Imagine that you are a play caller for a football team that is down by 30 points with, say, 8 minutes to go in the 4th quarter. What's your strategy?

Aggression. That's your strategy. Lot's, and lot's, of aggression.

Let's take a step back. What does a normal strategy in football look like, as far as play calling goes? On offense, it's a mix of passive and aggressive calls. You'll keep the defense honest with a few running plays. A few short passes. A few screens. These are all safe, passive play calls. They're likely to yield you a few yards, but unlikely to yield you a lot of yards.

This is a problem however when you're down by 30 with 8 minutes to go: if you pursue this strategy, even if you execute it successfully, you will lose. You will make slow but steady progress chugging your way down the field. It's probably enough time to score one touchdown. Two if you're very lucky. But when you're down by 30, you need four touchdowns. So no, this strategy will not cut it.

Ok, so what's the alternative? Simple: deep passes. You need lots of yards, and you need them fast. Throwing the ball far down the field to a receiver is the way to obtain that.

Note that this is scary, and perhaps counterintuitive. Such a deep pass probably won't be completed. It feels futile. Maybe there's a 95% chance that it is incomplete. Maybe it is futile.

On the other hand, the defense is probably willing to concede a nice, easy, 10 yard completion to you. Such completions are usually hard to come by. It feels very nice to make such completions. It feels like you are being productive. It feels like solid progress is being made.

Sadly, it is not. In the unfortunate scenario you find yourself in, there simply is not nearly enough time left in the game for such a strategy to work.

Oh, in addition to completing deep passes, you also probably want to conserve the clock. That would be helpful. Try to get it to your receivers in such a way where they could get out of bounds before being tackled. When that happens, the clock stops. When they are instead tackled in bounds, the clock keeps running, draining a very precious resource from you.

You might also want to try a few trick plays. Trick plays are an even more extreme version of the deep pass to some extent: they're even less likely to work, but when they do, they can yield a lot of yards.

One downside to trick plays is that you expose your hand, so to speak. They work because of some unconventional, unexpected wrinkle to the play. But after the opponent sees it once, it's not gonna work again. And all the other teams in the league are gonna see it too. But when you need to win, that is an acceptable price to pay.

Calling plays is only one part of what a coach does to help their team win the game though. Motivating players is another. Eg. with speeches.

Such speeches are a scarce resource, I believe. Here's what I mean. Consider a coach that gives the "It's now or never! Lay it all out on the line!" speech after every quarter, throughout the season. It gets old after a while. It feels like the boy who cried wolf. But when the calm and collected coach who never cries wolf suddenly gets amped up, it really means something, y'know?


Despite what they might claim, coaches rarely play to win the game. At least to my eye. They don't turn the aggression dial up as much as I'd expect if they were actually trying to win.

Perhaps incentives can explain this. When you just flail a handful of deep passes down the field that futilely fall incomplete and the team ends up losing by 30, you don't look very good. On the other hand, if you chug down the field and score two touchdowns, losing by 14 let's say, you don't look quite as bad. It wasn't as big a loss. Your troops pushed forward and made visible progress towards victory. The fans were entertained. The players padded their stats. And by an unspoken agreement, they didn't put each others bodies at as much risk either.

Perhaps there is wisdom in this.

Or, perhaps not. Perhaps one should, in fact, play to win the game. Is that not the goal?

Depends on the game. Sometimes, no, it isn't the goal. Sometimes it is just a small part of a larger whole. The actual goal is to win the Superbowl, and it might make sense to take your foot off the gas during a Thursday night game in the middle of the regular season where you're down by 30. Live to fight another day.

You can even argue that the Superbowl itself is just a part of a bigger whole. Suppose you have a young superstar quarterback like Patrick Mahomes and he has a bum knee in the Superbowl. You might not want to put him back in the game. He might further injure his knee, jeopardizing the bright future you guys have of winning multiple Superbowls over time.

And even that isn't where the rabbit hole ends. You could argue that this all is ultimately just a game, and that the real goal is, uh, being satisfied with life or something like that. I'm actually one to cheer "Sports Go Sports!". I believe that they can be a major component to such bigger picture life satisfaction and shouldn't be dismissed as childish or whatever, but sure, the Superbowls aren't the actual end goal.

At some point the rabbit hole bottoms out though, and you find yourself facing the actual end goal. This isn't a place where humans spend much time hanging out. We usually hang around the surface. Sometimes we venture a little further down, climbing down the ladder a few rungs, but with a rope tied to our waist and a trusted friend at the top holding it in case we fall. But sometimes we just get knocked down with no safety harness and awkwardly trip all over ourselves, stumbling until we fall all the way to the bottom. It is at this point that we must play to win.

New to LessWrong?

New Comment
5 comments, sorted by Click to highlight new comments since: Today at 5:40 AM

This is pretty well-known in gaming circles - when you're behind in a quantized-outcome game (where place matters, not points), increase the variance.  Take risks, be aggressive.  It doesn't matter if you lose by 1 or 10000 points - take any chance, however small, to pull out the win.

In reality, no game is ever one-shot.  Even for professional sports coaches, we care about a lot of factors beyond "who won".  In board or card games, you want to make sure your friends want to play future games with you, and that everyone has a good time.  In sports, they want the fans to be entertained enough to buy future tickets, and the coach wants to look competent enough to keep their job.  

It's VERY common for a team to go from down 30 to down 20, and the post-game analysis praises the choice of plays and how well the losers did in an impossible situation.  That's a big factor even when the coach doesn't know the betting spread and whether their knees will get broken if they make/blow it.  

I don't think that for most humans there IS an actual end goal.  Even on one's deathbed, one isn't talling up scores or making evaluations.  Some of us do have preferences for things long after our deaths, but even then they're directional (more happy sentient beings rather than fewer) rather than win-conditions.

Down by 30 I probably put in backups to make sure my best players don't get injured, and play to run out the clock. Winning would require a record-breaking comeback, and even if we go all-out to win the chances of pulling it off are tiny, maybe 1 in a million.

Though I guess if it's the playoffs then I keep playing for the win. Regular season it would be worth playing for the win if we're down by 20 instead of 30.

Generally teams do adjust their tactics in the right direction in these sorts of situations, but not by enough on average. NFL teams play faster when trailing by multiple scores, but they usually don't shift to the full-fledged hurry-up offense that they use when time is running low at the end of the half. Teams go for it more on 4th down, but there are still plenty of cowardly punts. Quarterbacks throw more interceptions due to trying to make more risky passes down the field into tight coverage - I guess I'm less sure about how to tell whether they do too little of that. We don't see many trick plays in these situations - something like a flea flicker might not be that effective when the defense is playing it safe, but I expect that using more laterals (like a hook & ladder play) would be worth the risk (though I'm not sure about the cost-benefit of putting those on tape vs. saving them for higher leverage situations).

Yup, I agree with all of that. Towards the end of the post I was trying to make that point about how most of the time, like in the regular season, you shouldn't play to win because there are those other considerations. And that really, it is quite rare that you actually need to play to win. Usually there are other considerations. (Which is why that clip of Herm Edwards is ignorant and funny.)

I wasn't planning on saying it explicitly, but I had AI alignment and Eliezer's recent post about Death With Dignity in mind as an exception. When the world is ending in ~10 years, that strikes me as a time when you actually do need to play to win. Other considerations pale in comparison to that. (And you need to throw some deep passes since you're down by 30.)

Also, funnily enough I did have the hook & ladder in mind when I mentioned trick plays. I know as much about X's and O's as the average fan though, so I was also thinking that a talented play caller might be able to come up with better ideas.

For non-Americans: full-court press and half-court 3-point attempts, shots on goal from far outside the penalty area, attacking despite being a player down due to a penalty... All variants of the Hail Mary play. 

I agree that it's ostensibly underused. The hard part is identifying what and when to play for how much risk. The same applies in spades to the AI x-risk.

For non-Americans: full-court press and half-court 3-point attempts, shots on goal from far outside the penalty area, attacking despite being a player down due to a penalty... All variants of the Hail Mary play.

Yeah, sorry about the America-centric assumptions here. To clarify a little bit, a Hail Mary is just the most extreme form of the more general aggressive strategy I'm trying to describe. Ie. the general strategy is one where the chances of an attempt working are low, but the benefits if it does work are high. With a Hail Mary in American football, the "probability of success" dial is turned all the way down, and the "benefits of success" dial is turned all the way up, basically. But I'm not advocating for that specifically here. I'm just advocating for the more general class of strategies that are high-risk, high-reward, when you happen to find yourself in a situation that justifies it.