I think self driving cars are much closer to 10 years out than the "today" that the author asserts, but I think everyone agree's that we've reached the point where they are coming no matter what.
To give the author credit I went into reading this thinking he was absurd but he does make a good point, some of the points I think are over sold, like planned obsolescence but I think his thesis is sound.
Both do a very good job of bringing the reader into the chaotic environment that occurs when startups are in trouble and both have a very similar message. When startups get in trouble having very powerful investors and mentors can make a huge difference.
If/When times get tough, they have people who have a vested interest in seeing them succeed. People who can get them talking to the correct people at the car manufacturers to convince them to use Uber's platform over an internally created one.
If self driving laws drag on many years before they are settled then Uber may be fucked in the end but there isn't any reason why they can't be a cash printing machine over the next 10- 20 years while self driving cars replace humans.
As to car companies cutting out uber, there is this.....
I had a hearty chuckle at the "It’s largely a solved problem." line. It does seem like in the next 5 years or so we'll be about 99% of the way to fully automated vehicles that can pick you up and take you anywhere a 'normal' car and driver can go... But that last 1% is a bunch of incredibly difficult problems that could easily take another 10-20 years.
Aren't self driving cars good for Uber? My understanding is that Uber's plan in the event that self driving cars become practical is to have a fleet of them which can be summoned by users on demand.
Google's an investor in Uber, it's not difficult to connect the dots - Google is interested in applications of its tech, not in owning the entire stack top to bottom.
Self-driving cars are very certainly the future of Uber, and once self-driving tech is mass-available Uber will have the same accessibility to it that Google or Tesla or whomever else does. At that point, they're competing on price and service quality, just like they do with Lyft and Sidecar right now.
I don't see self-driving tech becoming a major sticking point, because it's not like manufacturers are going to only be manufacturing self-driving cars for their own usage once the regulatory hurdles are cleared.
I would imagine that Uber would be investing in self driving cars. I don't think there's as much money in taxiing people around as Uber and others initially thought. Even with the taxi market being artificially constrained the number of taxi driver millionaires seems to be a little low.
At least in some countries and cities getting a taxi at the right times have been a hassle, and somewhat expensive. So even if, like Uber, you can expand the marked and steal customer from traditional taxies, how much money is there really to be made? Getting the profit that currently falls to the drivers seems like a logical next step, if they want to increase their own profit.
The conclusion at the end is that car manufacturers are better suited to benefit from driverless cars as Uber is not in the business of building or maintaining a fleet of cards. But by that same reasoning, manufacturers are not in the business of dispatching taxis.
I'm skeptical that manufacturers would be that nimble to capitalize on the inevitable. That's also not their business model. Building out such an effort in such a large organization would be a challenge. It would be much cheaper for car manufacturers to partner with or buy a company like Uber.
In regard to Uber not having data to create the machine learning models needed for self-driving cards, they could buy that data or start building it up themselves. Google doesn't have a monopoly on strapping cameras and sensors to cars. And Uber certainly has enough capital required. Besides, this is a core strategy for Uber while it's still a moon-shot for Google.
If we reach a day where there is some company, call them company U which has a successful dispatch app business, and another company, call them company C, which has just successfully demonstrated (and passed regulatory hurdles for) a self driving car, what do you think will be easier and cheaper and quicker? For U to develop/acquire/license self-driving car technology or for C to develop/acquire/license a dispatch app?
C could probably pay very little to license a dispatch app from any of the many vendors. Even if it's not popular before, it will be once it's the only dispatch app that dispatches SELF DRIVING CARS! But if U tried to license self driving from C, U would probably have very little leverage and end up signing a contract where U runs the show but ships most of the revenues to C. C would be like Microsoft in the 80s/90s, and U would be like a generic IBM PC compatible computer vendor.
This is a piece of amateur speculation making huge assumptions about advances in technology, and about Uber's ability to adapt to those advances.
One crucial thing to remember, when considering the threat Uber faces from traditional car manufacturers, is that Uber is a software company and they are not. It has a huge advantage in product development, and stands just as good a chance as they do of coming out with self-driving cars for ride-sharing.
Anyone can hire programmers, children can ship products, MLB created a Netflix-like platform to offer media companies as a service, Walmart have made important contributions to open source etc.
The number of steps between Uber's competitors becoming as competent as Uber at software is basically "1. decide to".
Attracting the talent to build a usable product is one of the most important steps in creating any business, or line of business in the case of a large company. That talent is scarce, and tends to prefer certain companies over others. I speak as a former recruiter and current employer, who has had to grapple with the difficulty of hiring good software engineers.
The software engineers I have known and tried to recruit tend to want to work for companies that are more like Google and less like General Motors.
The statement Anyone can hire programmers, children can ship products is manifestly untrue. It's hard to get the right people, and it's equally hard to coordinate all their activities to get a successful product over the finish line.
The companies that have healthy cultures of software engineering and digital product management and design are way, way ahead of everyone else. And those companies are in the minority.
You simply don't know what you're talking about. Effortlessly, first of all, is false. Secondly, while the hiring of thousands of software engineers over the last few years was trumpeted, that doesn't mean GM knows how to manage them. That applies, even more strongly, to the acquisition of Cruise that you cited. That happened three months ago, so it isn't clear how much talent will stay for how long. But the fact that they had to buy it meant they decided they couldn't build it themselves in a reasonable amount of time.
Effortless was referring to how quickly they positioned themselves to compete in an emerging landscape with lots of startups - one deal staffed them up. Whether they can manage those people effectively or ship big projects ... is a problem startups face too.
Yes, but my precise point is that startups lure talent and iterate on product better than most large corporations, with enclaves and exceptions in a few large-ish companies, and virtually none any where else. $1B is the monetary expression of a lot of effort, and it's unclear whether they can actually do anything with what they bought. Most mergers and acquisitions are failures.
Uber is in trouble, but not for the reasons cited here. Driverless cars are not coming as fast as the author thinks. Even in the example he gives, his friend has to drive the Tesla at some point. This means that it has a driver the entire time, even if he isn't doing much most of the trip. Widespread use of driverless (as opposed to self-driving) cars is still 10 years away. Uber's problem is not that they're going to get cut out of the driverless car business; it's that their customers are really the drivers, and Uber offers very little value to them for the cut they're taking. Uber is just a dispatch service, which is rapidly becoming a low-margin commodity business. To the driver, Uber is an app and a sticker. A driver just needs fares in their area, and as Austin has shown, with modern geo-enabled phones, it's easy to jump into the dispatch business to fill that need.
I think this is a point that often gets glossed-over when talking about the advances in self-driving cars:
Self-driving != Driverless
Self-driving cars is largely a technology problem, and one that we have made great advances in the last 5-10 years. Driverless cars are a legal/social/political problem that we have barely started to look at yet. Self-driving is a driving tool, like cruise-control, but the human driver is still ultimately responsible for the vehicle. Driverless on the other hand places the responsibility for the vehicle, and it's occupants, onto the vehicle itself. Driverless implies that you should be able to use it while you yourself are not capable of operating a vehicle. That is a massive step, and it is something that has never been done before. All other forms of transportation, despite advanced automation, still defers to a human in the end. I would say that true driverless cars is still a significant amount of time away, despite our advances in self-driving tech.
I don't think you are taking into account UberPool. HN comments almost never do, perhaps because programmers are generally upper income and do not feel a need to trade money for inconvenience, but UberX is basically a legacy product and now presented as more of an upgrade if you're in a hurry, rather than the default mode.
UberPool is everything. A network of cars buzzing around the city dynamically routing from one task (transport passenger, transport package, deliver food, etc) to the next, and each task is cheaper because it can share ride segments with other customers.
That is where the winner-take-all argument comes from - the bigger the network, the more efficiently Uber can pack multiple tasks into the same driver-hours, with less delay (more jobs means the car doesn't have to go as far out of its way to share capacity).
Uber is more than dispatch - it's quality control, reputation, dispute resolution, capacity planning, payment escrow, (soon to be) routing (it bought a few mapping companies), commercial insurance (try buying that on your own), and increasingly, the efficiency provided by combining multiple tasks in each car.
When people call Uber a taxi service - when is the last time a taxi offered you a cheaper ride if you agreed to share the car for 5 blocks with a guy 10 blocks down the road?
Disclaimer: I used to work there. Throwaway because I'd rather keep identifying information off my personal account.
I still don't think that most of those checkboxes is terribly important to either drivers or riders. If I were an Uber competitor, I'd thank Uber for screening the drivers for me. Now I can offer them fares at half the price Uber can offer them for. You have a point about delivery, but in the end that's just another fare. There will certainly be good dispatch networks, and Uber may be one of them, but none of them will be able to maintain high margins, certainly not margins that would support Uber's valuation.
> Car manufacturers will begin flirting with building, owning, and deploying their own cars…via their own apps.
Music cos. were supposed to do something similar to iTunes etc. Didn't happen.
> Cars will appear when ordered through the app. They will drive off and disappear when you step out of the car. When running low on fuel, cars will coordinate trips to tucked away re-charging depots to be smartly refilled with electricity.
You mean in about 20 years time?
> Then car manufacturers will drastically start reducing the cost of operating these vehicles.
Predictions will always look 100% certain if you presume the cause/problem will never be mitigated.
Search up "The horse manure problem:
"...In 1894, the Times of London estimated that by 1950 every street in the city would be buried nine feet deep in horse manure. One New York prognosticator of the 1890s concluded that by 1930 the horse droppings would rise to Manhattan’s third-story windows. ..."
Good luck for it to be legal that a drunk person can be in the drivers seat of a car. That's going to take 20-30 years. I'm sorry but there isn't "you can't operate an elevator while drunk" law and people in police elevators pulling you over to the first floor to give you a ticket.
Until they remove the ability for drivers to actually drive, they are going to be legally responsible. Uber is fine for the next 30-50 years.
To me, #1 is classic "inside the Valley" thinking - not recognizing that the world outside of the Valley is vastly different. As a simple example, as far as I know, no one is currently attempting autonomous cars in any place with more severe or varied weather conditions (e.g. snow).
In fact, Uber is. They have a research center in Pittsburgh, which has plenty of weather and terrible roads. Cars are seen regularly dodging students around CMU.
The author mentions the CMU acquisition. I think their contention is that it's not enough, because Uber doesn't have access to the kinds of vehicle telemetry data its competitors do that would be necessary to train good self-driving car models. Google, and other companies with existing advanced vehicles on the road, have thousands of cars in the wild now with suits of sensors collecting telemetry and phoning home. Uber contracts drivers who drive their own cars, so it's not benefiting in a vehicle-telemetry sense from the drives it's arranging in the way it could be and its competitors do.
I think this is probably a solvable problem, though.
Valley can start the process by promote the self driving cars only in the silicon valley + SF, similar to how AirBnB and Uber started within a small region.
Valley has better environment, friendly government/regulation support.
The next version Tesla (in 5 years) might allow the car owner to get to work and auto "uber out" the car to "earn $$$" for the own - start with just Bay Area.
Or Google can
Step 1: Build a few thousand self driving "beta" cars to shuttle only their employees to/from work in year N.
Step 2: Uber out (Guber out) those self driving cars to pickup/drop off anyone in Bay Area, in year N+1.
If the model works for Bay Area, I can see other congested cities will clone it immediately to solve their own traffic problem.
I totally agree, but that doesn't mean it's going to happen that fast, or that the public or local governments are not going to be obstacles to these initiatives going forward. Short version: I think the timeline the author has in mind is dramatically faster than it will actually happen. But hey, would love to be wrong.
Isn't Google invested in Uber? Didn't Uber order 100,000 units of Mercedes vehicles to do something with? Isn't GM partnered with Lyft to accomplish a self-driving taxi service?
I don't know, I think Uber and Lyft are well aware of the progress in self-driving vehicle technology and their actions point more towards them behind involved in the process as opposed to being left in the dust. The only group of people that are "fucked" are the current drivers whose jobs will be automated.
Let's say that's true; the folks at Uber are smart and they'll figure out self driving cars. Why does anyone think Uber will be able to maintain an advantage with that? What's to stop anyone from buying some self driving cars and opening their own service? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
Uber is in a low-margin business where the only way to really increase profit is to screw your customers or employees or both. If self driving cars are supposed to be the company's savior, good luck being the only organization on the planet that can use them. Having lived through Webvan, Groupon, Zynga, and countless others, I see disturbingly similar patterns in Uber.
You can basically just swap in any company in any industry to your comment...keeping up pace with technological changes, competitors entering your market, and maintaining a profit margin are just normal challenges that every business faces.
Let's see:
> Let's say that's true; the folks at Google are smart and they'll figure out artificial intelligence. Why does anyone think Google will be able to maintain an advantage with that? What's to stop anyone from buying artificial intelligence tools and opening their own AI service? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
> Let's say that's true; the folks at Apple are smart and they'll figure out personal computers. Why does anyone think Apple will be able to maintain an advantage with that? What's to stop anyone from buying some computer parts and opening their own computer store? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
> Let's say that's true; the folks at SpaceX are smart and they'll figure out re-usable rockets. Why does anyone think SpaceX will be able to maintain an advantage with that? What's to stop anyone from buying some re-usable rockets and opening their own space travel agency? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
Google: They have an advantage in search that I think would be hard for a new entrant to overcome (mostly because of the amount of data they've collected). I don't think their advantage is algorithmic at this point so much as lots of data + they're a verb. Uber is practically a verb at this point, but I think their data is less valuable. I guess it would be comparable to Foursquare's data, but much more of it.
Apple: The whole industry has become a race to the bottom, except for Apple which has built a phenomenal brand (also see Nike for example of a valuable brand). I don't think the Uber brand is worth all that much.
SpaceX: I think it's fair to say setting up a rocket company has significantly higher barriers to entry than building a taxi dispatching app.
But Groupon I think fits:
Let's say that's true; the folks at Groupon are smart and they'll figure out online coupons. Why does anyone think Groupon will be able to maintain an advantage with that? What's to stop anyone from hiring some web developers and opening their own daily deals website? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
In the case of Uber, I'm arguing that the barrier to entry for other companies will be buying several cars in a retail market, because various manufacturers will be selling them.
Uber does have access to car data - the cars Uber drivers use. I don't know if Uber is using this data to help adapt over the long term. But it's something they do have access right now.
By the way, robotics teams have traditionally been steeped in rules-based AI’s. Essentially enormous if-then decision logic tree. Even if they are using the most advanced self-learning algorithms, huge amounts of data are required to train, perfect, and ship such a system. Uber has access to essentially NONE of it.
Sorry but there's so much BS condensed in just these two sentences that it immediately disqualifies the whole article.
Of course robotics software isn't based on giant decision trees, wtf. Then what's a self-learning algorithm? Who says that Uber couldn't just license self-driving tech or data? And by the way the plural of AI is AIs, without the apostrophe.
Self driving cars that are beyond simply assisted driving or and overseen autopilot are probably 10-20 years away.
It would take quite a bit of additional improvement in tech and massive legislation for driverless cars to become a common thing.
I don't understand why people continue to mix self driving and driverless those are 2 separate things.
Autopilot or not I don't see most countries change the law of both hands on the wheel (unless you need to operate another system in the car in which case 1 hand might be taken off it) any time soon.
Infact with autopilot becoming more and more popular you might see actually increased enforcement of this law because It's another revenue stream for local authorities.
Also if anything Uber is the one that stems to benefit from driverless car since they won't have to pay the drivers any more.
With driverless cars I see their business model becoming something like "want to save 30% on your monthly car payments? sign up your car to Uber!", Uber can even offer financing options to people at considerably discounted rates in exchange for a mandatory opt in period.
Heck Uber can go Zip car and simply manage their own fleet when driverless cars become a true reality, it might even be easier for companies like Uber to get a driverless car before the general population, they've already built their platform and have a user base, the drivers are this point are a necessary liability for them the minute they can dump them they will.
And if the car makers would want to offer a similar service themselves well Uber can just come in as a service provider offer the car manufacturers access to it's user base and platform in exchange for a usage fee it would be even better for them since they could dump a huge chunk of their existing liability onto their partners.
As soon as there's strong evidence (from multiple credible sources) that self-driving tech reduces accidents and fatalities, the government conversation will switch from, "How shall we regulate this?" to "Oh! These are like seat belts. How shall we encourage them?" In other words, governments will speed up adoption, eventually.
The owners of the car fleets will be the car manufacturers because they're the ones who can pay the least per car. They don't have to pay any markup!
The self-driving tech will come from Google, Tesla, Apple, or the like, not Uber. See the blog post for the reasons why.
>As soon as there's strong evidence (from multiple credible sources) that self-driving tech reduces accidents and fatalities, the government conversation will switch from, "How shall we regulate this?" to "Oh! These are like seat belts. How shall we encourage them?"
Only 34 states have a primary seat belt law and that took decades of legislature to push through. Legislation takes time, if you think it would be 10 years before we have enough data on autopilot you are fooling yourself, and going from autopilot to driverless will not happen overnight.
If we take the US then again it's unlikely that the federal government would get involved on any reasonable time table, this means that the driverless car makers would have to lobby each state individually.
Until the big car makers would have a foolproof driverless car Apple and Tesla can do whatever they want they will not match the power of the auto industry lobby just look at how much problems Tesla is having over dealerships.
It doesn't matter who's making the driverless car, Uber has the user base and the platform saying that Tesla is making a driverless car is meaningless unless you would want to make the same argument that Uber should only onboard drivers that drive a Honda or a Ford today because they their cars come in blue.
If Uber survives until driverless cars really pick up (and BTW the 10-20 year window isn't only my opinion, I don't think even Tesla thinks that their car would become driverless road ready or road legal before 2025) and is not torn apart but anti-Uber legislation and sentiment then the driverless car would be the thing to save it not kill it.
1. Self driving cars are not coming fast enough, but their impact on Uber's business model is unclear. They've been investing in self-driving cars [1], but they've also been expanding into parcel delivery, which needs a human to perform the actual contact with the customer [2].
2. Like the article says, Uber indeed is collecting sensor data in Pittsburgh, to train self-driving algorithms. They may or may not be able to deliver an algorithm rivaling the quality of Google or other players. It's contradictory to say Uber has access to no car data, while pointing out that Uber is collecting car sensor data on their own.
The passengers who use Uber may not be very brand-loyal, but that assumes their business model is entirely predicated on customer fares. In truth, their business model is entirely predicated on burning VC money to establish first mover advantage and mindshare, and if they successfully pivot into profitability they will presumably alter business models. Licensing self-driving tech, charging for rush parcel deliveries, providing concierge services -- those are all viable directions in which they can pivot.
3. Being in the business of fleet maintenance is not so fun, and location-aware dispatching, load balancing, etc, requires the sort of secret sauce that, among a few others, Uber has developed. Car manufactures won't engage in this as their core business; they're far more likely to partner with some startup Uber competitor, or Uber itself. But car manufactures aren't like airplanes where the cost of a single unit is so high that it's much more economical to lease indefinitely. Many people in fact own cars -- the very, very profitable used car market is evidence of this, which unlike most new cars, actually has decent margins. While it's true that carmakers have subsidiary banks that offer financing, those instruments enable a larger portion of people to 'buy cars' than otherwise, increasing the total volume of cars delivered. Extremely widespread, mainstream car-hailing on demand would drastically drop the demand for car ownership (in terms of raw numbers) across the board, making it an awful value proposition for car manufactures.
> There’s a good reason cars start falling apart after their warranty period — They have been carefully engineered to specifically have this happen, a term called “planned obsolesce”. Manufacturers want to sell more cars, and if cars last longer, they sell way less new cars.
My experiences are vastly different. Neither my Mercedes nor my wife's Honda have "fallen apart" after the warranty period. MB is 216K miles and 18 years old and Honda is ~170K miles and 11 years old. Both have averaged under $200 in parts per year, mostly brakes and tires.
Indeed. Obsolescence does occur with cars -- just like most consumer items -- but it's either:
- technical or functional obsolescence (eg. no rearview backup camera, carburetors replaced with fuel injection), or
- style obsolescence (eg. station wagons, quirky headlamps) because styling is a subjective fashion statement. While this is arguably a sub-type of planned obsolescence, it affects every single object that has visual presence, from clothes to smartphones to household appliances.
The kind of 'intentionally crippled engineering designed to fail after the warranty period' the author proposes does not exist, because new cars are perfectly capable of getting themselves sold on their merits, and not because all old cars have been intentionally crippled. In fact, selling secondhand cars is quite often profitable, while selling new cars, outside of the luxury/SUV segment, has notoriously low margins.
Planned obsolesce was discussed by the late Alvin Toffler in his book 'Future Shock' in 1970. A great read! I believe it often occurs in the part manufacturing such as cutting costs. Let's replace this metal gear with a plastic gear and save $x.00 per unit. But every 'person on the street' knows that a metal gear will outlast a cheap plastic gear. Now we've just engineerd in the obsolesce due to bean-counter cost-cutting.
But think of Apple (or Dell of you prefer) and the iPhone or iMac. What is the lifespan of the avg. product? 3 yrs? 5 yrs? 10 yrs? 20 yrs? Computational devices are pretty much obsolete in 5 years (ten years for sure if you feel like arguing about 5 yrs.) as new technology is developed and surpasses the old. We now have to design something that works well for that expected time period and keep mfg. costs low.
Even if you accept the author's assumption that driverless cars are very close, his argument rests on the following premise:
Every single car manufacturer will launch their own uber-like app, market their app directly to the end-user, and maintain a fleet of taxis nationwide, in order to compete with Uber.
I find the above premise hard to believe. Why would a company like Nissan take on the massive marketing/logistics challenges required to run a nationwide taxi service? Do we really think the market can support 10+ car manufacturers each running their own Uber-like app?
It's so much more likely that each car manufacturer will simply continue trying to sell as much as possible, and let the buyers decide how they want to make use of the car. Even if they do decide to "put their inventory cars to work," they will do so by loaning the cars out to an aggregator like Uber, who will then use these cars in order to run a taxi service. The idea that every car manufacturer will decide to cut Uber out, and enter the taxi business directly, just doesn't sound plausible nor smart.
What I hope will happen is a Red Hat-style company emerging that will maintain an open-source car sharing app/webservice.
They would sell support or host each city's car-sharing service for a fee, or just let people set up their own.
This way, cities could run the service at cost (or maybe even subsidize it) and drivers would take home more. Or they could use the revenue for the city instead of having it leave the local economy.
Yeah, the car companies are also fucked. Prepare for massive consolidation on that front. One or perhaps two or three car manufacturers will swallow the whole market as demand craters. (We all own our own cars because it's a necessity at this point. It won't stay that way in a driverless future. Nor will cars be the status signifier they currently are.)
I actually think the form factor most cars come in will become obsolete-- I think they'll be mostly vans and buses to cope with economies of scale on the rider front and to handle rush hour surges. Who makes the bus you last rode on? Do you know? I know I don't, nor do I care.
I mean, Uber is totally fucked if this comes down the pipe and they're not running things, that's undeniable. Their value proposition is running this two sided marketplace that's going to turn into a one sided marketplace. And yeah, putting together a mobile app is a completely different universe than becoming a major auto manufacturer, or a robotics manufacturer of unprecedented scale. (Or in reality, both.)
I think we have some thoughts in common; however, I foresee a future where people pay a monthly fee for a individually-sized self-driving car (maybe on the order of a smart car) to pick them up, take them to work, and drop them off. Then do the reverse later in the day. This assumes the future requires people to actually travel to work.
Decoupling car use with car ownership also makes short-ranged EVs much more viable. Once the car drops you off, it heads back to the mother ship to recharge. A different car most likely would pick you up later anyway.
Why I think driverless cars will beat driverless buses is privacy and flexibility. A bus still has to run on a schedule, driver or no, and let's face it, sharing space with people has its downsides.
Hypothetically, Uber could still co-ordinate rides as a spot market, or get into the "car subscription" / "transportation as a service" industry. But the first movers in that space will likely have to own their fleets.
Demand may go down, but I don't see the car market necessarily consolidating (or going away) like you're implying here.
Technology and economic growth often increases the number of "single person/family" consumer products, rather than decreases them. The "sharing economy" isn't necessarily the default goal that everyone progresses towards for all products (although it may seem like that at times, living in a tech bubble), and "cars" have historically been one of those products that people buy more of, rather than less, as time goes on. Whether this will hold true in the future, who knows? But I don't see it as a given that the market will drastically decrease.
People like customizing their cars, storing things in them, the familiarity of them, and treating them as a "home away from home" even if it's just going to the grocery store. My husband even gets anxious when he's away from home without his car (if, say, we're on vacation or we took a cab somewhere. He never takes public transportation). It's a guaranteed safe place, a way back home, and something he can count on. Sure, you may feel very differently, but you are not every person.
And when you say: "Who makes the bus you last rode on?" Well, I did't have the dubious pleasure of experiencing a city bus until I was 22. Sure, now I ride one every day, but I know plenty of adults who have never been on a bus, much less have convenient access to a route. And while driverless vehicles may facilitate the expansion of public transportation to increasingly rural areas, at some point, you'll hit an absolute mathematical wall at which "the cost of the car transporting itself from customer to customer" will be greater than "the cost savings/reduced environmental impact of sharing the car"
Long story short: I think your assessment of the market is a little premature and somewhat egocentric. Sure, it could absolutely happen that way, but I could also see driverless cars becoming an even more "single user/personal consumer product" than they are today. It's hard to say.
Cars are expensive. Cars are also mostly sitting idle. These are objective facts. For most of the country, the only bill bigger than their combined car payment/maintenance is their mortgage. That's a lot of money.
I think a combination school-bus-for-adults/carpool would have a lot of traction if you could automate it. How much is that gonna cut into the car market? Remains to be seen. But not nothing, for sure. And moving cars into a service model is, in my opinion, gonna make a lot of the pricing more transparent. You want an SUV, you'll have to pay for it per ride. You want your own private car, pay per ride. I think consumers, when faced with spending the extra buck on an a la carte basis, are going to start looking into cheaper options.
Rush hour surges are gonna be a big strain on the system, and I really think that school bus model is going to take off. Everybody going more or less in the same direction at the same time, get on the bus and save a ton of money.
Maybe it just turns multiple car families (the default) into single car families. But even that would be a seismic shift in the auto industry.
An analysis of a business today doesn't mean much for the state of a business tomorrow. Companies and situations change. So, for the people happily working for Uber, probably don't quit your job as the author suggests.
> There’s a good reason cars start falling apart after their warranty period — They have been carefully engineered to specifically have this happen, a term called “planned obsolesce”.
I don't believe this, and I don't believe the author knows anything about cars, from reading this.
64 comments
[ 0.18 ms ] story [ 137 ms ] threadSomething about the Ninety-ninety rule seems appropriate here:
https://en.wikipedia.org/wiki/Ninety-ninety_rule
I think self driving cars are much closer to 10 years out than the "today" that the author asserts, but I think everyone agree's that we've reached the point where they are coming no matter what.
To give the author credit I went into reading this thinking he was absurd but he does make a good point, some of the points I think are over sold, like planned obsolescence but I think his thesis is sound.
One point I'll make on Uber's behalf......
Two of my all time favorite books are:
https://www.amazon.com/Startup-Silicon-Adventure-Jerry-Kapla...
This history of Go computing.
and Ben Horowitz The hard thing about hard things.
https://www.amazon.com/Hard-Thing-About-Things-Building/dp/0...
Both do a very good job of bringing the reader into the chaotic environment that occurs when startups are in trouble and both have a very similar message. When startups get in trouble having very powerful investors and mentors can make a huge difference.
Uber has some very influential backers. See:
https://www.crunchbase.com/organization/uber/investors
If/When times get tough, they have people who have a vested interest in seeing them succeed. People who can get them talking to the correct people at the car manufacturers to convince them to use Uber's platform over an internally created one.
If self driving laws drag on many years before they are settled then Uber may be fucked in the end but there isn't any reason why they can't be a cash printing machine over the next 10- 20 years while self driving cars replace humans.
As to car companies cutting out uber, there is this.....
http://qz.com/688003/ubers-self-driving-cars-are-on-the-road...
Self-driving cars are very certainly the future of Uber, and once self-driving tech is mass-available Uber will have the same accessibility to it that Google or Tesla or whomever else does. At that point, they're competing on price and service quality, just like they do with Lyft and Sidecar right now.
I don't see self-driving tech becoming a major sticking point, because it's not like manufacturers are going to only be manufacturing self-driving cars for their own usage once the regulatory hurdles are cleared.
See some nice photos here: http://qz.com/688003/ubers-self-driving-cars-are-on-the-road...
At least in some countries and cities getting a taxi at the right times have been a hassle, and somewhat expensive. So even if, like Uber, you can expand the marked and steal customer from traditional taxies, how much money is there really to be made? Getting the profit that currently falls to the drivers seems like a logical next step, if they want to increase their own profit.
I'm skeptical that manufacturers would be that nimble to capitalize on the inevitable. That's also not their business model. Building out such an effort in such a large organization would be a challenge. It would be much cheaper for car manufacturers to partner with or buy a company like Uber.
In regard to Uber not having data to create the machine learning models needed for self-driving cards, they could buy that data or start building it up themselves. Google doesn't have a monopoly on strapping cameras and sensors to cars. And Uber certainly has enough capital required. Besides, this is a core strategy for Uber while it's still a moon-shot for Google.
While car manufacturers are hardly the nimblest they're not brain dead and are already exploring related stuff (example: https://www.washingtonpost.com/news/dr-gridlock/wp/2016/07/1...)
If we reach a day where there is some company, call them company U which has a successful dispatch app business, and another company, call them company C, which has just successfully demonstrated (and passed regulatory hurdles for) a self driving car, what do you think will be easier and cheaper and quicker? For U to develop/acquire/license self-driving car technology or for C to develop/acquire/license a dispatch app?
C could probably pay very little to license a dispatch app from any of the many vendors. Even if it's not popular before, it will be once it's the only dispatch app that dispatches SELF DRIVING CARS! But if U tried to license self driving from C, U would probably have very little leverage and end up signing a contract where U runs the show but ships most of the revenues to C. C would be like Microsoft in the 80s/90s, and U would be like a generic IBM PC compatible computer vendor.
One crucial thing to remember, when considering the threat Uber faces from traditional car manufacturers, is that Uber is a software company and they are not. It has a huge advantage in product development, and stands just as good a chance as they do of coming out with self-driving cars for ride-sharing.
Anyone can hire programmers, children can ship products, MLB created a Netflix-like platform to offer media companies as a service, Walmart have made important contributions to open source etc.
The number of steps between Uber's competitors becoming as competent as Uber at software is basically "1. decide to".
The software engineers I have known and tried to recruit tend to want to work for companies that are more like Google and less like General Motors.
The statement Anyone can hire programmers, children can ship products is manifestly untrue. It's hard to get the right people, and it's equally hard to coordinate all their activities to get a successful product over the finish line.
The companies that have healthy cultures of software engineering and digital product management and design are way, way ahead of everyone else. And those companies are in the minority.
http://fortune.com/2016/03/11/gm-buying-self-driving-tech-st...
Self-driving != Driverless
Self-driving cars is largely a technology problem, and one that we have made great advances in the last 5-10 years. Driverless cars are a legal/social/political problem that we have barely started to look at yet. Self-driving is a driving tool, like cruise-control, but the human driver is still ultimately responsible for the vehicle. Driverless on the other hand places the responsibility for the vehicle, and it's occupants, onto the vehicle itself. Driverless implies that you should be able to use it while you yourself are not capable of operating a vehicle. That is a massive step, and it is something that has never been done before. All other forms of transportation, despite advanced automation, still defers to a human in the end. I would say that true driverless cars is still a significant amount of time away, despite our advances in self-driving tech.
I don't think you are taking into account UberPool. HN comments almost never do, perhaps because programmers are generally upper income and do not feel a need to trade money for inconvenience, but UberX is basically a legacy product and now presented as more of an upgrade if you're in a hurry, rather than the default mode.
UberPool is everything. A network of cars buzzing around the city dynamically routing from one task (transport passenger, transport package, deliver food, etc) to the next, and each task is cheaper because it can share ride segments with other customers.
That is where the winner-take-all argument comes from - the bigger the network, the more efficiently Uber can pack multiple tasks into the same driver-hours, with less delay (more jobs means the car doesn't have to go as far out of its way to share capacity).
Uber is more than dispatch - it's quality control, reputation, dispute resolution, capacity planning, payment escrow, (soon to be) routing (it bought a few mapping companies), commercial insurance (try buying that on your own), and increasingly, the efficiency provided by combining multiple tasks in each car.
When people call Uber a taxi service - when is the last time a taxi offered you a cheaper ride if you agreed to share the car for 5 blocks with a guy 10 blocks down the road?
Disclaimer: I used to work there. Throwaway because I'd rather keep identifying information off my personal account.
Opinion is my own, I do not speak for Uber, etc.
I still don't think that most of those checkboxes is terribly important to either drivers or riders. If I were an Uber competitor, I'd thank Uber for screening the drivers for me. Now I can offer them fares at half the price Uber can offer them for. You have a point about delivery, but in the end that's just another fare. There will certainly be good dispatch networks, and Uber may be one of them, but none of them will be able to maintain high margins, certainly not margins that would support Uber's valuation.
Music cos. were supposed to do something similar to iTunes etc. Didn't happen.
> Cars will appear when ordered through the app. They will drive off and disappear when you step out of the car. When running low on fuel, cars will coordinate trips to tucked away re-charging depots to be smartly refilled with electricity.
You mean in about 20 years time?
> Then car manufacturers will drastically start reducing the cost of operating these vehicles.
With magic?
Come to Boston.
Search up "The horse manure problem:
"...In 1894, the Times of London estimated that by 1950 every street in the city would be buried nine feet deep in horse manure. One New York prognosticator of the 1890s concluded that by 1930 the horse droppings would rise to Manhattan’s third-story windows. ..."
https://nofrakkingconsensus.com/2011/03/29/the-horse-manure-...
Until they remove the ability for drivers to actually drive, they are going to be legally responsible. Uber is fine for the next 30-50 years.
[1]: http://www.theverge.com/transportation/2015/5/19/8622831/ube...
I think this is probably a solvable problem, though.
Valley has better environment, friendly government/regulation support.
The next version Tesla (in 5 years) might allow the car owner to get to work and auto "uber out" the car to "earn $$$" for the own - start with just Bay Area.
Or Google can
Step 1: Build a few thousand self driving "beta" cars to shuttle only their employees to/from work in year N.
Step 2: Uber out (Guber out) those self driving cars to pickup/drop off anyone in Bay Area, in year N+1.
If the model works for Bay Area, I can see other congested cities will clone it immediately to solve their own traffic problem.
I don't know, I think Uber and Lyft are well aware of the progress in self-driving vehicle technology and their actions point more towards them behind involved in the process as opposed to being left in the dust. The only group of people that are "fucked" are the current drivers whose jobs will be automated.
Uber is in a low-margin business where the only way to really increase profit is to screw your customers or employees or both. If self driving cars are supposed to be the company's savior, good luck being the only organization on the planet that can use them. Having lived through Webvan, Groupon, Zynga, and countless others, I see disturbingly similar patterns in Uber.
Let's see:
> Let's say that's true; the folks at Google are smart and they'll figure out artificial intelligence. Why does anyone think Google will be able to maintain an advantage with that? What's to stop anyone from buying artificial intelligence tools and opening their own AI service? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
> Let's say that's true; the folks at Apple are smart and they'll figure out personal computers. Why does anyone think Apple will be able to maintain an advantage with that? What's to stop anyone from buying some computer parts and opening their own computer store? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
> Let's say that's true; the folks at SpaceX are smart and they'll figure out re-usable rockets. Why does anyone think SpaceX will be able to maintain an advantage with that? What's to stop anyone from buying some re-usable rockets and opening their own space travel agency? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
Apple: The whole industry has become a race to the bottom, except for Apple which has built a phenomenal brand (also see Nike for example of a valuable brand). I don't think the Uber brand is worth all that much.
SpaceX: I think it's fair to say setting up a rocket company has significantly higher barriers to entry than building a taxi dispatching app.
But Groupon I think fits:
Let's say that's true; the folks at Groupon are smart and they'll figure out online coupons. Why does anyone think Groupon will be able to maintain an advantage with that? What's to stop anyone from hiring some web developers and opening their own daily deals website? How does the whole industry not become a race to the bottom? Hell, given the price wars going on now, I'd argue it already has.
In the case of Uber, I'm arguing that the barrier to entry for other companies will be buying several cars in a retail market, because various manufacturers will be selling them.
Sorry but there's so much BS condensed in just these two sentences that it immediately disqualifies the whole article.
Of course robotics software isn't based on giant decision trees, wtf. Then what's a self-learning algorithm? Who says that Uber couldn't just license self-driving tech or data? And by the way the plural of AI is AIs, without the apostrophe.
Also if anything Uber is the one that stems to benefit from driverless car since they won't have to pay the drivers any more. With driverless cars I see their business model becoming something like "want to save 30% on your monthly car payments? sign up your car to Uber!", Uber can even offer financing options to people at considerably discounted rates in exchange for a mandatory opt in period. Heck Uber can go Zip car and simply manage their own fleet when driverless cars become a true reality, it might even be easier for companies like Uber to get a driverless car before the general population, they've already built their platform and have a user base, the drivers are this point are a necessary liability for them the minute they can dump them they will. And if the car makers would want to offer a similar service themselves well Uber can just come in as a service provider offer the car manufacturers access to it's user base and platform in exchange for a usage fee it would be even better for them since they could dump a huge chunk of their existing liability onto their partners.
The owners of the car fleets will be the car manufacturers because they're the ones who can pay the least per car. They don't have to pay any markup!
The self-driving tech will come from Google, Tesla, Apple, or the like, not Uber. See the blog post for the reasons why.
Only 34 states have a primary seat belt law and that took decades of legislature to push through. Legislation takes time, if you think it would be 10 years before we have enough data on autopilot you are fooling yourself, and going from autopilot to driverless will not happen overnight. If we take the US then again it's unlikely that the federal government would get involved on any reasonable time table, this means that the driverless car makers would have to lobby each state individually. Until the big car makers would have a foolproof driverless car Apple and Tesla can do whatever they want they will not match the power of the auto industry lobby just look at how much problems Tesla is having over dealerships.
It doesn't matter who's making the driverless car, Uber has the user base and the platform saying that Tesla is making a driverless car is meaningless unless you would want to make the same argument that Uber should only onboard drivers that drive a Honda or a Ford today because they their cars come in blue.
If Uber survives until driverless cars really pick up (and BTW the 10-20 year window isn't only my opinion, I don't think even Tesla thinks that their car would become driverless road ready or road legal before 2025) and is not torn apart but anti-Uber legislation and sentiment then the driverless car would be the thing to save it not kill it.
Yes, and Uber is working on them. On HN as the same time as this article: http://qz.com/688003/ubers-self-driving-cars-are-on-the-road...
2. Like the article says, Uber indeed is collecting sensor data in Pittsburgh, to train self-driving algorithms. They may or may not be able to deliver an algorithm rivaling the quality of Google or other players. It's contradictory to say Uber has access to no car data, while pointing out that Uber is collecting car sensor data on their own.
The passengers who use Uber may not be very brand-loyal, but that assumes their business model is entirely predicated on customer fares. In truth, their business model is entirely predicated on burning VC money to establish first mover advantage and mindshare, and if they successfully pivot into profitability they will presumably alter business models. Licensing self-driving tech, charging for rush parcel deliveries, providing concierge services -- those are all viable directions in which they can pivot.
3. Being in the business of fleet maintenance is not so fun, and location-aware dispatching, load balancing, etc, requires the sort of secret sauce that, among a few others, Uber has developed. Car manufactures won't engage in this as their core business; they're far more likely to partner with some startup Uber competitor, or Uber itself. But car manufactures aren't like airplanes where the cost of a single unit is so high that it's much more economical to lease indefinitely. Many people in fact own cars -- the very, very profitable used car market is evidence of this, which unlike most new cars, actually has decent margins. While it's true that carmakers have subsidiary banks that offer financing, those instruments enable a larger portion of people to 'buy cars' than otherwise, increasing the total volume of cars delivered. Extremely widespread, mainstream car-hailing on demand would drastically drop the demand for car ownership (in terms of raw numbers) across the board, making it an awful value proposition for car manufactures.
[1] https://newsroom.uber.com/us-pennsylvania/new-wheels/
[2] https://rush.uber.com/how-it-works
My experiences are vastly different. Neither my Mercedes nor my wife's Honda have "fallen apart" after the warranty period. MB is 216K miles and 18 years old and Honda is ~170K miles and 11 years old. Both have averaged under $200 in parts per year, mostly brakes and tires.
- technical or functional obsolescence (eg. no rearview backup camera, carburetors replaced with fuel injection), or
- style obsolescence (eg. station wagons, quirky headlamps) because styling is a subjective fashion statement. While this is arguably a sub-type of planned obsolescence, it affects every single object that has visual presence, from clothes to smartphones to household appliances.
The kind of 'intentionally crippled engineering designed to fail after the warranty period' the author proposes does not exist, because new cars are perfectly capable of getting themselves sold on their merits, and not because all old cars have been intentionally crippled. In fact, selling secondhand cars is quite often profitable, while selling new cars, outside of the luxury/SUV segment, has notoriously low margins.
But think of Apple (or Dell of you prefer) and the iPhone or iMac. What is the lifespan of the avg. product? 3 yrs? 5 yrs? 10 yrs? 20 yrs? Computational devices are pretty much obsolete in 5 years (ten years for sure if you feel like arguing about 5 yrs.) as new technology is developed and surpasses the old. We now have to design something that works well for that expected time period and keep mfg. costs low.
Every single car manufacturer will launch their own uber-like app, market their app directly to the end-user, and maintain a fleet of taxis nationwide, in order to compete with Uber.
I find the above premise hard to believe. Why would a company like Nissan take on the massive marketing/logistics challenges required to run a nationwide taxi service? Do we really think the market can support 10+ car manufacturers each running their own Uber-like app?
It's so much more likely that each car manufacturer will simply continue trying to sell as much as possible, and let the buyers decide how they want to make use of the car. Even if they do decide to "put their inventory cars to work," they will do so by loaning the cars out to an aggregator like Uber, who will then use these cars in order to run a taxi service. The idea that every car manufacturer will decide to cut Uber out, and enter the taxi business directly, just doesn't sound plausible nor smart.
They would sell support or host each city's car-sharing service for a fee, or just let people set up their own.
This way, cities could run the service at cost (or maybe even subsidize it) and drivers would take home more. Or they could use the revenue for the city instead of having it leave the local economy.
I actually think the form factor most cars come in will become obsolete-- I think they'll be mostly vans and buses to cope with economies of scale on the rider front and to handle rush hour surges. Who makes the bus you last rode on? Do you know? I know I don't, nor do I care.
I mean, Uber is totally fucked if this comes down the pipe and they're not running things, that's undeniable. Their value proposition is running this two sided marketplace that's going to turn into a one sided marketplace. And yeah, putting together a mobile app is a completely different universe than becoming a major auto manufacturer, or a robotics manufacturer of unprecedented scale. (Or in reality, both.)
Decoupling car use with car ownership also makes short-ranged EVs much more viable. Once the car drops you off, it heads back to the mother ship to recharge. A different car most likely would pick you up later anyway.
Why I think driverless cars will beat driverless buses is privacy and flexibility. A bus still has to run on a schedule, driver or no, and let's face it, sharing space with people has its downsides.
Hypothetically, Uber could still co-ordinate rides as a spot market, or get into the "car subscription" / "transportation as a service" industry. But the first movers in that space will likely have to own their fleets.
Technology and economic growth often increases the number of "single person/family" consumer products, rather than decreases them. The "sharing economy" isn't necessarily the default goal that everyone progresses towards for all products (although it may seem like that at times, living in a tech bubble), and "cars" have historically been one of those products that people buy more of, rather than less, as time goes on. Whether this will hold true in the future, who knows? But I don't see it as a given that the market will drastically decrease.
People like customizing their cars, storing things in them, the familiarity of them, and treating them as a "home away from home" even if it's just going to the grocery store. My husband even gets anxious when he's away from home without his car (if, say, we're on vacation or we took a cab somewhere. He never takes public transportation). It's a guaranteed safe place, a way back home, and something he can count on. Sure, you may feel very differently, but you are not every person.
And when you say: "Who makes the bus you last rode on?" Well, I did't have the dubious pleasure of experiencing a city bus until I was 22. Sure, now I ride one every day, but I know plenty of adults who have never been on a bus, much less have convenient access to a route. And while driverless vehicles may facilitate the expansion of public transportation to increasingly rural areas, at some point, you'll hit an absolute mathematical wall at which "the cost of the car transporting itself from customer to customer" will be greater than "the cost savings/reduced environmental impact of sharing the car"
Long story short: I think your assessment of the market is a little premature and somewhat egocentric. Sure, it could absolutely happen that way, but I could also see driverless cars becoming an even more "single user/personal consumer product" than they are today. It's hard to say.
I think a combination school-bus-for-adults/carpool would have a lot of traction if you could automate it. How much is that gonna cut into the car market? Remains to be seen. But not nothing, for sure. And moving cars into a service model is, in my opinion, gonna make a lot of the pricing more transparent. You want an SUV, you'll have to pay for it per ride. You want your own private car, pay per ride. I think consumers, when faced with spending the extra buck on an a la carte basis, are going to start looking into cheaper options.
Rush hour surges are gonna be a big strain on the system, and I really think that school bus model is going to take off. Everybody going more or less in the same direction at the same time, get on the bus and save a ton of money.
Maybe it just turns multiple car families (the default) into single car families. But even that would be a seismic shift in the auto industry.
I don't believe this, and I don't believe the author knows anything about cars, from reading this.