In 2011, Jason Garrett was the 45 year old first time head coach of the Dallas Cowboys. Garrett’s resume prior to becoming Dallas’ head coach was as follows:
· Quarterback at Princeton
· Back-up NFL Quarterback with 9 career starts
· Quarterbacks coach for the Miami Dolphins (2 years)
· Offensive coordinator for the Dallas Cowboys (4 years)
As offensive coordinator Garrett directed an offense that was generally packed with high profile names (Terrell Owens for instance) although critics would say that the results were uneven. The Cowboys were top five in the league in points scored twice during Garrett’s tenure and were closer to the middle of the pack the other years.
Garrett’s experience prior to becoming head coach can probably be summed up as 64 games coached as an offensive coordinator, 32 games as a quarterbacks coach, and over 100 games as a back-up quarterback. Garrett’s practice as a decision maker is probably about 64 games, while the rest of his games largely fall in the spectator category as back-up quarterbacks are not responsible for making any decisions during the course of the game.
Garrett took significant criticism during the 2011 season for one end of game situation that might have cost the Cowboys a win, in a year in which they missed the playoffs by just one game. In their game against the Arizona Cardinals, the Cowboys had a chance to win the game in regulation by kicking a field goal to break a 13-13 tie. That opportunity was set up by the following events:
Dallas got the ball for a potentially game winning drive at their 32 yard line with 2:54 left on the game clock. They needed only a field goal to win. In most situations that would be so much time that teams would be afraid of scoring too quickly and giving the ball back to their opponent. The result of their first two plays from scrimmage was 12 total yards of field position, but almost a full minute of game clock. Advanced NFL Stats (ANS) estimates the probability of each team winning a game following each play. According to ANS, Dallas started the drive with about a 70% chance of winning the game. After letting the minute of clock run, they had only a 60% chance of winning. Despite gaining field position, they had somehow become less likely to win. One of the plays that the Cowboys had run was a pass to tight end Jason Witten that actually lost a yard and ended with Witten being tackled in bounds and the game clock continuing to run until the two minute warning. Quarterback Tony Romo would have been better off throwing the ball into the ground so that at least the clock would stop. The yard that Witten lost would be damaging later as the Cowboys’ next play was one yard short of a first down. That meant that the yard that Witten lost required the Cowboys to run a quarterback sneak to pick up the first down. The quarterback sneak cost another 30 seconds of game clock.
At this point the Cowboys had moved the ball just 20 yards in roughly two minutes of game time. However, they had the ball with first down at the Arizona 44 yard line and they had two timeouts. There was still 1:06 of game time, so with the two timeouts and about 15 yards to gain to be in field goal range, the game was still very winnable. But the Cowboys then took a false start penalty which meant that first down with 10 yards to go became first down with 15 yards to go. The next play was an incomplete pass. After the incomplete pass, the Cowboys were flagged for delay of game! They weren’t able to run a play in the time allotted under the play clock. They were staring at 2nd down with 20 yards to gain for a first down. With the ball at their own 45 yard line, they were well outside of field goal range. They had started a drive with almost 3 minutes in game clock and had moved the ball a mere 13 yards of net field position.
The Cowboys next play was a nine yard pass to Dez Bryant, who was tackled in bounds allowing another 30 seconds of game clock to run before the Cowboys next play. Dallas snapped the ball with 31 seconds left on the game clock and completed another pass to Bryant at the Arizona 31 yard line. They were finally in field goal range. What happened next generated a lot of criticism. Tony Romo rushed the offense up to the line and spiked the ball to stop the clock with 0:08 seconds left on the game clock… even though Dallas still had their two timeouts left.
Dallas didn’t at that point use either of their timeouts to run another play in order to attempt to move the ball closer so that field goal kicker Dan Bailey would have an easier kick. They were apparently content with attempting a 49 yard field goal, even though attempts from that distance miss about 1/3 of the time. Bailey would miss this particular kick. Dallas would go on to lose the game in overtime.
It’s probably better that Dallas’ 49 yard field goal ended up in the “miss” category as we have what some might call a teachable moment. Had Bailey made that field goal, the abomination that was the Cowboys’ execution leading up to that point in the game might have gone overlooked. During those three minutes of game time the Cowboys put on a clinic in letting game clock run while not advancing the ball at all. They didn’t optimize their use of timeouts and in fact it’s probably likely that they didn’t even know they had the timeouts. If they had known, they wouldn’t have had Tony Romo rush the offense to the line to spike the ball.
Head coach Jason Garrett took considerable criticism after the game for the poor end of game management. Garrett refused to admit any mistake. Primarily the media focused on Garrett’s non-use of his timeouts and the fact that the Cowboys’ hadn’t tried to move the ball closer for the field goal. Garrett’s response to the criticism was that the Cowboys felt like they were in field goal range and he didn’t see a reason to risk running a play for a loss with just eight seconds on the game clock. Garrett would say of his decision making:
“We very well could have taken a timeout there. We felt like we were in field-goal range. We have yard lines that we use as guidelines before the game. We felt like we were in range at that point. Tony (Romo) had them on the line of scrimmage quickly, so we went ahead and clocked it and used that as a timeout. …You see so many situations where you have negative plays in those situations. We felt like we were in his range to give him a chance to kick the game-winner.”
Garrett is making the equivalent of a probability based argument. He’s saying that the Cowboys chances of winning were greater just kicking the field goal than they would have been if the Cowboys had tried to run another play. Garrett says you see “so many situations where you have negative plays”. The question is: Is Garrett right?
Are negative plays likely to happen in those late game situations? First of all, Advanced NFL Stats’ Brian Burke pointed out that Garrett’s explanation doesn’t even actually make sense. While holding on to timeouts, Garrett had the Dallas offense rush to the line to spike the ball and stop the clock. When they did that, they were snapping the ball and risking a false start or illegal formation penalty. That’s exactly what Garrett said he was trying to avoid.
But even if you ignore that Garrett’s logic fails on its own terms, it’s possible to look at his decision making and see that he missed an opportunity to maximize the Cowboys’ chances of winning.
First, as Burke pointed out in his analysis of the game, if the Cowboys had simply taken a timeout after Dez Bryant caught the ball at the 31 yard line, they would have had plenty of time left to run another play without having to worry about the clock expiring. The time it took Romo to rush the offense up to the line could have been saved.
Second, Garrett’s worry that another play could result in an offensive penalty should be balanced out by the reality that some penalty could also be called on the defense, which would increase the Cowboys’ chance of winning. Actually, the penalties probably cancel out at that point. We have to remove false start and illegal formation penalties from the things Garrett can be worried about because they weren’t avoiding the risk of a false start or illegal formation penalty by spiking the ball. If you remove those penalties from the equation, the probability of an offensive penalty is about the same as the probability of a defensive penalty. Since those things cancel each other out, all that’s left is that the Cowboys could have run another play and would have had a free shot at advancing the ball. They also would have had another timeout to use in the event that their next play didn’t result in the clock being stopped on its own.
Garrett’s performance at the end of the game against Arizona looks like the performance of someone who isn’t processing information as fast as they need to be. He may have been overwhelmed by the combination of his play calling duties and the need to manage the clock at the end of the game. If you think about it, he had an amazing amount of information to process. First, Garrett must come up with the offense’s play calls. He has to reason through which plays will be successful, weigh the odds that the defense might expect those plays, and then finally come up with what he thinks is the optimal play call. He also has to keep an eye on the game clock. He needs to be sure that he doesn’t score too early. But he also needs to be sure that he does score at least a field goal. He has to remember that he has timeouts. He has to listen to the other coaches talking in his headset. Garrett also has to process probabilities. What are the odds that a field goal does make it from a certain distance? What are the odds that a negative play puts them out of field goal range?
Garrett also had the appearance of someone who was learning, and who in the future might remember that 49 yard field goals aren’t automatic and might try to move closer for a game winning try. Even Garrett’s defense of his poor game management has the ring of an excuse from someone who has a suspicion that he screwed up, but doesn’t want to admit that it was because he didn’t know what to do.
The interesting question in this instance is whether Garrett should be expected to be any better at end of game management than he is. Has he been accumulating the deliberate practice that researchers say is necessary to become an expert? Would Garrett have gotten this experience in end of game situations during his career as a back-up quarterback? Would he have gotten this experience as a quarterbacks coach? Would he have gotten this experience as an offensive coordinator?
For purposes of understanding how much experience Garrett has as a coach, perhaps it would be helpful to compare him to another coach of similar age, but who also has a Super Bowl ring. New Orleans Saints coach Sean Payton is just three years older than Garrett, but has almost 15 more years of coaching experience than Garrett does.
Sean Payton Coaching Experience (Source: Pro-Football-Reference.com)
Jason Garrett Coaching Experience
In 1989 when Jason Garrett was the quarterback of the San Antonio Roughriders, Sean Payton was coaching college football. In 1995 when Sean Payton was the offensive coordinator for the Miami (Ohio) Redhawks, Jason Garrett was backing up Dallas Cowboys Hall of Famer Troy Aikman. The two actually worked for the same team from 2000-2002 when Garrett was a backup quarterback and Payton was the offensive coordinator for the New York Giants. In 2004 when Jason Garrett was a backup quarterback for the Miami Dolphins, Sean Payton was in his 17th year of coaching and was the assistant head coach for the Dallas Cowboys under the legendary Bill Parcells.
In short, while Jason Garrett was engaged in the deliberate practice of being a quarterback, Sean Payton was engaged in the deliberate practice of coaching football. Garrett is now going through the learning process that Payton has a 15 year head start on.
If Jason Garrett’s competition is a league full of coaches who have been coaching since they were in their early 20’s, does Garrett have any hope of obtaining enough experience to overcome that shortfall? Remember that the NFL has just 16 games per year. That’s 16 total chances to get practice at a real live two minute drill. It’s not very much. By comparison, an NBA coach has an 82 game season. That’s a lot more chances to practice varying aspects of game management. It’s realistic to think that an NBA coach could start the season deficient in something like end of game management and by the end of the season would have enough practice to be legitimately better at that aspect of the game. NFL coaches have no such luxury.
We saw the way that young poker pros overcame their experience shortfall by simply playing lots and lots of hands online. But is something like that possible for NFL coaches? It would be tough for NFL coaches to amass that kind of real life experience because their players can’t withstand the stress of practicing live on a regular basis. Because football is a collision sport, the injury potential stands in the way of meaningful practice reps. How does a coach practice what they are doing if they have to be directing players in their practice sessions?
The answer is probably computer simulations. Simulations are used by industries like the airline industry and the defense industry to train people on skills where practice is both extremely important and difficult to come by. Simulation is so important to the defense industry that the U.S. Department of Defense spends $4 billion annually on simulators and equipment. The comparisons between coaching a football game and doing something like flying an airplane or perhaps a drone aircraft are apt. Just as football coaches have no room for error, and have few opportunities to practice what they do, commercial airline pilots have no room for error. Simulators offer practice where a mistake doesn’t mean loss of life, it means starting over and trying again.
Some simulators are extremely expensive, like full cockpit replica simulators. But some simulators are essentially just off the shelf software. If you’re doubtful that a football coach could benefit from a computer simulation, consider that the U.S. Navy found that even just the use of Microsoft’s Flight Simulator substantially improved flight students’ chances of performing above average during real tests (the Navy also found that the use of MS Flight Simulator helped older pilots more than younger pilots).
Simulators are useful because they require the user to think about the problems to be solved. Users have to engage on the hypothesis creation and hypothesis testing activities that result in learning. Simulators allow students to train on precise tasks that are part of the overall skillset. For instance, in order to get 100 iterations of an actual aircraft takeoff you would also have to land the plane 100 times. But in a simulator you could practice a skill like taking off (or landing) that might be deficient. Perhaps a student is particularly good at landing, but has trouble taking off. That student can practice taking off over and over again. The basis for all learning is trying and failing.
Simulators reduce the cost of failure.
The military has also extended the use of simulators to skillsets like tactical strategy. In some cases they’ve developed their own software and in other cases they’ve opted not to reinvent the wheel by simply modifying software that video game makers had already produced. The military’s use of simulation falls in the same category of training as a football coach might need. They’re training skillsets where real live experience is difficult to acquire, yet mistakes on the job are extremely costly. The Army’s students at West Point can use simulators, or glorified video games, for things like practicing war games exercises. War games are similar to a football game in that they test a person’s ability to cope with a number of things happening at once and where speed of execution is critical.
The military’s study of simulation and the ways that video games can be used in simulations led them to an idea that they’ve confirmed with research. Video games improve reasoning ability, or fluid intelligence. Fluid intelligence is our ability to solve new problems.
In a study conducted by the Navy, video game players were found to perform from 10 to 20 percent higher on tests of perceptual and cognitive ability. Dr. Ray Perez was the officer in charge of the study. He says of the Navy’s interest in video games and how they can impact intelligence:
“We have to train people to be quick on their feet – agile problem solvers, agile thinkers – to be able to counteract and develop counter tactics to terrorists on the battlefield…It’s really about human inventiveness and creativeness and being able to match wits with the enemy…being able to work outside your present mindset, to think beyond what you have been taught, to go beyond your experience to solve problems in new and different ways.”
The idea that video games might actually improve fluid intelligence is an extremely important idea for the military because of the critical need to be able to think and reason through situations that are new, or novel. The military could spend significant resources teaching its soldiers to deal with the tactics that the enemy might employ today. But when the enemy figures out that the U.S. has adjusted its tactics, the enemy will also change. The military needs to train soldiers to be able to confront problems that they don’t know exist today.
The problem for the military in training its students is like the old saying that giving a man a fish feeds him for a day, but teaching a man to fish feeds him for a lifetime. By focusing on the problem of increasing fluid intelligence, the military has taken it one step further. They know that what they really need to do is to teach students how to figure out how to fish even if they’ve never been taught to do it.
The U.S. military is teaching both through the use of simulation, in order to provide students with repetition of skills they should have, and through the use of games, which are meant to address the students’ ability to confront problems they’ve never faced before.
The obvious objection for the use of simulation or video games in the NFL is that perhaps the NFL is too complex for the use of simulation. Perhaps it’s the case that simulation couldn’t reproduce an actual game setting and therefore isn’t useful. Except that would be the same objection that one might have for the military’s use of simulation and the military has said that games and simulation are useful. The other thing that’s important to note is that the military was so interested in the question that the conducted a study to make sure that games were useful at increasing reasoning ability.
Since simulations are useful for complex jobs like being an airline pilot, or for military strategists, it makes sense to look at the closest thing that football has to simulation, which are the football video games like the Madden NFL franchise.