Human polls should be eliminated - cougarguard.com — unofficial BYU Cougars / LDS sports, football, basketball forum and message board

jay santos · 10-09-2006, 06:49 PM

Computer polls blow away human polls in accuracy--based on looking at game outcomes.

Human polls will always have bias, and the little guys will always be screwed.

Looking at the top 38, counting the "others receiving votes", two non-BCS schools make it. Boise State at #19, BYU at #35

In my computer poll you have:

7 Boise State
23 Tulsa
29 Navy
31 BYU
32 Southern Miss
39 Utah barely misses it

Boise State is undefeated with a better SOS than #4 West Virginia, and similar SOS to one loss teams Iowa, Nebraska, Texas all ranked higher in the human polls.

That's called getting screwed by the system.

Sports media guys would say the computer has something wrong in its formula.

Power_rating=If(conf="bcs",true_power_rating+10,tr ue_power_rating)

That's the missing formula the computers are screwing up.

TheSizzle36 · 10-09-2006, 06:54 PM

At the same time, humans are the only ones who can actually watch the games and analyze what is happening. I think a better solution would be to release 1 pre-season poll, more of a rankings just for kicks and giggles for the fans, and then after 5 or 6 weeks, or say, October 1st, the first polls are released. Pre-season bias has hurt WAY too many teams in the past.

While the human element can hurt the little guys like Boise, and BYU, the human element also has the ability to look at external factors that may not show up in a computer poll like rivalries, injuries, players sitting out due to suspensions, etc. Computer polls should definitely play a part, but they should not be the end all.

jay santos · 10-09-2006, 07:03 PM

Quote:

Originally Posted by TheSizzle36

At the same time, humans are the only ones who can actually watch the games and analyze what is happening. I think a better solution would be to release 1 pre-season poll, more of a rankings just for kicks and giggles for the fans, and then after 5 or 6 weeks, or say, October 1st, the first polls are released. Pre-season bias has hurt WAY too many teams in the past.

While the human element can hurt the little guys like Boise, and BYU, the human element also has the ability to look at external factors that may not show up in a computer poll like rivalries, injuries, players sitting out due to suspensions, etc. Computer polls should definitely play a part, but they should not be the end all.

At the end of the season, there are over 700 game results. If any human has watched all 700 games, could pass a test showing memory of each one--who won, by how much, and who was at home, and could process that information with no preconceived bias towards specific conferences or teams or geographic regions, then that human should be allowed to vote.

The benefit of a human being able to weigh those external factors falls way short of the disadvantage the human has in bias and insufficient analysis of all games.

TheSizzle36 · 10-09-2006, 07:29 PM

Quote:

Originally Posted by jay santos

At the end of the season, there are over 700 game results. If any human has watched all 700 games, could pass a test showing memory of each one--who won, by how much, and who was at home, and could process that information with no preconceived bias towards specific conferences or teams or geographic regions, then that human should be allowed to vote.

The benefit of a human being able to weigh those external factors falls way short of the disadvantage the human has in bias and insufficient analysis of all games.

I don't need to watch all 700 games to have a grasp at what is going on in College Football. Week by week, one can certainly look at the games that are going on and understand what happens. If there are 50 games played in a week (assuming 116 NCAA teams, with appx. 16 teams having a bye week) one can certainly know what is going on enough to formulate a valid opinion on said teams.

I also think that there is room for a computer poll. But a computer poll should not be the only source of information.

jay santos · 10-09-2006, 08:21 PM

Quote:

Originally Posted by TheSizzle36

I don't need to watch all 700 games to have a grasp at what is going on in College Football. Week by week, one can certainly look at the games that are going on and understand what happens. If there are 50 games played in a week (assuming 116 NCAA teams, with appx. 16 teams having a bye week) one can certainly know what is going on enough to formulate a valid opinion on said teams.

I also think that there is room for a computer poll. But a computer poll should not be the only source of information.

This is a subject I've been passionate about for 20 years. I've always hated the human polls.

This is how I'd do it. Take a 30 computer poll average. Use the best 30 you can find based on fit to historical game scores. Throw out human polls.

Then if you want a selection committe to run a tournament of place teams in bowl games, I could go for that, taking into a human element, as long as the primary criteria was the human polls.

You think you can get a handle on 50 games, but you can't. You forget about games a few weeks ago, or you ignore games involving non-big name teams. You can't evaluate SOS unbiasedly, etc.

Detroitdad · 10-09-2006, 11:17 PM

I share your distaste for the human polls. The logical inconsistencies of voters are obvious. The potential for bias is huge, and the laziness in their compilation is stunning.

However, you and I both know that the human element will never be relinquished because they will deny that their biases exist. They never let the nerds construct the rules, whenever there is a populist streak in an item. That is why nerds gravitate toward engineering and law.

il Padrino Ute · 10-10-2006, 01:17 AM

You make a good point about human bias; however, a computer poll can have human bias as well depending on who is feeding the computer the data. If I were a betting man, I'd guess that those who run the BCS poll will manipulate data to their advantage to keep a team like Boise State out of it's top 12.

fusnik11 · 10-10-2006, 02:59 PM

Hey Jay when did computers start thinking for themselves?

How does your computer poll know that the SOS of West Virginia is less than that of Boise State?

Jeff Lebowski · 10-10-2006, 03:06 PM

Quote:

Originally Posted by jay santos

Boise State is undefeated with a better SOS than #4 West Virginia, and similar SOS to one loss teams Iowa, Nebraska, Texas all ranked higher in the human polls.

Sorry, but Texas would kill BSU.

myboynoah · 10-10-2006, 03:16 PM

Quote:

Originally Posted by Jeff Lebowski

Sorry, but Texas would kill BSU.

Perhaps, but I'd sure like to see that game, but we never will unless Texas sucks it up and loses 5 games. Then the game doesn't matter.

I felt for Ute Fans two years ago when they got served up Pitt. It would have been nice to see what that team could really have done. Much like good BYU teams of the past.

Polls, human and computer alike, provide nice conversation pieces, but what we really need is a playoff.

10-10-2006, 01:17 AM	#7
il Padrino Ute Board Pinhead Join Date: Jan 2006 Location: In the basement of my house, Murray, Utah. Posts: 15,941	You make a good point about human bias; however, a computer poll can have human bias as well depending on who is feeding the computer the data. If I were a betting man, I'd guess that those who run the BCS poll will manipulate data to their advantage to keep a team like Boise State out of it's top 12. __________________ "The beauty of baseball is not having to explain it." - Chuck Shriver "This is now the joke that stupid people laugh at." - Christopher Hitchens on IQ jokes about GWB.

10-09-2006, 06:49 PM	#1
jay santos Senior Member Join Date: Jan 2006 Posts: 6,177	Human polls should be eliminated Computer polls blow away human polls in accuracy--based on looking at game outcomes. Human polls will always have bias, and the little guys will always be screwed. Looking at the top 38, counting the "others receiving votes", two non-BCS schools make it. Boise State at #19, BYU at #35 In my computer poll you have: 7 Boise State 23 Tulsa 29 Navy 31 BYU 32 Southern Miss 39 Utah barely misses it Boise State is undefeated with a better SOS than #4 West Virginia, and similar SOS to one loss teams Iowa, Nebraska, Texas all ranked higher in the human polls. That's called getting screwed by the system. Sports media guys would say the computer has something wrong in its formula. Power_rating=If(conf="bcs",true_power_rating+10,tr ue_power_rating) That's the missing formula the computers are screwing up.

10-09-2006, 06:54 PM	#2
TheSizzle36 Senior Member Join Date: Jan 2006 Posts: 3,786	At the same time, humans are the only ones who can actually watch the games and analyze what is happening. I think a better solution would be to release 1 pre-season poll, more of a rankings just for kicks and giggles for the fans, and then after 5 or 6 weeks, or say, October 1st, the first polls are released. Pre-season bias has hurt WAY too many teams in the past. While the human element can hurt the little guys like Boise, and BYU, the human element also has the ability to look at external factors that may not show up in a computer poll like rivalries, injuries, players sitting out due to suspensions, etc. Computer polls should definitely play a part, but they should not be the end all.

10-09-2006, 11:17 PM	#6
Detroitdad Resident Jackass Join Date: Apr 2006 Location: Roswell, New Mexico Posts: 1,846	I share your distaste for the human polls. The logical inconsistencies of voters are obvious. The potential for bias is huge, and the laziness in their compilation is stunning. However, you and I both know that the human element will never be relinquished because they will deny that their biases exist. They never let the nerds construct the rules, whenever there is a populist streak in an item. That is why nerds gravitate toward engineering and law.

10-10-2006, 02:59 PM	#8
fusnik11 Senior Member Join Date: Aug 2005 Posts: 2,506	Hey Jay when did computers start thinking for themselves? How does your computer poll know that the SOS of West Virginia is less than that of Boise State?