I'll repeat my response from that thread: it's not so much about in-game computations (as many posters already observed, coaches and staff already to a lot of the statistical calculations the author describes, it's just all written on paper and wristbands), it's about gambling, specifically making sure that all viewers get the same information at the same time so that particular gamblers aren't advantaged/disadvantaged. Otherwise you have situations where players may be inadvertently tweeting/messaging injury information or game strategies. That's all.
In addition, its also possible that the NFL does not want the best possible and perfect game. The unpredictability and human element makes for more exciting games. That's one of the reasons why they too so much time to make video replays standard in games (and mandatory in the last 2 mins).
Much as the NFL love their dollars, I have to give them kudos for keeping the actual area of the game free of advertising. Here in Australia, in our local version of football, we will slap advertising on absolutely everything - the grass, the goalposts, the umpires (and players, of course)... even the ball sports McDonald's golden arches!
It also is a modest barrier to cheating. Tech that is networked can have stuff appear on it that recently was seen somewhere else, not so with printouts. The vulnerability being that someone on the opposing side looks at the plays and sends copies over to the other side.
I think many coaches already do this, however their predictive results are written on paper and carried around in their binders by assistants. Every coach has a list of scenarios when going for a 2pt convert is advisable, when going for 4th down is advisable, as well as dividing plays up according to yardage & scenario.
Additionally, the amount of film that is analyzed prior to each game, as well as the amount of free-agency these days means players are so familiar with one another and opposing team's play books, that the technology already exists - in their brains.
Maybe it's the football purist in me, but I really don't think iPads would add much. In the NHL, the Maple Leafs have an iPad or two on the bench during the game (the NHL has no restrictions) and they stink.
Yes, this is accurate. The coaches and stats guys do this before the games and they all know what they want to run in given situations and what they believe the other team will do. It's just written down on paper.
Trust me, NFL teams have stats and programmers on board.
Yes, they have stats and applications to apply some of the variables prior to the game. Even at the Division 3 level we had binders with probabilities based on formations. This was a fun brainstorm to see how far the technology could be applied.
You would think they would do this, but they certainly don't do a good job of it considering almost no coach in the NFL gets it right consistently outside of Belichick. Presumably most of them are still using old 4th down and 2 point conversion charts from decades ago, either that or their risk-aversion is just so high that they're afraid to make the obviously statistically correct call.
That's one of the surprises in NFL. If you read Gregg Easterbrook (Tuesday morning quarterback on espn), he frequently points out stastically obvious plays that coaches should have called, but they dont! It is hard to believe that NFL head coaches with their large entourage of assistant coaches do not have the same information that a writer on espn has!
The NFL actually boils down to players making plays. There is no rocket science involved (although coaches are regularly hailed as "geniuses"). If the playbook is made too complicated, players will have a hard time learning it. The trick (in my opinion) is to have your players accurately anticipate the opposition's moves. Belichick's playbook is not the most complex, but it definitely geared towards exploiting his opposition's weaknesses, and simple enough to be understood/implemented by the players on field.
It is hard to believe that NFL head coaches with their large entourage of assistant coaches do not have the same information that a writer on espn has!
It's not that the coaches don't have the info, it's that no play is 100% effective and coaches have to shoulder the blame if thet make a perceived "risky" call (even if it's the statisyicly best play) and it fails.
Remember the decision to go for it when the Pats were facing fourth and 2 on the Colt's 28 yard line in '09? It was statistically the best option, but since they didn't convert, Belichick's been hammered for that decision ever since. If a coach on the bubble makes that call and the team doesn't convert, he's probably gone the next year. As TMQ points out, it's much safer for the coach to punt in that situation, and if the team looses it will be the players who get the blame.
Yep I remember the 4th and 2 against Colts. However, that decision was more of a hit-or-miss.
Think of the countless times a team has the ball with 2-3 mins to go. Three incomplete passes and a punt. The other team scores in the last 2 mins and wins. Except, that it was 3 rushes, it would have taken enough time off the clock to keep the ball (if they get the first down) and keep the other team from getting the ball.
I've noticed a few teams in the NHL use iPads on the bench-- besides tracking ice time, and perhaps monitoring heart rate (I think the Sabres are the only ones who monitor this during games so far) what are some of the more obvious benefits of having a device at hand that isn't already accomplished from the pressbox and relayed down to the Assistant coaches?
They're recording the game and are able to tap into that recording. This way, they can show players replays of certain plays. Also, I believe they have quick access to stats like faceoffs won by certain players.
Whether or not this is true in every detail, I think the point that it is essentially true is undeniable. And there isn't anything objectively wrong with it, it's just not what we want to watch. Heck, I'm a computer programmer and it's not what I want to watch. If I want to see algorithms running against each other I've got all kinds of other places I can do that.
Where exactly can you watch (as a sport) two algorithms running against each other?
Wouldn't that be an awesome new sport? team algorithms fighting against each other. Maybe one was written by some random genius team of programmers to take on the big dog tech companies.
Or maybe a professional robot fight league within google/facebook etc.
WRT Core War: "The earliest published description of Redcode is found in the Core War Guidelines published in March 1984 by A. K. Dewdney and D. G. Jones."
That could be cool. I suggest watching a sport like F1 where the human element has been reduced significantly in favor of technology. I mean, they have cars with automatic gear boxes now.
As for football, I see it as coaches playing chess with pieces that work only some of the time. Without the player you loose that element of chance that leads to incredible 90 yard one had catches while failing backwards into the end zone and dragging both feet ;p
They're pretty interesting because there's an APM (actions per minute) cap imposed by the game itself. IIRC, once you approach 2,000 actions per minute, the game stops responding to commands properly.
You only have a small (for a piece of software!) amount of actions you can do for minute, so you have to choose wisely.
If you watch enough football, you get to see that playcalling isn't everything. You can call a winning game and the receivers will still drop passes or the linebackers will still make poor tackles or the quarterback will still overthrow the receiver.
Play selection is part of the game, and picking the right play under stress is a fundamental part of the game. Palming this off to machine computation is taking a fundamental part of the game and tossing it out. If you throw in runtime updates, it becomes even less about the guys on the field on the day, and even more about which franchise can afford the best statisticians.
I'm against just plugging tech into games simply because it's there, particularly if that tech is not available to all comers.
Instead of getting too far ahead of ourselves (battling AIs, hightech helmet visors, etc.) maybe it's a more mundane reason, they are afraid of having their playbooks hacked. A physical paper binder goes missing and people notice. Someone gains a backdoor into your iPad and it might not ever be reported. Or as Anechoic describes it, they don't want information leaked before it goes on air.
Humans still play football. They are free to change the play during the snap (which is what audibles are for) and even invent new ones on the field. I've never had a chess piece or even a Starcraft trooper decide to just get up and walk in a different direction. It took a roomful of supercomputer to defeat 1 human in a chessmatch. Imagine what would be required to anticipate the moves of 22 of them.
What a load of baloney. This is symptomatic of the believe that management adds overriding value to performance, whereas in practice the best thing they can do is bring the best and/or most cohesive individuals together.
An interesting factoid for football is that in nearly fifty years of the NFL a coach has never led a two different teams to a Superbowl win. So why overestimate the value of what can be done on the sidelines?
I think the use of technology in football (and sports in general) to aid play calling would only enhance the sport. For one thing, it's not as much of an advantage as the author thinks.
For example, in the NFL, there's already a wealth of ongoing research attempting to predict play calling using various methods. I just recently read one, published by Journal of Quantitative Analysis in Sports (see here: http://econpapers.repec.org/article/bpjjqsprt/v_3a7_3ay_3a20...) that uses a linear discriminant function with cross validation -- bottom line, the success rate was only 40.38% (you can read the abstract).
It's an incredibly challenging and fascinating line of research that's grown a lot in recent years. MIT Sloan Sports Analytics Conference is one notable forum where researchers have given talks on everything from speech analysis to predict quarterback performance, to network analysis explaining Bill Simmons' "Ewing Theory" in basketball, to motion tracking for collecting richer data on player performance.
But the other reason I don't think this would have such a "game changing" impact on the NFL is that there are many more elements to leading a team than employing the strategy with the highest expected "point value" in any circumstance. Often, coaches will make the less statistically effective decision purely based on how it will affect their players' psychology. For example, by now most coaches know it's often beneficial to go for it on fourth down -- yet it's still rarely done because in many cases, going for it on fourth down represents a lack of faith in your defense. I can think of more extreme cases but the fourth down issue probably plays out the most in the NFL. It may sound silly but as a coach, you're in charge of rallying your players, who more often than not are less persuaded by artificial neural networks than they are bravado.
Practically speaking, it's just not possible to lead a team by calculator, and that's why I think using tools to improve the "analytical side" of coaching will enhance the sport more so than change it into something completely different.
I think the point of sport isn't to be perfect, but to be visual, visceral, and exciting. Watching actual dudes running along a pitch with a ball is far more awesome than having a bunch of calculations done and observing the outcome.
It's the imperfection that is what is interesting to our brains, we like the analogue, the random, the warm, the fuzzy, the imperfect. Machines are bland in comparison to the rich inconsistency of our organic selves.
"The NFL has banned the use of certain technologies to be used prior the game. Specifically, iPads cannot be used two hours prior to a regular season game."
Cannot be used by who? The coaching staff? The crowd in the stands? The journalists covering the game?
It's to lessen the potential impact of a breach. a) you have a chance to recover from a hack in that two hours, b) it can reduce the speed with which leaked data gets to the opposition, and c) it reduces the likelihood of a hack getting valuable data.
And these are just a few reasons I can think of. There's nothing but benefits to this policy and very small negatives.
It's analogous to filling the moat before a battle. It's just a defense intended to slow an enemy. Try to remember that these guys are some seriously strategic thinkers, they're not morons and they're not technophobes.
If there were a computer algorithm proven to give even a slight advantage in play calling, I'm sure most NFL coaches would jump at the chance to use it. Using game theory and advanced statistics is already all the rage on most (American) football websites. While I'm sure most coaches don't follow such things, enough do that it would soon catch on. And if something successful catches on, NFL coaches have a proven track record of copying it.
Coaches, players and the like are now no longer needed. Okay, this last portion may be a bit of an exaggeration, but the point is proven, technology would, as in other industries, render the current status quo inept and inefficient. In conclusion, I plan to take the Robotics course from Udacity this month.
As an avid follower of this sport, I've found that a large part of success as a team is to be found in coaching the fundamentals. Just like in school, a good teacher can make all the difference in terms of outcome, even for a gifted player/student.
For example, there was a famous study a few years ago showing that NFL teams punt way too often. (http://fifthdown.blogs.nytimes.com/2009/09/17/a-new-study-on...). It didn't seem to have much effect on play-calling. Being conventionally wrong appears to be safer for coaches than being unconventionally correct.
Of course there are exceptions. To their credit, the Patriots let the Giants have a free touchdown near the end of the Super Bowl, rather than letting them run the clock out. An obvious tactic to most Madden players, but not one you see much in the pros, because That's Not The Way It's Done. And Belichick is one of the few coaches with the job security to do things his own way.
I am a football coach, and while this article seems logical. I believe there are too many variables, like stock trading.
And even if you could write a program to identify all useful variables, coaches will often evaluate themselves for tendencies, and call plays/defenses to go against their tendencies.
I have used breakdown software and it is good, but the sample size it often too small. Two or three games is not nearly enough plays/situations to analyze it using algorithms.
So, some might say, by the time the post-season comes along, the sample size would be big enough, but that is also not the case. Only the last few games are relevant when breaking down games. By the end of the season, early season games are useless.
Also, the team you are playing might not have played a team with a similar sceme to yours, so you have zero sample plays that are useful to you.
Cheating. New technology = new forms of cheating; and if the league allowed everything on the sidelines, it'd be near impossible for them to keep up with.
Predicting plays isn't an issue. Predict all you like. Stealing plays (& the play calls), that's the real worry, as it probably matters most in this sport than any other. Stealing signs in baseball is the only comparison I can think of.
The WSJ article linked by breakall mentions it a little, but nowhere do I see the word 'cheat'. I doubt the league would address a question about it seriously anyway as that would imply they don't trust teams' staff (which they shouldn't).
Mostly-agree. I think you're right that cheating, above everything else, is why the NFL is slow to adapt technology. But I do have to point out it's happening with the technology they're using already: http://sports.espn.go.com/nfl/news/story?id=3014677
Newer tech would make that easier, but it's going to happen one way or another.
"Cheating. New technology = new forms of cheating;"
100% correct on the cheating part, though the reason iPads are banned currently isn't even because of "new forms of cheating" but because they have cameras on them, allowing for a very old form of cheating in football.
I've always wondered what would happen in an extreme situation like the author describes. I can't imagine the type of game that would lead to. I figure at the end of the day it would come down to the human aspect of it: can the team execute their high-probability play vs. the expected high-probability defense?
I wonder what kind of race this would lead to in F1? If drivers were replaced by an apparatus that would drive the car and maximum driver assists and all telemetry measurements were allowed to be used would the races be boring or even more exciting?
Robot F1 could be very exciting. Faster races, lighter cars (probably could drop like <400kg if they opened up the rules), less need for safety regulations, etc.
Can someone help me out with the introduction to the argument?
"The theory is based on protective reasoning and that the NFL
wants to ensure the monopoly is held as a status quo."
What is the monopoly that is being held at the status quo?
Anytime we find that the technology and innovation create
a competitive advantage or an advantage that cannot
readily be seized by those in power, we see resistance.
Who could seize this advantage that is not already in power?
I have a hard time believing they don't already do this. Having computers/iPads banned on the field doesn't avoid outsiders using this and communicating with the coaching staff.
What they should really be afraid of is simulations of football becoming more interesting than recordings of the sport actually being played.
People already spend hours and hours playing NFL games... it's only a matter of time that a virtual league with probabilistic players and teams coupled with 'better than real' graphics entices more viewers than the actual sport. Imagine being able to see a great play from any angle, at any time -- the current NFL can't provide that.
I can't believe that some teams haven't already started doing this: hire a data scientist or two and turn them loose with statistics, film, and some assistant coach to help them make sense of the game and its nuances. Apply some machine learning and you'd have a really serious tool to apply in predicting your opponent's play calling.
There's no way to avoid technology, it will catch up to them whether they want it or not. This is the same mentality that Hollywood has up until now and is now catching up to them - as we all know too well.
There is a particular camera feed (called All-22) that monitors all 22 players at the same time. This particular camera feed is only available to teams, not outside data-collection agencies.
Meh, aren't them in communication all the time with people who watches the game 100km from there? Couldn't they print the papers generated all the weeks before?
Nice conspiracy theory, but it ignores one extremely important point: Football is a sport, not a technological arms race. They won't allow iPads and LCD visors for the same reason they won't allow jetpacks and rocket shoes: it wouldn't be a sport anymore.
Not to mention the fact that iPads don't have the power to analyze such large volumes of information in a useful time frame anyway. Getting all that info through an iPad would require communication with some outside computing power, which they could accomplish with their headsets already if it were allowed.
I don't understand why athletic competition precludes technical competition?
As far as the ipad computational power goes I think its safe to assume that when the author says use ipads he assumes they are a frontend to a big processing backend.
I'll repeat my response from that thread: it's not so much about in-game computations (as many posters already observed, coaches and staff already to a lot of the statistical calculations the author describes, it's just all written on paper and wristbands), it's about gambling, specifically making sure that all viewers get the same information at the same time so that particular gamblers aren't advantaged/disadvantaged. Otherwise you have situations where players may be inadvertently tweeting/messaging injury information or game strategies. That's all.