My Son's First Waymo Ride
On the self-driving S-Curve, and will my children ever learn how to drive
“Daddy, can we take the car-with-no-driver to the game?” my son asked.
We are visiting San Francisco this month, and while he was disappointed that he won’t get to attend a Golden State Warriors game — the NBA season doesn’t start until October — my son got excited about seeing the Golden State Valkyries, the new local WNBA team. And if we could also use Waymo to get to Chase Center, that would make it an even cooler experience.
Our first conversation about self-driving cars was before my son even celebrated his third birthday, back when we were still living in San Francisco. And now, at age 8 - he got to ride in one.
“Can you get in your seat by yourself?” I asked. It was the spring of 2020. The preschool was closed due to COVID-19, and I used to spend mornings with my son in Golden Gate Park. As we headed back to the car one day, he pointed at the front seat, next to the driver. “Can I ride here?” he asked.
“Sorry, buddy, I need to look up the exact requirements, but I think it’s going to be at least ten more years before you can sit there,” I replied. Then he pointed at the driver’s seat itself. He liked sitting there while the car was shut off, pretending to drive.
“When could I ride there?” he asked next.
“Well, you’ll need to get a license first, and you can only start learning once you’re 16 or so, I think. So it’ll probably be 14 or 15 years before you can drive yourself.”
Then I paused.
I thought of all the self-driving cars that were already roaming the streets of San Francisco back in 2020.
“Actually, you’ll probably never drive. By the time you’re 16, the computer will do the driving for you,” I revised my answer.
I imagined diving into explanations about how computers work, or the history of automating different tasks that had been done by humans in the past. But he wasn’t particularly interested. “Can we listen to Elmo’s Song on the way home?” he changed the subject.
Fast-forward to this month — five years later — and my son was eagerly waiting on the street for the Waymo car. He got jumpy as it pulled over to pick us up. I let him push the button in the app, and the door handle popped out. So did my son’s eyes, after he pressed Start Ride and the car started rolling down the street. He pointed at the wheel spinning by itself. My son was closely following every small move: stopping at the crosswalk line, figuring out when it was OK to go again, making the turn, climbing a hill with limited visibility, handling a delivery truck blocking half the lane.
Until it was no longer interesting. The excitement cooled off. Waymo’s driving is so smooth that it’s actually boring. We switched to talking about the game.
Valkyries fandom justifies its own blogpost. Chase Center was packed. Tickets were expensive, and the lines were long. The atmosphere felt like a Warriors game. And the match was great1! It was awesome both from an equality standpoint — women can definitely play basketball! — as well as from a business point of view. Talk about TAM expansion: the demand for basketball is there even during the summer. And there are twice as many potential players.
We called a Waymo again on the way home; this time not as a tourist attraction, just because it was 40% cheaper than Uber. A few minutes into the ride though, we hit our first autonomous-driving issue: congestion.
The exodus from the stadium area caused a major traffic jam. Have I mentioned how popular the Valkyries are? Cars — with human drivers at the wheel — were aggressively muscling their way to turn right from 4th Street onto Mission Bay Blvd. Our Waymo was just sitting there, signaling, while human drivers going bumper-to-bumper, and over the bike lane, kept bypassing us.
It’s a fascinating challenge. Waymo vehicles are programmed to obey traffic laws. Even those that aren’t always followed by humans. A Waymo making a traffic violation would be a gift to the Luddites. The general public, and regulators in particular, are closely monitoring the self-driving behvaior. Just like my son did when he first entered a Waymo. The engineers can’t allow it to bypass traffic through bike lanes or squeeze too closely to other cars. I kept staring, thinking of the engineers who have to deal with this type of situation.
“Daddy, why are we the only car that isn’t going?” my son asked. He was tired at this point. It was getting late, and the ball-game energy had worn off. I quit pondering and contacted support.
The representative quickly took over our route. Waymo started signaling left. Into an empty lane. It was a small detour, but at least we didn’t have to deal with traffic. “I apologize for the inconvenience; you will receive credit toward your next ride,” the representative concluded the conversation, as our Waymo was smoothly making its way through dark, empty streets.
Carrying my son out of the car and into his bed, I was reminded of the other day in Golden Gate Park. It’s been more than five years since, and I’m no longer so confident in my prediction. Perhaps he is going to learn how to drive after all.
“If the computer is doing all the driving by the time I’m 16, what’s going to happen by the time I’m 8?” Imagine my three-year-old son asking a hypothetical follow-up question back in 2020. Working for Google a decade ago, I had seen the self-driving videos. They were already mind-blowing back then. I would have guessed self-driving to be far more widespread by now. Like way-mo(re) than the ~1,500 Waymo vehicles operating in five U.S. cities.
Bill Gates famously observed that people overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten. But as I, in retrospect, overestimated the change that occurred in the last decade, the only reasonable conclusion2 is that predicting change is hard!
Don’t get me wrong, Waymo’s exponential growth is highly impressive. It’s just that — when it comes to revolutionary new technologies — the exact shape of the S-curve is extremely hard to predict. How long will it take before we reach the inflection point — in terms of the technology, the cost structure, the regulation — that allows autonomous driving to really take off?
The long-term vision is, of course, 100% autonomous. There are no traffic signs. No lanes. No traffic lights or speed limits. Computers govern everything, and the cars coordinate through digital communication and APIs. To fully get there, though, we’ll have to slog through the hybrid middle-ground period. As we do now.
How long does it take? Asking for a friend (my son, in this case).
The percentage of young people getting a driver’s license has already been decreasing for a while now. The Atlantic wrote about The Decline of the Driver’s License back in 2016. This trend may help accelerate things; fewer human drivers on the roads makes it easier to adopt the fully-autonomous vision. On the other hand, watching Waymo struggle last week, considering that it still doesn’t do highways or airport pickups, and that each Waymo car is still absurdly expensive — all make me question whether my 2020 forecast (about my son never learning to drive) will hold up.
He may still drive, only as a hobby. Just like people enjoying a metalworks class today, though few actually earn a living as blacksmiths. Or my son may work for a company like Waymo as a specialized human driver. Remotely taking over stuck rides and getting them out of traffic jams.
Whether he drives or not, my son would have great about the evolution of driving, which he could tell his children on the way to basketball games together. Can you believe my dad literally had to steer a wheel and push pedals for the car to move? I just hope their self-driving drone, or whatever, doesn’t get stuck in traffic on the way back home.
Even though the Valkyries lost. By a single free throw shot, with one second remaining.
This is inspired by Daniel Kahneman's Thinking Fast & Slow: the best thing we can learn from surprises that make our forecasts go wrong is that, the world is surprising and hard to predict.
Enjoy reading.