Why Driverless Cars Won’t Be Better Than Humans: A Sports Analogy
If you read my blog on the regular, then you already know I’m gaming you with the headline. But, whether you’re a newcomer or a frequent reader, rejoice in this rare respite: for today, we’re not talking urban planning or transportation policy or corporate chicanery.
We’re talking basketball. We love that basketball.
The average NBA team makes somewhere between 45-50% of the shots they take. Let’s err on the low end after factoring in turnovers, and say that for any given play, there’s about a 45% chance a team will succeed in achieving their goal: to score points.
Now, because sports are arguably the pinnacle of existential diversion, there are endless analyses of shooting percentages throughout a season and comparative data points going back for decades. Yet, even amidst this incredibly rigorous oversight, no one sees a problem with shooting 45%, even though it means a team fails 55% of the time. It’s the norm. It’s an assumption about how the game works, and we’ve built an entire sport around it.
A team has to deviate from that 45% rate by about a third before coaches start rethinking their strategies and opportunities during a game. Deviating from the 45% clip by half or more during a game (shooting 68% or 23%) is almost unheard of — and again, we’re talking about a single game; a moment in time.
Across the more consistent timeframe of a full NBA season, a team’s ability to perform the game’s primary task — making a shot — falls between 36% and 55%. Pick any team in history across 1,500+ team seasons; they’re all inside this performance window. These are the boundaries of reality in the NBA.
So, what would happen if a player showed up on the scene with the ability to make 100% of his shots, no matter where he stood on the court? A player who literally could not miss? Well, he’s here. Let’s call him Otto. Otto Mattick.
To make it interesting, let’s say Otto has absolutely no talent for the other facets of basketball. He doesn’t know how to dribble or pass, he’s baffled by the idea of running a play, and his defense is about equal to a sack of potatoes. The only thing he can do is catch a ball and shoot it.
Since Otto makes 100% of the shots he takes, he drastically changes how the game is played. For one thing, there’s no point in even going down the court anymore, because getting closer to the basket is just a strategy for imperfect shooters. Same goes for running complex plays to get the best shot. Otto doesn’t care if the shot’s difficult — he’ll drain it every time.
With that in mind, let’s compare how the game changes from the perspective of a shot chart. Here’s a typical game chart for an NBA team: they shoot 47% (38 for 81) over the course of the game, for a total of 84 points from field goals. Note how they shoot better when they’re closer to the basket, which is part of the whole battle between offense and defense. Note also that they have a few missed shots from the backcourt area, which are usually Hail Marys heaved up in the final seconds of a quarter. Again, standard shot chart.
Now, here’s the shot chart for a game Otto played in. 100% shooting (50 for 50), for a total of 150 points — they’re all three-pointers after all, because Otto simply caught the inbound pass and immediately threw up his shot from the backcourt. Looks like he ran around a bit to get open for the passes, but that’s all the drama there was.
Otto took 31 fewer shots, yet generated 66 more points than the entire aforementioned team. It’s efficiency the likes of which the NBA has never dreamed of, let alone seen.
That all sounds well and good, except for the fact that no one would want to watch this game. Maybe the first five minutes or so, sure… but once it becomes obvious that Otto will never miss, what’s the point?
What’s the point. Remember that for later.
The disruption Otto brings to the NBA would obviously breed some unusual reactions at first, as the league tries to grapple with this new reality. Maybe opposing teams start recruiting huge football players to keep Otto from getting open for the inbound pass; in turn, his own team might recruit similar giants to keep the opposing team at bay. Sounds like what we have now is more like football, with Otto as the quarterback and the basket as the receiver.
Maybe the league even makes new rules to return the game to a watchable state, like requiring that each player get an equal number of shots, or forcing each quarter’s leading scorer to switch teams, or whatever. There’d be a lot of panic about how to solve for Otto.
But all of this assumes there is only one Otto.
Let’s say instead that Otto knew the trick to making every shot, and started teaching it to everyone. Eventually, every team could be filled with Ottos, and no one would ever miss a shot — rendering the game entirely, and finally, obsolete. No more professional basketball. Otto ruined it because he was unwaveringly successful in a game predicated on the necessity of failure.
WHAT THE HELL DOES THIS HAVE TO DO WITH DRIVERLESS CARS BEING BETTER THAN HUMANS
Right, here we go.
Is Otto the best basketball player in the world?
Pit him against LeBron James in a 1-on-1 game, and LeBron would demolish him. Remember, Otto can’t dribble, so LeBron knows what’s coming every time. He’d just reach up to block Otto’s shot, grab the ball, and drive to the hoop virtually uncontested (since Otto can’t play defense either.)
So LeBron would probably shut Otto out in a 1-on-1, but even if the game panned out differently, you’d be hard-pressed to say Otto was a better ball player, right? LeBron is dynamic, powerful, athletic… his basketball intelligence is commendable, and he’s mastered so many components of the game. Otto is a parlor trick by comparison.
But here’s the rub: professional basketball isn’t an individual sport. You have to consider how all the players on the court interact.
The best player is the one who brings the greatest benefit to a team, and that player is Otto by a long shot. You know this because a league full of LeBrons wouldn’t change the NBA, whereas a league full of Ottos changes everything.
All those components of the game LeBron excels at — dribbling, passing, dunking, rebounding, defending, strategizing — they only exist because making a shot is difficult. When shooting no longer involves failure, none of those other skills matter. And that, friends, is how traffic works.
Autonomous vehicles have the benefit of being unwaveringly vigilant in all directions of traffic, far and wide — along with being connected to other vehicles and infrastructure they can’t even see. Their parlor trick is that they know what every other vehicle on the road is going to do before it happens, and because of that, they render every other component of driving obsolete.
Virtually everything that makes someone good at driving in traffic is based on reaction or anticipation of bad driving, the congestion and accidents caused by bad driving, and the unforeseen situations that lead others to drive badly. The failures. Being the LeBron of driving means navigating the failure of others. A driverless vehicle isn’t better, and it doesn’t need to be, because it simply removes the chance of failure. It’s Otto.
Again though — put one Otto into a world of imperfect drivers, and a lot of confusion ensues, because he’s screwing up the familiar world full of failure in which we’ve all learned to interact. And here is where depart from the basketball analogy, because of something you told yourself further back in this article:
What’s the point?
Basketball is a virtual reality we created to entertain ourselves. We wanted to insert excitement, drama, joy, struggle, friends and enemies into our easy and dull lives. The point of professional basketball is to exist for our amusement. We wouldn’t be amused by an army of Ottos, which is why they would never play in the NBA.
But traffic… what’s the point of that? Congestion, pollution, land occupation, wasted time, lost wealth, injuries and deaths abound in traffic. All of it is caused by our failures as drivers, and all of it we would happily do without. There is no point to traffic — we tolerate it because we believe it’s a necessary consequence of transportation. And, like LeBron, everything that makes you a good driver is founded on the notion that traffic involves failure.
Yet here we are on the brink of inventing machines which virtually can’t fail at driving, and the only reason they would is because of our own refusal to get out from behind the wheel and stop causing failures that don’t need to exist.
Being “better” than a self-driving car means nothing in traffic, because you have to consider how all the drivers on the road interact. The best driver is the one who brings the greatest benefit to a traffic pattern, and that driver is the robot, by a long shot.
The crux of this analogy — why being better is a red herring — is that many of the skills we possess as humans (whether it’s in basketball or driving) have been developed as workarounds for inherent failures in the task. So, we see being better as being more highly skilled in working through failure. But, when we eliminate failure as a potential outcome, then the set of skills which made someone better than another are obsolete, even though the person still possesses those skills.
That’s the argument as to why it’s unnecessary that driverless cars develop the skills to out-drive humans in traffic. Remove the failures — human error — and the skills to work around them are no longer needed.
