Pollspeak: Humans vs. computers

2009-09-08T11:21:08-04:00

2008-12-01T15:24:16-05:00

2008-12-03T15:51:52-05:00

BamaBTT

Redfield 624217

43 These discussions about "the great SEC tiebreaker" are silly. The Big12 lets a poll determine their winner (highest in BCS ranking). The SEC lets a poll determine their loser (throw out the lowest ranked and use head-to-head on the two remaining). Either way, you are not determining the winner on the field. The reason you are not is simple: IN A 3-WAY TIE, THERE WAS NO CLEAR WINNER ON THE FIELD (if you only consider winning vs losing)! If you consider only what happened on the field, you have to take score differentials into account...but only the scores of the three games involving these three teams...not scores with common opponents which encourages meaningless blowouts of cupcakes. So you consider this coding: T=Tech, O=Oklahoma, X=Texas and place them in rank order so TXO would be Tech>Texas>OK and would indicate Tech is the best. The six possible rankings are TOX, TXO, XOT, XTO, OTX, and OXT. Now set the score differentials as OT=44 (TO=-44), OX=-10 (XO=10), TX=6 (XT=-6). You assume each score has a possible fluke error and assign a common variability to the likelihood of a fluke in a given game. You want to find the ranking the maximizes the log-likelihood of these scores. A few simple assumptions finds that the ranking with the maximum log-likelihood is OXT=-1200. TOX, TXO, and XOT have log-likelihoods of -1350. OTX and XTO have likelihoods of -1494 (these are constrained unnormalized log-likelihood scores). So the conclusion from the evidence is that the rating is: OK>TX>TT. Of course, this is what everybody thinks was the ranking except for the fact that they keep searching for some weird head-to-head way out using only wins and losses that does not exist in a 3-way tie (or else it would not BE a three-way tie), then getting tangled up in letting the BCS decide, then insisting that we shouldn't look at final scores. So is relying on polls to resolve this better than relying on the score differentials involving only the teams involved? I can't see it. The other less subjective BCS standard is to look at the strength of schedule...but OK wins that one, too. By the way, I am an Alabama/SEC fan and think both Alabama and Florida would match up better with Texas than Oklahoma...I have no dog in this hunt. But I think it is a bad idea on the parts of all the conferences to let a poll determine their winner...might as well do one-potato-two-potato. 3525844

2008-12-02T14:30:36-05:00

1595

Klee Irwin

Nome 582256

Klee, don't scare me with logic!Seriously though, I think the humans would move UF to #1, to offset whatever nonsense the computers end of doing. Computers only look at numbers, they don't watch the game. They can't see the utter domination the Gators have on both sides of the ball. They don't see how Tim Tebow toys with defenses, or how Harvin, Demps, and Rainey blow by DBs like they were going the other direction.How can UF not gain ground by beating the #1 team on the last day of the regular season? It would signal the absolute END of the BCS and incite riots for Pete's sake!

33 what if Florida barely escapes with a last second FG win, or gets lucky and wins on a controversial call? According to the computers, Texas is ranked way ahead of Florida. They would need a unanimous human poll sweep to be assured of being in the title game, and so its not necessarily a slam dunk 3511154

2008-12-03T01:36:57-05:00

2448

Sandy Underpants

Los Angeles 481510

what if Florida barely escapes with a last second FG win, or gets lucky and wins on a controversial call? According to the computers, Texas is ranked way ahead of Florida. They would need a unanimous human poll sweep to be assured of being in the title game, and so its not necessarily a slam dunk

42 Klee, the BCS final poll will be rigged to make sure it isn't a Texas-OU re-match. That year when Michigan and Ohio St. finished the season 1 vs. 2. The day after they were still 1 and 2, and it wasn't until 2 weeks later AFTER the SEC championship game that Florida slipped past Michigan into the #2 position because of "the computers". "The computers" will again ensure no re-match because "The computers" are 6 guys who can manipulate the outcome of the results to ensure that same division/same conference re-matches don't happen in the BCS championship game. It's telling that undefeated Alabama is ranked #3 today in the Computers and Florida is #6, goes to show how strong the mighty SEC schedule really isn't. And ESPNites and sports reporters should be talking about Alabama's #3 computer ranking every day on every BCS BS roundtable. As a CFB observer I too would rank Alabama behind several other teams because I don't think being undefeated against a bunch of tomato-can, trash bag, bottom-feeding loser teams like Georgia, LSU, Western Kentucky, and Arkansas St. don't mean crap in the real world. 3517330

2008-12-02T22:48:41-05:00

http://img.fannation.com/upload/user_profile/image/622/804/thumb/IMG_0082.jpg 69

IrishAllDay

623444

41 Very true. Human judgment (with its own biases and opinions) has no place in college football. Let's all petition for a play-off! 3516639

2008-12-01T16:18:38-05:00

http://img.fannation.com/upload/user_profile/image/461/788/thumb/chicken.gif 685 NH

Rikj

Derry 462228

1 This is the problem with human polls. A week ago, both the coaches and AP polls had Oklahoma ahead of Texas. This week, after Oklahoma, if anything, improved the merit of their ranking by beating a tough top-15 opponent on the road while Texas played a laugher against a bad team at home, the human polls both have Texas' standing improving, closing the gap to 1 pt in one poll and actually moving ahead in the other. Nothing that happened this week on the field changed anyone's mind. Rather, it was just politicking and debate that apparently caused some voters to move Texas ahead of Oklahoma. The computer rankings have no such bias and cannot be influenced by debate. They go strictly by on the field performance. They had Texas ranked higher last week, and Oklahoma justifiably moving ahead this week due to their performance on the field. Nothing else. 3498783

2008-12-02T22:04:26-05:00

marvel1

Tulsa 600541

This is the problem with human polls. A week ago, both the coaches and AP polls had Oklahoma ahead of Texas. This week, after Oklahoma, if anything, improved the merit of their ranking by beating a tough top-15 opponent on the road while Texas played a laugher against a bad team at home, the human polls both have Texas' standing improving, closing the gap to 1 pt in one poll and actually moving ahead in the other.Nothing that happened this week on the field changed anyone's mind. Rather, it was just politicking and debate that apparently caused some voters to move Texas ahead of Oklahoma. The computer rankings have no such bias and cannot be influenced by debate. They go strictly by on the field performance. They had Texas ranked higher last week, and Oklahoma justifiably moving ahead this week due to their performance on the field. Nothing else.

40 I think the current BCS formula is as reasonable as anything availalbe because it has an element that is free from emotions, politics, regional/conference/coaching tree loyalty and differnt biases. I understand Mac Brown was on air wave (TV & radio) 5-6 times to plead with voters his case for texas even during OU-OSU game time. Bob Stoops did the same once after TT game and declined to be interviewed afterwards. Most of people may not know that Stoops declined to be interviewed during Texas-Tx A & M game and other opportunities. Human polls changed for Texas due to politicking and pleading and different emotional ties to Mac and Tx program even after OU showed a big game against #12 ranked team on the road. It is the sign of contamination on human judgement by emotions. 3516356

2008-12-02T16:37:01-05:00

Tom from Dallas

Dallas 623296

37 You mentioned using the SEC tiebreaker in the Big 12. I don't think it would help the situation. Oaklahoma is 2nd, Texas is 3rd, & Tech is 7th. That puts them within five spots of each other by my math. With Texas-beats-Oaklahoma-beats-Tech-beats-Texas, which team would wind up going? You'd most likely wind up falling back a step to another tie-breaker which would most likely be best BCS ranking & wind up with Oklahoma going as they are under the current situation. 3513429

2008-12-02T17:03:41-05:00

1595

Klee Irwin

Nome 582256

You mentioned using the SEC tiebreaker in the Big 12. I don't think it would help the situation. Oaklahoma is 2nd, Texas is 3rd, & Tech is 7th. That puts them within five spots of each other by my math. With Texas-beats-Oaklahoma-beats-Tech-beats-Texas, which team would wind up going? You'd most likely wind up falling back a step to another tie-breaker which would most likely be best BCS ranking & wind up with Oklahoma going as they are under the current situation.

39 The SEC throws out the lowest ranked team nits. Can people really not read or is this just too difficult for some peabrains to grasp? 3513822

2008-12-02T16:37:01-05:00

Tom from Dallas

Dallas 623296

2008-12-02T16:59:00-05:00

http://img.fannation.com/upload/user_profile/image/884/14/thumb/al.jpg 81 CA

Franco

San Francisco 88714

You mentioned using the SEC tiebreaker in the Big 12. I don't think it would help the situation. Ok is 2nd, Texas is 3rd, & Tech is 7th. That puts them within five spots of each other by my math. With Texas-beats-Ok beats-Tech-beats-Texas, which team would wind up going?

38 There are several options that would work. In the Pac 10, for example, were the Trojans to lose to UCLA (unlikely) you have a three way tie with the Duck, Beavers and USC. They each own head to head loses against one another (Ducks vs. Beaver, OSU vs, USC, Trojans vs. Oregon); in the Pac 10, we go to the next team - however, in the Big 12 South, each of the top teams beat each of the bottom three teams (the Cowpokes, Baylor and A&M); so you can look at the Big 12 North: Nebraska and Missouri are both tied, but the Tigers win the head to head. The Sooners and Red Raiders both defeated Nebraska but didn't play the Tigers, which did play (and lost) to the Longhorns. Unfortunately, this gives us another SOL scenario. I still believe head to head should count for something, but this was an unusual and unfortunate situation that should NOT have been left to the BCS. 3513711

2008-12-02T16:37:01-05:00

Tom from Dallas

Dallas 623296

2008-12-02T13:08:17-05:00

http://img.fannation.com/upload/user_profile/image/622/455/thumb/oklahoma_1280x1024B.jpg 4 OK

SoonerBredEd

Jones 623095

31 TEXAS45OU35, Spoken like a true texas big mouth. You and all shorthorns know that you squeaked by OU on a bad day. You and all shorthorns know that the last thing in the world whiney Mack Brown and team wants is a rematch. The Sooners would crush you this time like we did Tech (and on another neutral field). You say Tech squeaked by you and at their house. Home field advantage is good for 3 points. So how good is Texas tying them or winning by one? Our 44 point pounding of the same team a couple weeks later speaks volumes. Yes it was at our house. Subtract 3 and it???s a 41 point pounding. The Texas win over Oklahoma looks much more like a temporary let down by a great team to a good team, than as proof that Texas is a better team. Stop running Tech down as a team. It only makes Texas look worse. The body of work since that date indicates Oklahoma is the better team. Pray for a rematch shorthorns. 0.935 to 0.922 settled on a neutral CPU. Gotta go now, on my way to Kansas City. 3509889

2008-12-02T16:17:02-05:00

http://img.fannation.com/upload/user_profile/image/485/588/thumb/hd.bmp 1307

GhostofTCUFanFuture

486028

TEXAS45OU35,Spoken like a true texas big mouth. You and all shorthorns know that you squeaked by OU on a bad day. You and all shorthorns know that the last thing in the world whiney Mack Brown and team wants is a rematch. The Sooners would crush you this time like we did Tech (and on another neutral field).You say Tech squeaked by you and at their house. Home field advantage is good for 3 points. So how good is Texas tying them or winning by one? Our 44 point pounding of the same team a couple weeks later speaks volumes. Yes it was at our house. Subtract 3 and it???s a 41 point pounding.The Texas win over Oklahoma looks much more like a temporary let down by a great team to a good team, than as proof that Texas is a better team. Stop running Tech down as a team. It only makes Texas look worse. The body of work since that date indicates Oklahoma is the better team.Pray for a rematch shorthorns. 0.935 to 0.922 settled on a neutral CPU.Gotta go now, on my way to Kansas City.

36 what a numbnuts, if everything you say is true then why has OU lost at Tech the last two times 3513015

2008-12-02T11:50:46-05:00

http://img.fannation.com/upload/user_profile/image/568/563/thumb/txwhiner__2_.jpg 28 WA

FIR3SIDE

Tacoma 569043

NYsooner,That's the pot calling the kettle black. You are taking time out of your ever so important day to tell Texas fans to get over and and to get a life. If it really was no big deal to you either way, you wouldn't bother to post a thing here. As such, you negate your own derisions by your appearance on this board deriding Longhorns. OU deserved a shot, but the Longhorns simply deserved it more. No one is talking about forgetting the TT loss and in fact, should we not choke in that game, OU isn't even in the discussion. A one second loss on an away field against an undefeated team, or a 10 point loss against an undefeated rival on a neutral field, or a 35 point loss on an away field? Which game was the anomaly? The Tech OU game was the biggest, not by merit of the victory, but because of the away scenario and because Tech folded up in unique fashion because of their lack of big game experience. As such, Tech isn't part of the discussion in any logical scenario and their performance against Baylor underlines this reality. That brings up OU's 10 point loss. Again, I'm not saying Ou isn't deserving, they just aren't as deserving. Tech saved your chances by pulling of a miracle at home against a superior team that stumbled and still gave Tech hell on its way to a narrow and, yes, a loss that wouldn't happen again.

28 Wow. I like how you said it was an anomaly. 45 penalty yards in McCoy flopping as if he we're Bill Laimbeer, a early celebration by our cornerback who intercepted McCoy in the endzone that was ruled out just because he didn't 'stand up' with the ball, 2 of our top defensive players out(Reynolds and English) which leaves us with 1 starter from last year, Demarco Murray not completely 100%. Anyway who flops in football as if it were the NBA? Oh yea McCoy does. You TX fans are still in denial about the Tech win. I am praying that no other team in the Big XII gets a BCS bid. That way OU can show you that "......a loss wouldn't happen again." Yea sure Bradford's non throwing hand has torn ligaments and our top 2 middle linebackers are out for the season and Englishis questionable. But f$%k it. 3508604

2008-12-02T16:08:05-05:00

http://img.fannation.com/upload/user_profile/image/485/588/thumb/hd.bmp 1307

GhostofTCUFanFuture

486028

Wow. I like how you said it was an anomaly. 45 penalty yards in McCoy flopping as if he we're Bill Laimbeer, a early celebration by our cornerback who intercepted McCoy in the endzone that was ruled out just because he didn't 'stand up' with the ball, 2 of our top defensive players out(Reynolds and English) which leaves us with 1 starter from last year, Demarco Murray not completely 100%. Anyway who flops in football as if it were the NBA? Oh yea McCoy does. You TX fans are still in denial about the Tech win. I am praying that no other team in the Big XII gets a BCS bid. That way OU can show you that "......a loss wouldn't happen again." Yea sure Bradford's non throwing hand has torn ligaments and our top 2 middle linebackers are out for the season and Englishis questionable. But f$%k it.

35 lol,which sooner fan was just screaming about TX fans wanting to rewrite history? it was a horribly officiated game on both sides of the ball. and your DB didn't drop the ball getting up celebrating. even the commentators that day said the replay got it right - you have to control the ball when you hit the ground.change it how you want, TX still won 3512835

2008-12-02T14:48:02-05:00

Oscar Holguin

Odessa 166606

34 I think that Texas should have gone ahead of Oklahoma on the simple fact that they played the starters up until the score was out of hand (11 minutes left in the 4th qtr against Texas A & M) while Oklahoma still had their startes in until almost 5 minutes left when they defeated Texas Tech. Oklahoma was ahead at 42-7 at half time and had the game in control by the start of the 4th qtr and should have pulled the starters out but instead the ran up the score for "style points" as to where Mack Brown pulled his starters out to let his back ups get some playing time or to avoid injury. 3511439

<div class="photo_container image_right"> <div class="photo"><img src="http://sportsillustrated.cnn.com/2008/images/12/01/colt-mccoy-swp.jpg" alt="Double Click to select a Photo" /></div> <div class="photo_attributes" style="text-align:right;">Colt McCoy and Texas had a slight edge in the human polls.</div> <div class="photo_attributes" style="text-align:right;">AP</div> </div> <p><em>By Hugh Falk, <a href="http://www.pollspeak.com/" target="_blank">Pollspeak.com</a></em><br /><br />The BCS rankings took on extra importance this week when the Big 12 used them to determine its South Division champion, as Oklahoma is ranked No. 2 in the BCS ahead of No. 3 Texas. Last week I predicted an Oklahoma "victory" if all three Big 12 South leaders won their games, which they did. However, I expected the human voters to give them the edge, and I was wrong on that one.</p> <p>Texas actually has a slight (almost negligible) edge in the human polls, although both teams average a No. 3 ranking. Texas is ranked No. 3 in both polls while Oklahoma is No. 4 in Harris and No. 2 in the Coaches' Poll. The real difference is in the computer rankings, with Oklahoma getting three first place votes to Texas' one.</p> <p>As I've mentioned in the past, the computers have no long-term concept of head-to-head match ups (though arguably, human voters don't, either). So, it doesn't surprise me that Oklahoma ended up on top. However, if the Big 12 (or Texas fans) don't like how the tiebreaker played out, they shouldn't blame the BCS. Instead, they should enforce a head-to-head rule just like the SEC does. The SEC tiebreaker is similar to the Big 12, but if the teams are within five spots of each other, the head-to-head results are the deciding factor instead. That's a great safety valve to override the voters and computers who ignore head to head results.</p> <p>While the BCS' top four is always interesting, the bottom four this week is also interesting:</p> <p>No. 22 <a href="http://pollspeak.com/pollstalker/pollstalker.php?r=T&s=5&p=8&w=15&t1=71&t2=0&v=0" target="_blank">Northwestern</a> was ranked No. 22 in both human polls, but only cracked the top 25 in one computer rating, Anderson/Hester, which ranked the Wildcats No. 25. Yet that was still enough to keep them at No. 22 in the BCS.</p> <p>No. 23 <a href="http://pollspeak.com/pollstalker/pollstalker.php?r=T&s=5&p=8&w=15&t1=80&t2=0&v=0" target="_blank">Pittsburgh</a> similarly keeps the same ranking it received from human voters in the BCS.</p> <p>Meanwhile, the last two, No. 24 <a href="http://pollspeak.com/pollstalker/pollstalker.php?r=T&s=5&p=8&w=15&t1=30&t2=0&v=0" target="_blank">Florida State</a> and No. 25 <a href="http://pollspeak.com/pollstalker/pollstalker.php?r=T&s=5&p=8&w=15&t1=112&t2=0&v=0" target="_blank">Virginia Tech</a>, are just the opposite.  They were not ranked by either Harris Interactive or the Coaches, yet both teams still snuck into the BCS based on the strength of their computer rankings. That is a rarity since the computers have been relegated to one-third of the BCS formula, but it shows that the computers can make a difference at the top and the bottom of the BCS.</p> <p>As for the humans, most people obviously voted Alabama No. 1, but if you look at who received the second most first-place votes (overall) it was neither Texas nor Oklahoma, but rather Florida. The Gators dismantled the Seminoles in the swamp (not the "Swamp" in Gainesville ... the temporary swamp in Tallahassee caused by constant rain), and have impressed at least a few new voters in Florida and elsewhere.<span style="font-size: 11pt; line-height: 115%; font-family: Calibri;"><span> </span></span> Looking at the AP ballots, <a href="http://pollspeak.com/petition.html" target="_blank">which are the only ones made available each week</a>, the <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&v=0&t1=27&w=15&r=T" target="_blank">three No. 1 votes for Florida</a> come from two California voters and one Virginian. If Florida can beat Alabama in Atlanta, the Gators will surely get several more No. 1 votes.</p> <p>One last note about Florida. The Gators were the only SEC winner in the four SEC vs. ACC matchups this weekend. The ACC definitely proved itself a solid league. The Georgia Tech victory over Georgia was particularly impressive, and even though both teams now have the same record, <a href="http://pollspeak.com/pollstalker/pollstalker.php?s=5&p=9&t1=32&t2=33&v=0&w=15&r=T&o1=diff" target="_blank">three voters still ranked the Bulldogs over the Yellow Jackets</a>. However, with Florida beating FSU soundly, the ACC also showed it has no top 10-caliber teams this year. Sagarin ranked the ACC as the strongest conference last week, and Anderson/Hester did this week, too. Still, it's a dubious honor to be the strongest conference but to not have had a shot at the national championship since week one. Although, one could argue that if Georgia Tech and Boston College didn't play in such a tough league (their only losses are to ACC teams), they could be undefeated right now. But let's table that argument until after the bowls.</p> <p>For more poll analysis, go to <a href="http://www.pollspeak.com/" target="_blank">pollspeak.com</a>.</p>

The SI Staff

New York City 21244 28043 <div class="photo_container image_right"> <div class="photo"><img src="http://sportsillustrated.cnn.com/2008/images/12/01/colt-mccoy-swp.jpg" alt="Double Click to select a Photo" /></div> <div class="photo_attributes" style="text-align:right;">Colt McCoy and Texas had a slight edge in the human polls.</div> <div class="photo_attributes" style="text-align:right;">AP</div> </div> <p><em>By Hugh Falk, <a href="http://www.pollspeak.com/" target="_blank">Pollspeak.com</a></em><br /><br />The BCS rankings took on extra importance this week when the Big 12 used them to determine its South Division champion, as Oklahoma is ranked No. 2 in the BCS ahead of No. 3 Texas. Last week I predicted an Oklahoma "victory" if all three Big 12 South leaders won their games, which they did. However, I expected the human voters to give them the edge, and I was wrong on that one.</p> <p>Texas actually has a slight (almost negligible) edge in the human polls, although both teams average a No. 3 ranking. Texas is ranked No. 3 in both polls while Oklahoma is No. 4 in Harris and No. 2 in the Coaches' Poll. The real difference is in the computer rankings, with Oklahoma getting three first place votes to Texas' one.</p> <p>As I've mentioned in the past, the computers have no long-term concept of head-to-head match ups (though arguably, human voters don't, either). So, it doesn't surprise me that Oklahoma ended up on top. However, if the Big 12 (or Texas fans) don't like how the tiebreaker played out, they shouldn't blame the BCS. Instead, they should enforce a head-to-head rule just like the SEC does. The SEC tiebreaker is similar to the Big 12, but if the teams are within five spots of each other, the head-to-head results are the deciding factor instead. That's a great safety valve to override the voters and computers who ignore head to head results.</p>