Hit enter to search or ESC to close

Who Will Win The Self-Driving Game?

A box for a board game about autonomous vehicles with one in white and another in teal against a yellow backdrop

I’m often asked who is going to win the self-driving car race, so in my last column I explained that self-driving isn’t a race, but a game. Thinking that games are races is what kids do when they don’t understand the game being played, like my divine 3-year old baby angel Coco. The first time she saw the tiny silver car sitting unused in the Monopoly box, she took it, marched over to where the adults were playing, drove the car around the perimeter of the board, raised her arms and yelled “I WIN.”

Racing a toy car around a Monopoly board is a perfect analogy for how some view the self-driving business, like the Tesla fans who say that once Elon has “solved” autonomy, it’s game over. But a game isn’t solved, it’s played, and everyone has a different spin on why they’re “winning.” Some are impressed with driverless rides late at night in a quiet neighborhood, yet others have been doing that during the day, for years.

So, who’s winning? That depends on what the game actually is.

It makes no sense to compare companies based on their confidence, or the picking of low-hanging fruit, or how long they’ve been in the game. None of the promises, milestones, or beliefs teach us anything about what’s actually happening, because they lack context. It’s not clear what game(s) most self-driving technology companies are playing, or think they’re playing. Is there one game that might give us that context, and maybe even predict the industry’s future? If so, what are its rules? What are its victory conditions?

Let’s Define “Self-Driving Industry”

When I say self-driving industry I mean everyone trying to build a business in, on, for, or using vehicles that don’t require a human driver behind the wheel. From sensor and chip makers to ride-hail and delivery platforms, mappers, shippers, distributors, vehicle manufacturers, and of course those developing the AI that will actually do the driving, thousands of companies are competing in the self-driving game.

But they’re not all fighting the same fight, which is why claiming any one company can “win” self-driving doesn’t mean anything…unless they’re at the top of the food chain. A company building smart tire valve stem caps or vomit sensors could theoretically “win” their piece of the self-driving market, but Twitter rarely lights up over secondary battles. People want to know who might win the big self-driving game, where a handful of players glue everyone else’s products together to commercialize the technology at scale.

The Game(s) Self-Driving Isn’t

Because the self-driving industry has so many sub-verticals, and the potential market is so vast, overly simplistic game analogies just aren’t helpful. What does Tic-Tac-Toe teach us? Move first and take the center square. That’s great for Tic-Tac-Toe, but it won’t help you master checkers or chess, let alone scale up to real world complexity.

A Go game board and checkerboard

Even chess — arguably the most popular strategy game of all time — isn’t really a useful lens through which to understand the real world. When people use chess analogies, it’s rarely chess they’re comparing something to, it’s a chess move. “Chess” has become the catchall for anything considered complex, and checkmate just means winning move. Declaring “X played checkers while Y played chess” doesn’t teach us anything about checkers or chess, or how Y checkmated X, other than make sure you know what game you’re playing.

If we want games to illuminate the real world, then we need to look at games whose field of play resembles it, and mechanics mirror it.

And The Self-Driving Game Is…

The self-driving industry’s game is the classic board game Risk. Any company that didn’t see this five years ago faces some serious issues in the near future, if they don’t already.

For those who’ve never played Risk, let’s break down the original game, then discuss why and how it explains the future of the self-driving industry so well.

A board game with a colored map of the world with a brown backdrop

Risk is played on a map of the world. There are many game variants, but the classic one is World Domination, designed for 2-6 players. Officially, there are two phases: Setup and Combat. Unofficially, there’s a third I call Equilibrium, but we’ll get to that later.

The Rules of Risk

Risk is unlike traditional games like checkers or chess, where players start with identical forces and mirror image positions before combat begins. In Risk, the game is largely — but not absolutely — won during the setup phase.

Setup starts with an empty map. Each player receives an equal stack of armies, and one-by-one take turns placing a single army on an empty territory. When all 42 territories are claimed, players take turns placing one additional army on territories they already occupy, stacking multiple armies in anticipation of launching attacks on enemy territories.

Once setup is complete, the combat phase begins. Battles are won by both chance and attrition — players roll dice to determine the outcome of battles — so superior numbers are essential. New armies are generated at the beginning of every combat turn. Hold more territories, generate more armies. Hold entire continents and you get a flywheel effect.

The better your initial setup, the shorter the path to spinning up that flywheel. The key is not only to recognize which continents might be easiest to seize, but how to delay and deprive opponents from seizing continents of their own.

The country pairs in yellow below are the most important territories to seize during setup, representing the choke points essential for attacks over sea. For example, hold Iceland, and Europe is safe from North American attacks via Greenland. Hold Europe AND Greenland, and the North Americans are deprived of their flywheel while you exploit yours:

A board game showing a colorful map of the world with a brown backdrop, with several regions encircled in yellow

If only one player wisely completes Setup, the Combat phase is a formality, and winner-take-all is inevitable. If two or more players set up wisely, combat will eliminate the weak, and the dominant players keep fighting until they get tired, declare a draw, and literally go home. Why? Because when a bunch of friends get together to play Risk, the game isn’t fun once the two strongest players reach Equilibrium, which is the unofficial third phase after Setup and Combat.

Risk Equilibrium is winnerS-take-all by another name. Elimination of all competitors might be possible, but it’s tedious and unnecessary. Victory is defined not by eliminating all competitors, but by making elimination of oneself cost and time-prohibitive. If your empire can stand the test-of-time, or at least make it through the end of game night, you’ve won.

But you’re unlikely to be alone.

Let’s Talk Risk: Autonomous Vehicles (AV) Edition

Great games never die, they get expansion packs and licensed variants. In my fantasy version — let’s call it Risk: AV Edition — trust, safety, and technology that “works” aren’t goal posts, they’re table stakes.

By “trust” and “safety” I mean a vehicle I’m comfortable putting Coco in alone to get to school. I’m already there. By technology that “works,” I mean a vehicle that does not require a human in the driver seat. (Coco’s been in a Waymo, an Argo ride is next, and she thinks being driven around by what she calls “Mr. Robotaxi” is normal.)

Risk: AV Edition is played simultaneously on multiple maps stacked like pancakes. Each of these maps represents different areas in which AV companies can operate different types of autonomous driving services, including but not limited to: sidewalk bots, robotaxi, last mile, middle mile, and trucking. (A future expansion pack could include privately owned autonomous cars, which is a very different game, with different mechanics.)

Here’s just one vision for what Risk: AV Edition might look like:

A stack of board games with colorful world maps rising high, each labeled with different names including Trucking, Middle Mile, Last Mile, Robotaxi, and Sidewalk Bot
How many Risk AV boards is company (X) competing on?

Different maps may have slightly different territories. Sidewalk bots, robotaxi and last mile maps consist of cities. In the spirit of the original Risk’s simplified countries, only large cities are depicted. Middle mile and trucking maps consist of cities/distribution centers linked by routes, much like the ocean crossings in the original game.

Setup occurs simultaneously across all maps. Which business do you want to dominate first? It might be global robotaxi. Or maybe robotaxi and last mile on one continent. Or maybe it’s just middle mile and trucking on two continents.

Armies still function like armies, but they’re effectively business units. The first army placed on any territory is your operations hub. Move it or lose it, and you lose that territory. Every additional “army” is a business unit composed of everything and everyone else, from fleet vehicles, to safety drivers, and local policy, strategy and marketing teams.

You still gain armies from holding territories, but some are far more valuable than others, like cities and routes with consistent and optimal regulations, safer human drivers and better weather.

Flywheel effects may be generated in two dimensions: horizontally, by holding an entire continent on any one map, or vertically, by holding two or more of the same territory on different adjacent maps. The horizontal flywheel is a function of partnerships, reliability, and brand/app loyalty: it’s nice to know Uber and Lyft work anywhere in the United States, but that isn’t necessarily true everywhere. The vertical flywheel is a function of efficiency: for example, if your robotaxi and last mile operations can share a hub, you’ve got economies of scale:

An illustration of two game boards with colorful world maps stacked atop one another labeled with the text Robotaxi and Last Mile
Share an operational hub across boards and spin up that flywheel!

Like original Risk, the AV Edition is largely won in the Setup phase, yet unlike it, the Combat phase can begin on any map before Setup is complete. Why? Because AV Edition players don’t start with enough units to fill up all the territories on every board. On the contrary, they have to pick geographic areas of focus, and combat begins the instant one player chooses to “compete” with another…even if empty territories remain.

An illustration showing two brown game boards with colorful world maps atop one another labeled with black text reading Robotaxi and Last Mile

Vertical efficiencies or horizontal combat? In AV Risk, the choice is yours.

Victory conditions are the same as Risk: winner-take-all, but given the 5x complexity of the game map pancake, so is the overwhelming likelihood of Equilibrium. Even if only one player sets up wisely, winner-take-all is unlikely. If two or more players set up wisely, Equilibrium is guaranteed.

Who Is Going To Win Risk: AV Edition?

The next time anyone suggests the self-driving race is almost over, they should consider what Winston Churchill said in late 1942, when many thought the Allies might lose the Second World War:

“Now this is not the end. It is not even the beginning of the end. But it is, perhaps, the end of the beginning.”

The game of Risk: AV Edition isn’t close to over. The Setup phase is barely underway, and Combat hasn’t begun anywhere. The different map territories are less than 50% filled out with armies. In San Francisco, for instance, stacks of rival robotaxis are being assembled to face off in just ONE territory, on ONE board.

Meanwhile, large and attractive territories are being taken unopposed across multiple boards by companies like Argo, who are quietly spinning up those flywheels.

Scale Wins

Obviously, the first player or players to scale up unit production will win. But Risk: AV Edition has one more game dynamic I call flyscaling, and it’s the shortest path to victory.

In traditional Risk, there is only one board, with six continents and one flywheel each. But in Risk: AV Edition, multiple boards and vertical flywheels mean hundreds of potential flywheels, which can power each other. Spin up a lot of flywheels across boards and you get to scale very, very quickly.

Flyscaling replicates a dynamic that doesn’t exist in traditional 2D Risk. The AV Edition isn’t about destruction, but creation. Whereas horizontal flywheels replicate production economies of scale, and vertical flywheels operational economies of scale, their sum replicates the compounding benefits of global business, where scale begets scale. Lower production costs → bigger fleets → more cities → more happy customers → more data → better technology → lower production costs.

Once a player begins flyscaling, they become almost impossible to stop.

I’ve long said that less than 5 companies would dominate the global self-driving industry’s inevitable equilibrium state. I didn’t say which, or when, but 3 years ago I jumped the wall and took my first corporate role at the only self-driving technology company whose strategy I could see in three dimensions, ten years out. Time and events have shown how right I was.

Who else will make it?

You can’t buy Risk: AV Edition, but if you buy six copies of Risk, some sharpies and a ream of paper, you’re ready to become a self-driving technology strategist. The next time an autonomous deployment is announced, try to figure how many autonomous vehicles that press release actually represents, and place the appropriate number of units on the right territory. Then sit back, look at all the boards, and ask yourself: when and where will competition begin? Who will generate those flywheels? How many companies’ Setup phase was executed wisely?

I can only think of three companies who understand Risk: AV Edition.

If Argo is one, who do you think the other two are?

Alex Roy loves board games, driving, self-driving, and commuting in his Tesla. He is also the Director of Special Operations at Argo AI, host of the No Parking & Autonocast podcasts, editor-at-large at The Drive, founder of the Human Driving Association, author of The Driver, and Producer of APEX: The Secret Race Across America. He held the Cannonball Run record from 2006-2013. You can follow him on Twitter and Instagram.

Choose your lane

How Autonomous Vehicles Distinguish Between Bicycles and People Who Ride Them

How Autonomous Vehicles Distinguish Between Bikes and People

When it comes to how autonomous vehicles see the world, humans come first, literally. Autonomous vehicles (AVs), like the kind operated by Pittsburgh-based Argo AI, use Machine Learning to detect and classify the objects in their surroundings, identifying people...
Why The League of American Bicyclists is optimistic about autonomous vehicles

Why a Leading Cycling Advocacy Group Is Optimistic About Autonomous Vehicles

As autonomous vehicle use grows, AV companies and the League of American Bicyclists are collaborating on how to ensure cyclists and motorists can share the roads safely, even if the “motorist” is artificial intelligence software. As part of the...

Self-Driving Is Arriving Right On Time. Just Like Ice Cream Did

Seven years ago, I was a self-driving skeptic. Not of the technology. Of all the “experts” promising autonomous vehicles would be everywhere by 2020. You didn’t need to be Nostradamus to know that was ridiculous. All you needed was...
Illustration of a futuristic parking deck turned into a mixed-use space, with AVs driving by

How Autonomous Vehicles Could Help Transform Parking Lots

Researchers say it’s likely that autonomous vehicles (AVs) can help reduce the need for parking lots, opening more room for grass and trees and other elements of nature. It may not seem like it when you’re circling the block...
An illustration of an Argo autonomous vehicle in teal against a backdrop of buildings, a bicyclist, and research papers

7 Big Breakthroughs From Argo Research at CVPR 2022

The 2022 Conference on Computer Vision and Pattern Recognition (CVPR 2022) is nearly here. Thousands of computer scientists, software engineers, and researchers from around the globe will gather in New Orleans to review and discuss their latest work in...

Researchers Predict the Future With Lidar Data

Researchers at Carnegie Mellon University’s Argo AI Center for Autonomous Vehicle Research, a private-public partnership funded by Argo for advancing the autonomous-vehicle (AV) field, say they have come up with a way to use lidar data to visualize not...

Must Reads