> When we think about our next cities, Los Angeles jumps out,” said Waymo’s co-CEO Tekedra Mawakana. “LA is a remarkable, vibrant place – and Waymo’s experience leaves us best positioned to tackle its driving complexity.
That's a funny way to say the next biggest place with 300 days of sun per year.
> With approximately 13 million residents, the Los Angeles metropolitan area is one of the largest ride-hailing service areas in the world and the third largest in the U.S., with an estimated market opportunity of $2 billion in 2022. As a commercial opportunity, Los Angeles is the equivalent of a dozen smaller U.S. ride-hailing market opportunities combined.
I don't understand the division of CEO roles into two CO-CEOs. Is this generally effective? I would imagine this would make tough decision making nearly impossible? Who takes the blame for bad decisions vs. good decisions? Anyone have some history on why Waymo has two CEOs?
This is a really good question...its unusual but I didn't process it as such because it read as a smart & effective split. Last CEO was from Detroit, co-CEOs are effectively technical lead and ops lead. Makes sense for what is a one of kind tech lab hitched to a ride hailing business.
You also made me think deeper about the unusual situations where I've seen co-leads at a FAANG. Honestly? I think it helped.
There's a very strong "bullshit" culture that can set in. It is a mere step beyond your average upper class Americans cultural mileu, so its underrated how degenerate it turns with that one more step: suddenly, leadership is never leading, and being extremely territorial about general subject matter while never doing anything ambituous enough to fail can leads to cylical ineffectiveness.
With two leads, there's two people to appeal to, and two people who need to argue and agree eventually. They can't sidestep that conflict is a crucial part of polishing.
I guess that when the last CEO was leaving, he had two candidates to replace him, but both treatened to leave if the other is made the boss, so that CEO made the solomon decision: he made them both "ceo".
A company the size of Alphabet doesn't even understand the meaning of the adjective "chief", as in, "most important." Also, the "co-CEO" is a womens' studies lawyer that spent 16 years managing "public policy" at each failing company de-jour of the past decades: AOL, Yahoo, eBay, and is now "co-CEO" of an extremely technical company? Yikes.
Waymo does work in rain and fog. They are permitted to operate in both of those conditions in San Francisco. They also do extensive weather testing, including areas with snow, so presumably they are close to cracking that as well.
...yet. I think Waymo's approach of launching the service in the cities when it is easy first, and over time expanding to more troublesome climates is much more sensible than Tesla's "humans can do it, so AI should be able to, with just cameras".
> Waymo's approach of launching the service in the cities when it is easy first, and over time expanding to more troublesome climates is much more sensible
Is there a fundamental answer to this question? Start with low-hanging fruit and then build up? Or begin with the most challenging bits before gliding forward? The risk with the former is you end up in a dead end. The risk with the latter, finding the hard problems intractable without the experience of solving the easier ones.
From a business perspective it's probably not even close. Start with the easiest + biggest. Scale matters here so you can't leave big easy markets there for the taking.
Correction: Tesla FSD doesn't work in rain or snow, or even for that matter all that well in regular daylight.
Cruise is slightly better than FSD, but not by much. It's prone to driving really slowly and phantom braking, but at least it doesn't swerve into light poles or oncoming traffic like FSD does.
What happens to the passengers if you're on the freeway and it starts to rain. Does the car just dump them on the shoulder? Or is it smart enough to find a good place nearby to kick them out?
Great news! It does seem like the 5th gen hardware will enable scale and real revenue generating operations for Waymo.
The videos coming out of driverless rides in SF [1][2] and Downtown Phoenix [3] (both employee-only currently) look very impressive. So I don’t doubt they can crack LA.
It will be interesting to see how Waymo progressively reduces deploy time in a new city.
I expect the cultural impact of launching in the world's media capital to be high. If they're good it'll be strong advertising and everyone will want it. If there are prominent things to criticize you can be sure they'll be criticized heavily.
I assume (hope) they're now reasonably sure the service doesn't suck and preparing for further expansion.
Very exciting news, as more people use autonomous vehicles, the less parking needs there will be and hopefully more parking lots can be reclaimed with plant life, that benefits the whole earth with better air and calming energy.
It doesn’t seem likely that parking is even half of street surface let alone street plus sidewalk.
That would require more parking width than travel lane width on any given segment with parking (to account for areas where parking is not permitted plus intersections). Few streets have less travel width than parking. Many have a multiple.
>That would require more parking width than travel lane width on any given segment with parking
Not sure about Houston, but that is the case in many dense US city-suburbs. They'll have two-way roads with parking on both sides, and the roads will be about three cars wide. The result is cars parked on both sides, with drivers sharing the middle lane for both directions of travel.
When you see someone driving towards you, it's a game of chicken to see who will temporarily pull off into a vacant parking spot. It seems to happen because the lots aren't big enough for 2-4 car garages and driveways.
And for our friends across the ponds, yes: US households really do demand space for at least 2-4 cars per family.
I believe you'll find that the difference between a machine and a human driving the same car can be dramatic. A lot of humans just drive like complete dicks. Switching the throttle map on my car to "eco" and driving in a sensible way gets me +50% MPG compared to setting the throttle map to "sport" and driving like a teenager.
Why do you believe that? These taxis will increase the number of vehicles on the road, because
> more people use autonomous vehicles
Those "more people" will be bus riders and the housebound, and extra trips by people who already use cars in one form or another. The latter's own cars will be parked somewhere. Everyone commutes at the same time of day, no-one wants to share with randos.
Or more housing. Or virtually anything else would be a better use of land than parking lots.
The amount of valuable land that widespread deployment of autonomous vehicles could unlock is just staggering. Not to mention the avoided costs of building undercover garages.
They mention unprotected left turns. I'm really curious how this will be handled in LA because there are a lot, and if you aren't at least a little aggressive, you will miss lights constantly. And some lights are miserable if the cars turning left in front of you are passive, and only one makes it through per light change.
Missing lights isn’t a problem when your time and attention isn’t consumed by the pointless toil of driving. It’s just the same 2 minutes it takes to microwave some popcorn.
That’s the same experience as being in a taxi though, and plenty of people get frustrated by slow taxi rides. Nobody wants to spend more time than they have to in a taxi, whether driven by a human or computer.
Yea, that's a fair point. Though I'd still be upset if I'm waiting in a line of Waymos turning left that go once per light change when I'm not in a self-driving car.
A big move for sure. The LA demographics are much less tech heavy than other cities. Hopefully they succeed and have a model where transportation become cheaper and more efficient and an individual owning a car becomes an option rather than a necessity.
> have a model where transportation become cheaper
Multiple billions have been spent on developing this technology, its hard to imagine they won't squeeze out maximum profit with a multitude of dark patterns if they do become mainstream.
Google is known for a lot of things, like cancelling products, but "dark patterns" is definitely not one of its things.
Of course they'll squeeze out maximum profit, like every other company does. Fortunately, you can just check out the Wikipedia article on "profit maximization" to understand why maximum profit doesn't come from maximum prices, but generally comes from a middle price that attracts more customers.
In any case, it's always going to be cheaper than a ride with a human driver, so I don't know what you're being so negative about.
More likely than other large companies to not have dark patterns, is probably the more correct sentence. It's especially correct with some of their similarly "non-core" products.
I remember moving off of Google Fi a couple years ago and it being incredibly easy to cancel and move my number. I didn't even have to call anyone. A few clicks and it was done. If you've ever tried to do anything with another phone service, you'll know that's unheard of.
You’re both right. Google
doesn’t use dark patterns to nickel and dime like most companies do, so it seems like they don’t.
They absolutely use all means necessary to preserve the data feeds and tracking that enables their core advertising business, such as the fact that incognito mode, well, isn’t incognito.
But you're replacing low-paid drivers that supply their own cheap cars with very high marginal-cost cars and sensors (and low marginal cost software). Its hard to imagine them turning a profit from owning and operating a fleet of cars - it seems like a business model closer to Hertz than Uber, and neither are terribly profitable.
Huh? Seems like the current Waymo cars run a little under $200K with all the sensors. Obviously that's going to come down at scale to something closer to $100K let's say.
And yes you can contrast that with a $25K Uber vehicle, but the average US Uber driver makes $42K a year, and since taxis are traditionally split into 3 shifts, we can assume the cost of a 24-hour driver is closer to $100K. In NYC, Uber drivers make something like $30/hr, which means $263K/yr for a 24/7 driver.
So a "low-paid driver" costs as much in a year as one or two single autonomous vehicles. In which case, it's hard to imagine Waymo not turning a profit. The reason Hertz and Uber aren't particularly profitable is because they have tons of competition, there's no real competitive moat. While Waymo is going to undercut driver-based rideshare drastically. It's going to make tons of money once the R&D is recouped.
On a per mile basis they might not be expensive per car.
Hertz and Uber struggle because of competition, not business model. If Waymo is the only self driving player and hence lowest cost mode of transportation, they'll make a fortune. If there are several players, they won't.
They have plenty of competition though - if Waymo is operating a taxi service, they are explicitly competing with Uber/Lyft/traditional taxis/bikes/buses/all the other ways to move around - generally pretty low-margin businesses where their competitors set a pretty strict ceiling on the price you can charge (who would pay a lot more for a waymo ride vs a competitor? maybe some, but not many). Waymo would have high capital costs for the cars, plus still need centralized service infrastructure (re-fueling, re-charging, who cleans a waymo when a drunk person throws up in it?). It would also still have the same problem other companies have - lots of people might want to travel to/from work, and on weekend evenings (or to big events), but not much the rest of the time. An uber vehicle might serve double duty as a personal vehicle during off-peak times (where the driver might even have a different job), but a waymo car needs to pay for itself, so it needs to be driving a lot. I could see them licensing their software to other companies like android auto or something, but that also doesn't seem like a way to print money (how much does android auto earn for google?). It just seems like a very hard business to transition from 'fun software project' to 'clearly profitable independent business.'
Agree with most of what you are saying, except that the driver is a big enough component of the cost structure that they should be able to price something like 20% below services that require a car. I just took a $65 uber yesterday which took I think 42 mins. Assuming the driver cost $15/hr and is running at 75%(??) capacity, the labor cost baked into my ride was around $15. If Waymo offered the same ride at $55 (ie leave $5 on the table for some of the issues you highlight), Uber etc are in a lot of trouble.
It has to be cheaper than human drivers and compete with other self driving offerings. That said this is definitely a difficult market to break into (at least for now). At the same time, an open source model isn't completely out of the question. Datasets of specific cities can be built pretty easily these days (even if expensive). If that ever happens, self driving will have no added cost.
I think the Hollywood market is actually an interesting one. Having a private door to door transportation option for celebrities is likely a very enticing option, and I would be curious to see if they specifically try to go after that market. I know I personally would choose a private waymo vs an Uber/Lyft/taxi even if it meant a few extra bucks. The ability to sit privately and not have to make awkward small talk is a big selling point IMO.
aside from such a wide range of parts, there is also a wide discrepancy in what LA natives do and care about, compared to what transplants do and make reference to. almost no overlap.
I think we can correlate what I'm referring to with the monthly rent though
Can confirm having seen Waymo vehicle's driving in my hood several times this past week. I'm concerned that Waymo will never be able to drive "the LA way" and ultimately cause congestion.
For example it's standard in LA that when the light turns yellow then two cars are able to make a left turn in time to not back up the perpendicular Lane.
Or if you are turning left you should try to make room on your right, so that those going straight can get around you. No way Waymo vehicles will be doing that.
All in all there are certain behaviors that are necessary in a city with traffic like LA that would certainly be outside the strict safety parameters that (I assume) Waymo operates under.
> Or if you are turning left you should try to make room on your right, so that those going straight can get around you. No way Waymo vehicles will be doing that.
Why do you say that? In this video you can see it doing exactly that, pulling to the side a bit to make room on the left for a vehicle doing a weird u-turn inside an intersection: https://youtu.be/rGfV_3HPnYE?t=442 If anything, common behaviors should be among the easiest for an AI system to learn.
Probably not as it’s not a meaningful comparison. One is completely driverless (L4) and the other requires a driver with hands on the wheel ready to take over (L2). Removing the driver (safely and reliably) is like 99.99% of the problem in self driving.
That means FSD has to work only once for a YouTube video, but a Waymo has to work every single time as there’s no driver to rescue when things go wrong. So apples to oranges.
The difference is that Waymo works and FSD does not work, so it's going to be a short video. FSD cannot take a trip without a man behind the wheel, full stop. It is not any more "self driving" than any other car.
I hope they gain enough xp from this and other cities to test out harsher cities. Additionally, if this offering becomes attractive enough more challenging cities like NY and Chicago may themselves get involved in this process.
To be frank, especially for anyone who hasn't lived in LA, the LAPD doesn't do anything for the public unless it's something involving guns and/or gangs.
It's not unlike how developers mark most bugs as "won't fix".
I've seen active break-ins and robberies in LA, and called the police only to have nothing being done in those situations. While anecdotal, I'm afraid this is a really common experience that has only gotten worse with COVID-flation.
Nah, breaking a Waymo is property crime against a large corp, and that makes them pay attention; they fall asleep when it's just the general population and small business, but there's backing and funding at stake when capital is attacked. Police departments over the last decade or so of protest have demonstrated willingness to defend Apple stores, Louis Vuitton, etc. They mostly don't succeed at stopping burglaries-in-progress but they do tend to go through camera footage and pick up people after the fact in those incidents.
There's some precedent for other unattended things like rental scooters. If it becomes a nuisance for the police, they may choose to ignore self-driving car vandalism.
The solve rate for grand theft auto is almost always the lowest rate across felony crime types, just below other property theft. The highest solve rate is for murder, and LA hovers around 50% for that.
This will be interesting to see if they can crack LA - drunk and overly aggressive drivers, crack addicts running around the streets, and vandalism.
You can already see this "tragedy of the commons" type of behavior with the unusable public transport system (it's a nightmare to use the trains between Koreatown and DTLA, especially for women) and in the private sector with Lime and Bird scooters.
Very exciting if it works out, but I don't have high hopes.
I’m really disappointed at the kind of generalizations and misconceptions you’re using here to describe a vast and diverse metropolitan area.
You’re essentially blaming “elements of sin:” it’s all about the crackheads and the crazies.
You’re magnifying a very small portion of the population while diminishing and disrespecting the mostly-normal law-abiding citizens who actually live in LA.
You want to know why LA has drunk drivers and scary bad transit? It’s not because of all of the vague sin of the “city problems” you describe.
It’s at least in part because of LA’s unsustainable car-focused development. (In that respect, Waymo might act as a welcome band-aid, but it’s still a band-aid).
Why are there drunk drivers? Because you can’t walk or take a train/bus to the bar.
Why is transit scary? Because only the truly desperate actually use the barely-useful mass transit in LA, because LA was built for cars.
Why are the crackheads out vandalizing stuff? Maybe because car-focused development exacerbates wealth inequality [1] by encouraging the development of socioeconomically segregated communities, and by gating economic opportunity behind the most expensive form of transportation, requiring everyone to buy and maintain personal vehicles at high rates (at least one vehicle per person). Don’t have your own car? Hope someone can drive you to work! Want to interview for that job? It’s far away, hope you can pay for gas! Disabled? Elderly? Guess you’re not going anywhere!
Regarding crack addicts, LA isn't that much worse than SF where they have been testing for a while. Yes, LA is bigger and has more crack addicts, but the density of crack addicts per square kilometer is about the same between the Tenderloin and Skid Row.
Waymo will probably just geofence out the worst areas anyway.
That's true, there are definitely ways to make this work if we're confining these to very restricted spaces with active monitoring.
I look forward to a future LA where car ownership is optional. There's a lot of aggressive urban planning that needs to happen and implemented for it to work. Downtown Culver City is a good example.
Most of Los Angeles is a weird sort of "urban lite" environment. All the problems you describe are much worse in the areas Waymo is already operating in in San Francisco.
I completely agree. And autonomous vehicles solve none of the issues of regular cars, except maybe injury rate. The rest is the same -- emissions, inefficiency, expense, noise, traffic congestion, unwalkable cities, and so on.
They're electric, so that solves the emissions and noise problems.
Once self-driving cars are common people will figure out that it's safe to jaywalk in front of them, and nobody minds inconveniencing a robot, so I expect jaywalking to rise substantially too.
A self-driving hire car isn't idle 23 hours a day like a normal car so that should solve some of the inefficiency and expense issues.
All I'm left with on your list is congestion, and that'll remain, even if the self-driving cars start communicating with each other to allow closer spacing. Making something cheaper and/or easier inevitably increases usage.
> They're electric, so that solves the emissions and noise problems.
Past a certain speed, the majority of noise is coming from the tires, which also result in emissions. An EV is certainly better than an ICE, but there's no way to make a vehicle per person more efficient than a vehicle per tens/hundreds/thousands of persons.
> All I'm left with on your list is congestion, and that'll remain
Smart car sized AEVs, rather than Hummer sized AEVs. Individual "pods" the size of motor cycles, rather than cars.
Today we all buy "peak usage cars". i.e. "Once a quarter I drive 300+ miles", "Once a week I drive my whole family to the park/hike/...", "I'm not yet an off-roading type of person, but might as well spend the extra $10k just incase I wanna get into it. (never gonna get into it)", "What about something that will go fast on the track, just incase I need that". Then 99% of the miles this giant 5+ person car with storage will ever drive is 1 person trips to and from work bumper to bumper.
When the car isn't owned, you don't need to buy "peak usage cars". You can order a car sized for the usecase.
I'm pro-electric car, but I wouldn't say they solve the emissions problem. Tailpipe emissions, sure. But tire particulates are a surprisingly large fraction of current automotive emissions, and they're not going away.
In fact, if real self driving ever becomes widespread, it's likely to induce demand (sure, why not live a 2 hour drive from work if you can relax the whole drive home?) which will lead to increased tire emissions.
>They're electric, so that solves the emissions and noise problems.
60% [1] of the electricity in the US is produced by burning fossil fuels (21% coal!), and this will not change in a long while. In California specifically it's a bit better, with ~40% fossil fuels [2]. Also, battery production specifically is a tremendously energy intensive industrial process that also involves a lot of emissions [3]. Factoring in efficiency losses it results that electric cars, while maybe better than ICE cars, are inferior to biking, walking, and electric trains.
As for the noise, as other commenters pointed out, at average city speeds most of the noise a regular car makes comes from the tires rolling, not the engine.
>Once self-driving cars are common people will figure out that it's safe to jaywalk in front of them, and nobody minds inconveniencing a robot, so I expect jaywalking to rise substantially too.
If this starts to happen, congestion will only get worse as the average speed of roads goes down. Cars suddenly stopping in the middle of the road isn't good for traffic flow. It also doesn't address the fact that most people in LA can't leave their homes and be anywhere just by walking.
>A self-driving hire car isn't idle 23 hours a day like a normal car so that should solve some of the inefficiency and expense issues.
I concede that this is strictly true -- having a single shared vehicle is more efficient than several privately owned. However, most people need to use their cars at the same time for their commute. Demand still peaks in the morning and in the afternoon, and if you have enough vehicles to cover that, during the rest of the day they will be idle, same as now.
>All I'm left with on your list is congestion, and that'll remain, even if the self-driving cars start communicating with each other to allow closer spacing. Making something cheaper and/or easier inevitably increases usage.
So poor people still don't have access to transport, and the congestion is even worse than before? Why would this be a good solution to the car problems of LA?
> As for the noise, as other commenters pointed out, at average city speeds most of the noise a regular car makes comes from the tires rolling, not the engine.
If that's true, cars are driving too fast. EV's are silent to about 30mph, and any speeds above 20mph are lethal to pedestrians.
> So poor people still don't have access to transport,
Poor people are the population most reliant on taxis & Ubers. Because they don't own a car, they take taxis for emergencies. They can't afford to do it often but it is a key mode of transportation for poor people.
> and the congestion is even worse than before? Why would this be a good solution to the car problems of LA?
Because it solves the problem for rich people. If a robo-taxi is the best way of getting around for rich people, then some rich people will ditch their cars. And once you've ditched your car you stop using a personal car for everything and start choosing the best option based on circumstance. And sometimes that option will be the train, or walking and not necessarily a robo-taxi.
In some trite sense what you're saying is true, but if autonomous vehicles are widely available and cheap they'll drastically reduce vehicle ownership rates, which will help improve the status quo for most of the problems you listed.
That's a funny way to say the next biggest place with 300 days of sun per year.