Comment by emdowling | Hacker News Reader

Comment by emdowling | Hacker News Reader

Parent Full thread

It's fascinating to see Google and (purportedly) Apple move in a similar direction away from developing their own actual cars and instead building the software platform. It makes a lot of sense for Google, who have always struggled with selling products direct to consumers. It is more surprising for Apple, but makes sense if you look at how they can add the most value.

I'm looking forward to seeing how this plays out. Are cars going to be (even more) commoditised at the low end and differentiated by the platform on top of it (Apple, Google, Tesla) in the same way phone carriers were commoditised by the iPhone? My gut is that they won't (industrial design is a huge part of choosing a car manufacturer), so the dynamics between platform and carrier (Toyota, Fiat Chrysler, etc) are going to be fascinating to see unfold. The traditional car industry moves at such a glacial pace, yet you have two tech titans ceding a lot of control to them.

Of course, you then have the "apps" that will run on top of them, like Uber. How they interplay with it all is another dimension too. Tesla has fired the first shot here by restricting usage of non-Tesla car pooling services. It is a worrying thought to think that buying a BMW might mean I'm locked into only renting out my car through Apple-approved car riding services (and while I can't see Apple creating their own car pooling service, never say never).

Tesla seems to be the only player doing the entire stack from top to bottom. Very reminiscent of early Apple.

I wonder if a fully automated car might actually be a different product category than a "regular" car. I could actually see a decent market for (relatively ugly) self driving cars that are mostly seen as a means to get from A to B. In fact I'm not sure if the ownership model is right for this category. I could see renting/leasing these vehicles on a monthly base as a service and the competition being buying a monthly commute ticket for the train or something. I think my train ticket is in the vicinity of 180 Euro/month and all inclusive (insurance, maintenance) leases for midelevel cars are in the 300 Euro/month+gas range but prices go up a lot if you want to lease for <24 month. I could see paying 500-600/month for a self-driving car that I can also use for grocery shopping and other random stuff.

I can see that during a relatively short transition period but hope it won't remain that way. Once all cars are autonomous we can do incredible things. For example traffic lights to regulate traffic between cars will be obsolete. Cars can just adjust their speed way before the intersection so that they just miss each other at the intersection. Likewise we can avoid congestion by cars registering for their route in advance. We could raise the speed limits on freeways and other roads that only have cars on them or even remove them entirely since the cars will know better what their limits are than some blanket regulation. It also will be much safer, since self driving cars are likely to quickly become safer drivers, but we will also remove erratic behavior that humans display. For example there won't be no more people who cannot zip properly (currently the majority of US Americans), no more cars not speeding up properly on the freeway on-ramp and then slowing everyone down on the freeway, no more weird stopping and slow driving by drivers who are lost. This will make the overall traffic situation much more predictable and safer.

Your description of how things would work with no traffic lights would scare me to death. One errant car (mechanical failure, software bug or hacked) could cause multiple fatalities. Systems need to be resistant to disaster from perturbations like these.

"One errant car" already kills scores of people today. It really isn't that much different with autonomous vehicles. A stop light isn't going to prevent a distracted driver from entering an intersection. Mechanical or software problems already cause fatalities today.

That said I would still prefer to have a hands-on option with my car. Driving around an unfamiliar neighborhood looking for someplace to eat or maybe house hunting are examples of when you can't just pick a destination ahead of time. You want to drive slowly and decide where to go turn-by-turn.

Hands-on? You'll just tell your car, like a passenger would tell a hired driver.

Not to mention the fact that non-autonomous cars will continue to exist en-masse. People value the experience of driving a car (especially sports cars or classic cars, but even just the every day sedan) and I don't think that the advent of autonomous cars will eliminate that for a very, very long time.

Unfortunately you are probably right with that, especially in the US where people value individual freedom over pretty much anything else. I expect Japan and China to be first with having meat driven cars and the US and Germany to ban them last. Not banning them will come with a tremendous economic cost.

Sometimes stop lights are there so that people can walk across the street.

That's why I said "traffic lights to regulate traffic between cars"

Sorry I'm being a bit glib. What I'm poking at here is what I feel to be undue optimism about self driving vehicles. It almost appears as if people are describing the Beetle from Fahrenheit 451, with no irony intended.

What I'm getting at is that there seems to be a lack of critical thinking about the potential negatives of self driving cars. We know for example that transit by car is extremely price elastic, and it is easy to infer that a self driving car imposes a very low cost on drivers, so we should expect trips by car to increase if cars are self driving. Is this good or bad? Its hard to say. But we do know that more car trips increase demand for roads. Is this good or bad? If all self driving cars are going to and from similar locations, eg exurbs to job centers, will congestion decrease? If one no longer suffers the penalty of having to drive in congestion, will more people voluntarily sit in traffic, thereby saturating infrastructure?

I totally agree that those are all valid concerns.However, I am confident we will get those things figured out. Maybe the congestion won't be that bad because autonmous driving services might encourage ride sharing. I for one already sometimes use Lyft Ride for longer trips. Maybe we can increase incentives for that? Communities could create a fee for single passenger rides. In a way car pool lanes are similar to that and already might accomplish something like that. On the less punitive side I can see something like convoys decrease fuel consumption. My university in Paderborn, Germany was working on something like that for self driving trains over ten years ago. Of course it's much harder with cars, but we also have come a long way.

All was said is that traffic will be more predictable and safer, not that it won't have any possible negative effects.

>market for (relatively ugly) self driving cars that are mostly seen as a means to get from A to B.

You mean like a Volvo ;)

Perhaps not the prettiest car ever made, but looks OK to me: https://www.media.volvocars.com/image/low/194846/1_1/5

>> Are cars going to be (even more) commoditised

Self driving cars as a service are much better deal for the consumer. And they'll be more commoditized[1], same as taxis are, from the consumer perspective. On the other hand, building the car with the highest reliability , and the lowest cost of operation , is a hard challenge.

But: if electric cars are going to win, cars may become mostly batteries on wheels, so a lot of power will go to the battery companies, maybe even to the point they'll build the cars(like Samsung is working on).

And about Tesla: a lot of it's value is about brand. cars as a service kill branding(if everybody transports with a Tesla,it's nothing special) . Maybe that's one reason why they don't want their vehicle as a service.

[1]One thing i'm curious about : is taxi sharing realistic in a world of self-driven cars ? one guess: to make it viable, someone would have to build cars that are both shareable, but offer privacy for each of the passengers. And actually, this doesn't require self-driving to offer great value, so i wonder why aren't we seeing work on this?

Sharing isn't viable. I don't want my 20 minute taxi/CaaS ride to become 30+ minutes as we drive off my route to pick up another rider (and wait for them to come downstairs, pick their way across the slush, close their umbrella, and FINALLY get in the god damn cab)

> Sharing isn't viable. I don't want my 20 minute taxi/CaaS ride to become 30+ minutes as we drive off my route to pick up another rider

I drive for Uber part time and the majority of riders do choose the POOL option so I think you're wrong here.

Uber says Pool is at 20+% of the rides. Huh, I stand partially corrected!

I think most people would prefer no delays and a private ride. But the savings of pooling is too large for many to ignore.

So you mean, most people absolutely do prefer no delays and a private ride, since the other ~80% of Uber riders don't do the POOL option.

The market exists for both though. It's a smart move to offer this option. It's more energy economical and fiscally responsible. There's certainly even people who have the time but not the money.

Pool is not an option everywhere (even in the same city). In areas where it is available it represents a majority of my trip requests.

a shared ride has to have at least two occupants. With 4 passengers on average, 20% shared rides would mean 50% of user.

The Uber Pool service doesn't require you to actually be matched. If there doesn't happen to be another rider going in the same direction, you still get the cheaper fare and it's counted as an Uber Pool ride.

Pool also appears to be the default option now, so I on a recent trip accidentally got a pool a couple times. I didn't know the first time, and forgot the second.

Second time was also a 'pool' with no additional riders, so it was a happy accident for a cheaper ride. Perhaps people bet on that as well?

If we're imagining a pleasant future where self-driving cars have solved congestion, isn't some kind of car pooling going to be necessary?

It's fundamentally silly for millions of commuters to each travel in their own car, and robot cars aren't going to fix that. Buses make massively more efficient use of road space (when they're reasonably full, anyway).

Robot cars should reduce accidents, and parking problems will go away, but they won't magically reduce congestion on commuter routes. But car-pooling will.

Valid point, car pooling would definitely be a good thing. And I think people would be more willing if it was more efficient than it is. Uber Pool does not seem to match people very well. Scheduled rides could probably be very optimized, so its two minutes out of the way, for each person, and no one really minds. Also optimizations like, if the rider isn't there yet, don't even go for them and pass them on to the next car. Definitely plausible, but a pain in current formats.

You just described a bus.

Something like Uber for buses would be great!

Imagine an existing excellent bus network, like London's, but with completely dynamic routes, and a variety of bus sizes suitable for different times of day. Open the app and tell it where you're going, and it tells you which bus stop and bus number you need.

Uber tried this in Toronto already (called Uber Hop). I believe they had to stop as it is illegal to compete with the municipal bus system eyeroll.

https://newsroom.uber.com/canada/meet-uberhop/

Not quite. Buses have fixed static routes. I described dynamic routes optimized to minimize delays for all users, while being picked up and dropped off at the exact address they want to be at, rather than some static bus stop.

> If we're imagining a pleasant future where self-driving cars have solved congestion, isn't some kind of car pooling going to be necessary

Maybe, maybe not. Small one-person "pod" cars that link up into road trains automatically are also a possibility, and have basically the same benefit for congestion as sharing without actually sharing the same pod.

That could work, but I'd bet against it! In the medium term at least.

It would be a big R&D problem to make it work at all. Even if it worked, is a train like that really going to be as efficient as a bus? And finally, even if those obstacles are overcome, it seems like everyone has to buy the same model (or at least compatible models, and who's going to make that happen?) to get decent economies of scale.

On the other hand -- in-flight refueling is a kind of similar problem that sounds similarly unlikely, but actually works!

On the other other hand, that's a very specialized (and I'm guessing very expensive) system that isn't used by commercial airlines. I think pod-trains will go roughly the same way.

Though, with a self driving car sharing would not necessarily mean you both need to use it at the same time. The car could pick you up, drive your 20 minutes, then drive to the other person on its own and drive their 20 minutes. "Serial sharing" not "parallel"

> someone would have to build cars that are both shareable, but offer privacy for each of the passengers

The type of sharing you are writing about is not what the comment I was replying to was writing about.

I agree FWIW and I think you'll see CaaS co-operatives start up.

> I think you'll see CaaS co-operatives start up

That's assuming that autonomous cars as a product (rather than a service) enter the free market at all.

Most of people outside the 5% value cash over times. I've seen a lot of people begin to use UberPool in India who would have earlier commuted by partly open and definitely unsafe motor rickshaws. With pooling, they get an air conditioned car ride.

In Honiara, moderately well off people will own a car and then lease* it to a taxi driver so that it can make them money throughout the day. It works out really well if you need to get your kids to school, as the taxi driver will take them or pick them up without disrupting your work commute.

I think it depends where you live. In a dense city like New York existing pool services work very well indeed - the diversion to pick up is usually brief, and the savings considerable.

Whether Apple pursue the hardware route remains to be seen. They only changed their strategy, probably to develop the core self-driving technology first then worry about the design of the car later.

Reply on news.ycombinator.com