.

Welcome to an imperfect system my friends; don't be suprised when screwy things happen.

 

[Sloopy45]

Zurp: "Yeah. I think Auburn has the right to bitch a little after last year."

I agree somewhat. I think Auburn had a really good team last year, and they could've beaten OU or USC, IMO. However, why is the Citadel on their schedule? I mean, really ... you're a major program, in a major conference ... there's no need for a bush league move like that. Schedule a bottom rung BCS team (like Missouri, Iowa State, Michigan State, etc.) & the game would be equally as winnable, and you're not hurting your computer points.

I just can't cry for Auburn when their Out of Conference slate includes Southern Baptist A&M of Nevada.

"I think it needs to be a nearly equal combination between computers and human voters."

The problem with the human voters is that you can't filter the bad apples out of the bunch. There's always going to be a Mike Lupica or Bill Rhoden-type with an axe to grind, or many sportswriters who don't watch the games or follow what's going on.

 

[Zurp]

I just can't cry for Auburn when their Out of Conference slate includes Southern Baptist A&M of Nevada.

I agree. The BCS picked the best 2 teams last year. Auburn has a legitimate complaint, but when three teams go undefeated, you need to look at each teams' wins. If Auburn beats The Citadel (I'm assuming your facts are right) and Oklahoma and USC beat Div I-A teams, they're doing more to prove they belong in the NC game.

The problem with the human voters is that you can't filter the bad apples out of the bunch. There's always going to be a Mike Lupica or Bill Rhoden-type with an axe to grind, or many sportswriters who don't watch the games or follow what's going on.

Yeah. "Of course I'm putting USC #1 on my ballot - they haven't lost a game in 16 years." I'm rather torn between human voters and their off-the-wall voting and computer polls and their straight black-and-white decisions.

 

[Sloopy45]

Zurp: "If Auburn beats The Citadel (I'm assuming your facts are right) and Oklahoma and USC beat Div I-A teams, they're doing more to prove they belong in the NC game."

Auburn's OOC slate inculded Louisiana Monroe (31-0), the Citadel (33-3), and Louisiana Tech (52-7). Apparently, Disco Tech wasn't available for a 4th Out of Conference game.

"Yeah. "Of course I'm putting USC #1 on my ballot - they haven't lost a game in 16 years." I'm rather torn between human voters and their off-the-wall voting and computer polls and their straight black-and-white decisions."

I'll tell you this right now, coming from the NorthEast: nobody here gives two shits about CFB. Its gets no attention, and people look at you cock-eyed when you say its your fav sport. The Big East is considered a major conference with major teams up here. That being said, why does Mike Lupica have a ballot? Or anyone from this area for that matter? All these Boston College alums are supposed to be representative of the CFB media in this region? Give me a break.

When is the last time ANY of these guys put their fat-ass in a plane and flew down to South Bend, or Columbus, or Gainseville, or College Station, or any other CFB hot-spot for a game on a Saturday?

 

[Buckeyeskickbuttocks]

Zurp: "If Auburn beats The Citadel (I'm assuming your facts are right) and Oklahoma and USC beat Div I-A teams, they're doing more to prove they belong in the NC game."

Auburn's OOC slate inculded Louisiana Monroe (31-0), the Citadel (33-3), and Louisiana Tech (52-7). Apparently, Disco Tech wasn't available for a 4th Out of Conference game.

"Yeah. "Of course I'm putting USC #1 on my ballot - they haven't lost a game in 16 years." I'm rather torn between human voters and their off-the-wall voting and computer polls and their straight black-and-white decisions."

I'll tell you this right now, coming from the NorthEast: nobody here gives two shits about CFB. Its gets no attention, and people look at you cock-eyed when you say its your fav sport. The Big East is considered a major conference with major teams up here. That being said, why does Mike Lupica have a ballot? Or anyone from this area for that matter? All these Boston College alums are supposed to be representative of the CFB media in this region? Give me a break.

When is the last time ANY of these guys put their fat-ass in a plane and flew down to South Bend, or Columbus, or Gainseville, or College Station, or any other CFB hot-spot for a game on a Saturday?

Of topic a bit, I suppose, but I've witnessed the don't give a crap about CFB first hand. Was up there for the Sandy Ego State game 2003 and was fraeking out cause I didn't know if I'd get to see the game. Found a bar (in the bottom of a hotel I wasn't staying at) which had the game on, walked in, asked dude if I could watch the game (decked out in full OSU regalia) and he said sure. Flipped the game on, and I'm drinking and watching and all is good. Well, as time goes on the bar starts to fill a bit more and I start overhearing people and they're all "Hey, the Sox are on, can't we chaneg the channel?" Fortunately there were enough TV's to accomodate, but here it was, me and my wife watching OSU football, no one watching Michigan - Houston (except me, when OSU was at commercial) and 50 or 60 folks hootin and hollarin about the Sox game. It's cool and all, just different. You go to a bar in C-Bus, and you'll be lucky to see anything BUT football on a Saturday.