Pollspeak: Humans vs. computers

2008-10-21T17:24:23-04:00

2008-10-20T14:35:16-04:00

2008-10-21T17:24:23-04:00

533

beachgator

Lucedale 174427

22 I don't get it...where's Maryland?!? They hate the turtle? Go Terps!!! 3112435

2008-10-21T14:59:07-04:00

http://img.fannation.com/upload/user_profile/image/359/778/thumb/USA_rowing.jpg 42 CA

Stewart L

Oakland 360218

19 "Likewise, Craig James, Kirk Herbstreit, Maurice Patton, Glenn Guilbeau and Jason Franchuck rank Kansas over USF even though the Bulls have a better record and beat Kansas." Kansas lost by 3 to USF at USF (Home teams on average have a 3 point advantage). Kansas schedule, though weak, is tougher than USF. Sagarin's "Predictor," which is the most accurate measure of a team (by accurate, it is the best predictor of future performances), has Kansas ranked ahead of USF. Kudos to James, Herbstreit, Patton, Guilbeau and Franchuck 3110548

2008-10-21T16:30:11-04:00

1595

Klee Irwin

Nome 582256

"Likewise, Craig James, Kirk Herbstreit, Maurice Patton, Glenn Guilbeau and Jason Franchuck rank Kansas over USF even though the Bulls have a better record and beat Kansas."Kansas lost by 3 to USF at USF (Home teams on average have a 3 point advantage). Kansas schedule, though weak, is tougher than USF. Sagarin's "Predictor," which is the most accurate measure of a team (by accurate, it is the best predictor of future performances), has Kansas ranked ahead of USF. Kudos to James, Herbstreit, Patton, Guilbeau and Franchuck

21 so frickin what? they played each other. That trumps everything. Not only that, Kansas has 2 losses. head to head is always the initial tiebreaker if teams have the same record. Here, the teams don't even have the same record. What a ditz 3111916

2008-10-21T15:55:23-04:00

Pollspeak

169617

20 While I admire the Sagarin Ratings, these voters don't vote in the Sagarin Poll..it is the AP Poll. The AP guidelines clearly state "Pay attention to head-to-head results." As for SOS, one team can play the toughest teams in the nation and not beat any of them. That doesn't make them better than a team with a weaker schedule and more wins. We have to look at the strength of the teams beaten. The best team Kansas has beaten is Colorado. The best team USF has beaten is ...well, Kansas. It seems like a pretty open and shut case (this week...it can change when Kansas plays more difficult teams and wins). Finally, even Sagarin ranks USF over Kansas in both the ELO_CHESS and standard ratings. Good rankings don???t try to tell the future (which is what the Predictor rating tries to do). They report on actual performance (which is another AP guideline). There is no reason to predict the future. There are several weeks left for rankings to change based on actual performance. If voters would rank teams based on what they see instead of what they feel, the polls would be less biased and would still all work out in the end. 3111455

2008-10-21T14:59:07-04:00

http://img.fannation.com/upload/user_profile/image/359/778/thumb/USA_rowing.jpg 42 CA

Stewart L

Oakland 360218

2008-10-21T13:00:19-04:00

ChunkyChuck

570649

18 The problem with a playoff is that any team MIGHT beat any other team one time (i.e. Ole Miss/Florida, Oregon State/USC, Giants/Patriots). What we really need is a playoff that has a best out of 3 series for each match-up. More like baseball. OK, that's completely unrealistic, but the point is that football play-offs aren't perfect either, and sometimes the better team doesn't get to advance. Right now in college football, the whole season is a playoff, and that's why it is so fun. 3108909

2008-10-21T12:38:28-04:00

CJ MN1

Saint Cloud 589039

17 This is why the NCAA and various conference commissioners want nothing to do with a playoff. What other major sport brings out this much passion, anger, anguish, etc... ??As much as many of us think this BCS is total BS---why would they (ref: Communist NCAA and conference leaders) change when it brings in so much attention (read: money). What do they care, as long as one of their own conference members is in a major B-S Bowl to rake in some major $$$$... They are laughing their way all the way to the bank. Meanwhile, the players and the fans get bum-doped. Hey... until there is a true playoff (minimum 8 teams)---the "BCS" championship trophy is only worth as much as the crystal it is made of. Hey, if the NFL played this way... New England would be the "champs" not the Giants... (the Giants wouldn't even have been in the big game)... The BCS reminds me of being a kid... "my daddy is tougher than your daddy"---- my team is better than your team... ha!ha! ha! What fools we are!! Go Golden Gophers!! 3108630

2008-10-20T18:08:48-04:00

norcalfella

Dublin 536285

5 I do understand voting teams ahead of teams they have beaten but... while it seems easy to say GT then BC then VT in the ACC let's not forget VaTech beat Georgia Tech! It doesn't always work out so I wouldn't give these voters too hard of a time. On the other hand, Craig James and "Herbie" should have their voter card revoked for putting Kansas ahead of USF. Come on now. 2 losses for the Jayhawks makes it even worse. Losing is still losing even if the game is relatively close. 3103132

2008-10-21T12:06:22-04:00

http://img.fannation.com/upload/user_profile/image/587/801/thumb/200901091229066d5626975ag7_1_.gif 10 GA

GlennW

Smyrna 588401

I do understand voting teams ahead of teams they have beaten but... while it seems easy to say GT then BC then VT in the ACC let's not forget VaTech beat Georgia Tech! It doesn't always work out so I wouldn't give these voters too hard of a time.On the other hand, Craig James and "Herbie" should have their voter card revoked for putting Kansas ahead of USF. Come on now. 2 losses for the Jayhawks makes it even worse. Losing is still losing even if the game is relatively close.

16 You may not be aware, but GA Tech beat Boston College at Boston College and then lost to Virginia Tech at Virginia Tech in the final moments of the game after having the lead when a questionable call gave the Hokies a first down when GA Tech's defense had stopped VA Tech and would have iced the game for GA Tech. This allowed Virginia Tech to move into position to score their final points to take the lead. Georgia Tech still almost won the game when a sure touchdown pass was missed by inches. If that would have happened, GA Tech would be undefeated, including road wins at Boston College, Virginia Tech, and Clemson. At this juncture, however, their only loss is one on the road, at Virginia Tech, and by less than a touchdown. And, as a simple reminder, Paul Johnson has accomplished this with one of the youngest teams in the BCS, a limited number of players on scholarship (all of which were recruited for a pro-style of offense), a late recruiting class, and having implemented new offensive and defensive schemes since taking over the program the first of this year. Bodes well for the program, wouldn't you say? 3108206

2008-10-21T11:30:16-04:00

Tzarius

Omaha 66882

15 Well The BCS is better than what we had in the past lol. 3107706

2008-10-20T23:22:08-04:00

172

ceteris paribus

Athens 532894

Computer polls DO have bias. Statistics don't tell you everything. A team can play lousy early in the season or just the "wrong" game early, and the person(s) composing the data for a poll putting weights on what he/she thinks is significant can still screw up (or just skew) the poll and wind up with really strange (aka stupid) rankings. The human polls are supposed to balance the statistics with the intangibles... in theory. Theory regularly fails when it comes to polls. Remember, even the computer polls are designed by people, so they're just as subject to errors as the human polls are. Supposedly, the computer poll errors apply equally to every team, but still, I don't trust any poll or any combination of polls to determine which team is really better than any other team--not at the beginning of the year, not at the end of the year. Just let the players decide it on the field. Period.Strength of schedule is a myth--there is no Transitive Law of football.

13 Strength of Schedule does not mean Transitive Law of Football. It does, however, attempt to evaluate the level of competition each team faces. Check out Jeff Sagarin's football rankings on the USA Today website. It gives a rough explanation of how he determines strength of schedule: it is a function of the rank of the teams played and the location of the game . . . he does claim the accuracy improves as teams become "connected" but I'm fairly sure that is because it provides a basis for performance not a direct prediction of a result. That is to say, if you have the SEC East where Georgia and Florida play at least three common opponents, if, on average Florida wins by 15 points in those games whereas on average Georgia wins by 12 points (against a sample of like opponents) it suggests Florida is the better team. 3104517

2008-10-20T23:58:07-04:00

222

GlennAllen

Richmond 185259

Strength of Schedule does not mean Transitive Law of Football. It does, however, attempt to evaluate the level of competition each team faces. Check out Jeff Sagarin's football rankings on the USA Today website. It gives a rough explanation of how he determines strength of schedule: it is a function of the rank of the teams played and the location of the game . . . he does claim the accuracy improves as teams become "connected" but I'm fairly sure that is because it provides a basis for performance not a direct prediction of a result. That is to say, if you have the SEC East where Georgia and Florida play at least three common opponents, if, on average Florida wins by 15 points in those games whereas on average Georgia wins by 12 points (against a sample of like opponents) it suggests Florida is the better team.

14 And that's why conference standings mean something: for the most part, all the same teams play each other (sort of--divisions complicate the issue somewhat). Conference standings are based on the teams' records against each other, so they mean something. So, which Oregon St. team did USC play? The same one that Stanford played? I don't think so (or, clearly, Stanford will have no problem beating USC... then again, they might :D). Conference standings mean something; polls mean nothing. The BCS is worthless as anything other than filling the rosters for a bunch of bowl games; the BCS doesn't determine any national champion. 3104611

2008-10-20T21:57:18-04:00

222

GlennAllen

Richmond 185259

that's why you use multiple computer polls which don't have any bias and do weigh the games from 6 weeks ago as well as one week ago nitcakes. the problem is they put too much emphasis on the media polls

12 Computer polls DO have bias. Statistics don't tell you everything. A team can play lousy early in the season or just the "wrong" game early, and the person(s) composing the data for a poll putting weights on what he/she thinks is significant can still screw up (or just skew) the poll and wind up with really strange (aka stupid) rankings. The human polls are supposed to balance the statistics with the intangibles... in theory. Theory regularly fails when it comes to polls. Remember, even the computer polls are designed by people, so they're just as subject to errors as the human polls are. Supposedly, the computer poll errors apply equally to every team, but still, I don't trust any poll or any combination of polls to determine which team is really better than any other team--not at the beginning of the year, not at the end of the year. Just let the players decide it on the field. Period. Strength of schedule is a myth--there is no Transitive Law of football. 3104098

2008-10-20T23:22:08-04:00

172

ceteris paribus

Athens 532894

By Hugh Falk, <a href="http://www.pollspeak.com/" target="_blank">Pollspeak.com</a> Today's blog answers some burning questions: WHO NEEDS THE BCS? The big poll news this week is the <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=3&t2=&v=406&w=9&r=V" target="_blank">BCS standings</a> were released, but the even bigger news is this is the first week we don't need the BCS to tell us who should play for the national championship. For the first time in BCS history, we have a consensus No. 1 (<a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&v=403&t1=94&w=9&r=T" target="_blank">Texas</a>) and No. 2 (<a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&v=403&t1=3&w=9&r=T" target="_blank">Alabama</a>). Every BCS component and even the AP poll agree on the top two teams. This could last as long as both teams keep winning. Judging by recent history, that might only be a week. DID THEY SEE THE GAME? While the overall rankings are in agreement, that doesn't mean the people who vote in the polls are.  Once again, we present our list of people who have forgotten or ignored what happened on the field: <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=13&t2=112&v=47&w=9&r=V" target="_blank">Myron Patton</a>, <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=13&t2=112&v=26&w=9&r=V" target="_blank">Jason Franchuk</a> and <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=13&t2=112&v=147&w=9&r=V" target="_blank">Marcus Fuller</a> <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=112&t2=13&v=47&w=9&r=T&o1=diff" target="_blank">rank Virginia Tech over Boston College</a> even though the Eagles have a better record and just beat the Hokies. I can't think of a good reason for someone to vote that way unless they're following PETA's agenda and striking a blow for turkeys against the pilgrims' hometown. Likewise, <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=41&t2=106&v=14&w=9&r=V" target="_blank">Craig James</a>, <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=41&t2=106&v=40&w=9&r=V" target="_blank">Kirk Herbstreit</a>, <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=41&t2=106&v=398&w=9&r=V" target="_blank">Maurice Patton</a>, <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=41&t2=106&v=23&w=9&r=V" target="_blank">Glenn Guilbeau</a> and <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=41&t2=106&v=26&w=9&r=V" target="_blank">Jason Franchuck</a><a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=41&t2=106&v=26&w=9&r=V"></a> <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=41&t2=106&v=0&w=9&r=T&o1=diff" target="_blank">rank Kansas over USF</a> even though the Bulls have a better record and beat Kansas. Interestingly, the only BCS component that agrees is <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=106&t2=41&v=403&w=9&r=V" target="_blank">The Billingsley Report</a>, which is also the only computer that starts the season with the rankings from last season. <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=106&t2=41&v=403&w=1&r=V" target="_blank">Billingsley's preseason ballot</a> has Kansas ranked No. 2 and USF No. 33, which is a significant gap. It could be suffering from preseason bias along with some of the AP voters. <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=80&t2=106&v=0&w=9&r=T&o1=diff" target="_blank">A majority of AP voters</a> still <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=106&t2=80&v=403&w=9&r=T&o1=diff" target="_blank">rank USF over Pittsburgh</a> even though they both have one loss and the Panthers beat the Bulls at home. Again, only the human polls and Billingsley agree, and every other computer poll has Pittsburgh ranked better. Finally, in the "we forgot about that" group we have Georgia Tech going to Chestnut Hill and beating Boston College in the second game of the year. The Yellow Jackets also have one more win than the Eagles and yet 17 voters still <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=33&t2=13&v=0&w=9&r=T&o1=diff" target="_blank">voted BC over GT.</a> JAMES MADISON? <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=3&t2=0&v=402&w=9&r=V" target="_blank">Sagarin's ELO_CHESS rankings</a>, which are used in the BCS, have James Madison at No. 39 over the likes of Kentucky, California, Nebraska and Notre Dame. Ironically, at 6 -1 the Dukes only loss came to Duke, who is also ranked lower than James Madison. Unfortunately for the Dukes, the BCS doesn't accept FCS schools, and they can never realize their BCS dreams. HOW DID THEY GET IN? While there were no surprises at the top of the BCS standings, there were a few at the bottom.  Some teams were probably surprised to make it in the top 25 including: No. 25 <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=30&v=406&w=9&r=T" target="_blank">Florida State</a>, No. 23 <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=41&v=403&w=9&r=T" target="_blank">Kansas</a> and No. 21 <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=16&v=403&w=9&r=T" target="_blank">BYU</a>. To get BCS points from the computers you have to be in their top 25, and these teams are not computer favorites. FSU and BYU are only ranked by two computers and Kansas by one.  Luckily for Kansas, the Billingsley Report's high No. 15 ranking kept it in the BCS standings. Billingsley ranked BYU at No. 16 and the Colley Matrix helped with a No. 24. Meanwhile, FSU just squeaked in with a No. 23 ranking from Colley and a No. 25 ranking from Anderson and Hester. For more poll analysis, go to <a href="http://www.pollspeak.com/">pollspeak.com.</a>

The SI Staff

New York City 21244 15941 By Hugh Falk, <a href="http://www.pollspeak.com/" target="_blank">Pollspeak.com</a> Today's blog answers some burning questions: WHO NEEDS THE BCS? The big poll news this week is the <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&t1=3&t2=&v=406&w=9&r=V" target="_blank">BCS standings</a> were released, but the even bigger news is this is the first week we don't need the BCS to tell us who should play for the national championship. For the first time in BCS history, we have a consensus No. 1 (<a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&v=403&t1=94&w=9&r=T" target="_blank">Texas</a>) and No. 2 (<a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=8&v=403&t1=3&w=9&r=T" target="_blank">Alabama</a>). Every BCS component and even the AP poll agree on the top two teams. This could last as long as both teams keep winning. Judging by recent history, that might only be a week. DID THEY SEE THE GAME? While the overall rankings are in agreement, that doesn't mean the people who vote in the polls are.  Once again, we present our list of people who have forgotten or ignored what happened on the field: <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=13&t2=112&v=47&w=9&r=V" target="_blank">Myron Patton</a>, <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=13&t2=112&v=26&w=9&r=V" target="_blank">Jason Franchuk</a> and <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=13&t2=112&v=147&w=9&r=V" target="_blank">Marcus Fuller</a> <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=112&t2=13&v=47&w=9&r=T&o1=diff" target="_blank">rank Virginia Tech over Boston College</a> even though the Eagles have a better record and just beat the Hokies. I can't think of a good reason for someone to vote that way unless they're following PETA's agenda and striking a blow for turkeys against the pilgrims' hometown.