The thing the will benefit driverless vehicles is the increasing poverty of the middle class. As wages continue to stagnate, more and more people will be unable to handle the capital costs of new cars.
Nobody will ever choose driverless Uber-like cars for a long time. Paying by the drink (ie paying the $0.60-0.80 + margin that it costs to operate a car) will never be a good deal for the consumer. People will rent because they need to.
The current situation with Uber and Lyft is an exception. Uber has revealed itself to be a criminal enterprise and both companies rely on VC subsidy and pushing liability to operators to survive. That’s not a model that will stand the test of time.
> Paying by the drink (ie paying the $0.60-0.80 + margin that it costs to operate a car) will never be a good deal for the consumer.
The economics of this are the opposite. Having cars sit unused in driveways and parking spots, and be maintained by non-experts, is highly inefficient. Just like cloud infrastructure, it is more efficient to centralize and rent out capacity as needed.
One difference from centralized computing is that a lot of the costs associated with cars are distance-based rather than time-based. If I only drive a car a few thousand miles a year and have it sitting in my driveway/garage in a suburb/the country, it doesn't cost me a lot.
That said, a lot of people (such as people who lose money driving for Uber) don't really properly account for their per-mile costs. I expect a lot of people would think twice about their driving habits if they had to feed a meter at 50 cents per mile every time they got in the car. (And, yes, that number includes some amortization of fixed costs, will be lower some day with electrics, and is probably lower today with an econobox--but the point stands..)
The marginal cost of cars is not just a function of mileage. There is non-trivial wear from just sitting, and you have to pay interest on the capital cost. This is at least 5% operating cost per year which is mostly eliminated with shared vehicles that are operated continuously.
Huh? You paid X dollars a long time ago for the Honda. If instead you and your identical neighbors had been sharing a sequence of Hondas which were driven continuously and then replaced, your contribution to the purchase prices (i.e., the total price of all those cars divided by the number of neighbors) would have been less than X dollars (since your Honda is still running). Furthermore, it would have been spread out in payments over 14 years rather than being paid upfront in 2003. Therefore, you could have kept that money in the bank stock market earning interest.
That's the cost I'm talking about. The maintenance cost of raw sitting time (rust, freeze-thaw cycles, etc.) are additional.
I’m almost 40, live in upstate New York (ie. Cold, snowy). Maintenance for a regularly driven car is all about mileage. Suspension and electronic motors start falling off at $150k miles.
Per mechanic friends, most better cars go until about 250k miles with modest maintenance issues, and after that things like transmissions fail and are not a good value.
The problem with the pooling is that any outlier blows the financial model. Whatever interest I would make would be wiped out by a single event where I’d need to rent a car.
There’s a place for things like Uber (or traditional taxis) and zip car. But unless you barely use a vehicle or cannot afford one, it’s a pain in the ass to share.
I'm not making any claims about the overall value of ride-sharing an autonomous vehicles. I'm only rebutting the idea that the cost of capitalization are trivial.
Not just but mostly. The biggest time-based costs are:
- Insurance. Some policies have surcharges or discounts based on mileage but it's mostly an annual cost for a typical driver. (Plus excise tax and state inspection.)
- In northern states, corrosion due to salt in the winter happens even if you don't drive a car many miles. My Honda dealer is always amazed when I bring in my old '97 Del Sol with minimal rust; I've just garaged it in the winter for about the past decade.
- Some fluid, belt, etc. maintenance is time-based
- If you want to have a relatively late model car then of course that's a time-based cost. But, if you hold onto a car until it becomes unreliable/expensive to operate then you're back to auto costs being mostly a result of mileage.
Yea, I was trying to be extremely conservative with my 5% number, and I think it's much larger due to some of the effects you mention. But I'm a little skeptical time effects are actually larger than mileage ones in total. Do you know a good quantitative argument (even if approximate) for this?
I'm arguing that mileage effects are much bigger unless you regularly flip cars when, say, a lease expires.
AAA publishes car ownership cost numbers. But the big challenge of quantifying this in general is that the relatively fixed component is a lot more for someone who gets a new leased vehicle every three years vs. someone who owns a car for 10-15 years and 150K miles.
Whether costs are mileage or time based vary heavily on the environment.
They don't salt the roads in the winter in Silly-Valley so maintenance costs are primarily mileage based.
In somewhere like NY it's going to be more time based because the vehicle will corrode at a fairly steady rate and salt solutions will counteract lubricants in places that can't be cleaned out (e.g. wheel bearings and ball joints) at a fairly fixed rate.
> Just like cloud infrastructure, it is more efficient to centralize and rent out capacity as needed.
Cloud infrastructure is probably a bad example. It essentially got started at the same time as x64 virtualization was becoming a thing, so you had people comparing the cost of operating 25 physical servers to what Amazon charges for 25 virtual machines and then the virtual machines are obviously a better deal.
But one VM host can run all 25 VMs, and if you operate it locally then the latency is much lower (<1ms vs. ~40ms), there is much more local bandwidth (symmetric gigabit or 10G vs. 50Mbps/20Mbps or whatever your site's internet link is), you don't have to pay anyone for those bytes, local servers are still accessible when Comcast is down, your data stays on site instead of being stored who-knows-where accessible to who-knows-who, etc.
There are use cases for cloud hosting, high volume static web hosting being the major one because then the data is public, the users aren't local and (this is the main reason) last mile ISPs charge uncompetitive prices for fast WAN connections. But people massively overuse it for other things. Companies that put the local file/print server in The Cloud are doing it wrong.
No way. It’s inefficent for a business trying to make money. As an individual, I don’t have to recognize depreciation nor do I get to deduct depreciation as a non taxable business expense.
If it were more cost effective to rent cars, we’d be going to Hertz every month. By the time you build in the overhead, profit, risk, etc, you always pay more. A middle of the road car costs <$0.60/mile to operate, insure, fuel and maintain. Uber is easily 10x that, even with Uber losing money on almost every ride.
I’ve been in the business of centralizing technology services for twenty years now. It’s a clear cost savings when you’re offering something that is a commodity that can obtained on the open market. It’s clearly cheaper to rent email. If you have the volume and capital, it’s clearly cheaper to own the fiber between facilities.
Outsourcing business process is only cheaper as long as your use profile is static — the change orders kill you. Transportation and people are similar to that. Taking the bus to work is often a money-saver. But paying a $30 uber tax or overpaying for Amazon groceries is just a drag. A family of 4 probably saves enough in paper towels alone to make 1 car payment.
There are a lot of benefits to a driverless Uber-clone that I think you're overlooking.
1. Driverless Uber means that you don't have to pay for an expensive human driver
2. I expect that the number of driverless taxis will be far more than the current number of taxis+Ubers/Lyfts, because suppliers are limited only by the capital costs of the cars.
2. As a passenger relying on driverless taxis, you no longer have to pay car insurance.
3. You no longer have to pay for parking.
4. You no longer have to find parking.
5. Businesses will no longer have to pay for parking.
6. You don't have to pay Uber for the entire trip -- you can get a taxi from home to a bus stop or light rail station.
The benefit of renting a driverless taxi over owning a car is that most commuters are only using their cars for an hour or two a day. This turns car renting into a by-the-hour spot market, which is far more granular than renting a Hertz car for a month.
But I think that freedom from parking will be the really big incentive that pushes driverless cars forward. Cars are really big, land in cities is really scarce, and car storage is a really wasteful use of land. We currently have homes and malls and offices surrounded by miles of parking[1]. I expect most of that parking to go away within a decade or two of driverless cars becoming popular, as owners reclaim it for more profitable uses of the land. And without abundant free parking, driverless cars & taxis will become more popular.
If you rent from Hertz every month, the car is still just sitting in your driveway most of the time, so there is no efficiency increase.
If you Uber everywhere, you need to pay someone else to drive you around all the time, that wipes out any possible cost savings.
Get rid of the cost of the person driving you and it will almost certainly be cheaper to rent than buy unless you have a very high utilization rate for the car.
In an era of low interest rates, the capital cost of a car is not huge.
I think people are also over-focusing on monetary costs; there are lots of other considerations like convenience, cleanliness, predictability, reliability, and finally social status and taste.
> In an era of low interest rates, the capital cost of a car is not huge.
In the limit of a high utilization rate (i.e., the increased fraction of the time the car is driving, which is probably 5-10), the capital costs are roughly given by the interest rate compounded over the lifetime of the car. Even at just 2% interest, this means a 21% savings if you would otherwise buy a new car every decade.
> I think people are also over-focusing on monetary costs;
I definitely wasn't suggesting capital costs will be the most important effects, I was just pointing out that an effective 20% discount on car purchase price is non-negligible.
I described a procedure for a rideshare co-op in which the members provide their own capital, which, from a capital cost perspective, is a clear win (>20%) compared to the status quo of individually owned cars. The members may in addition decide it's even better for outside investors to provide the capital in exchange for the members paying a monthly fee, but this means the total benefit to members is even larger than the member-capitalized co-op scenario.
How do you conclude that paying by use is not going to be a good deal? In my mind it's quite the opposite, it now makes financial sense to rid your life of daily dependence on your own car while enabling economical occasional car use.
Many people currently drive daily just because the marginal costs are so low (and aren't very environmentally conscious).
That's assuming no cost for parking which is a correct assumption if you are commuting from a suburban house to a suburban office park. But if you are commuting from a SF/NYC apartment to a SF/NYC downtown office and you want parking in both places then its safe to add a cost of $150 per month for each which is about $3600 for parking. Its all about context.
The obvious question is how would those numbers change with a shared car. It's quite possible it could be cheaper for many even with a higher per mile cost.
Hmm. So my question was "How do you conclude that paying by use is not going to be a good deal?" and you quoted by fixed cost observation back at me without making any additional argument.
To interpret this charitably, I guess you may be trying to say that it is still a good deal for people who drive a lot in any case and don't mind the fixed cost? I agree, but I think these people are going to be a shrinking segment, in contrast to your claim that it will "never" be a good deal for the consumer.
To reiterate, many people don't want to invest this fixed cost and drive everywhere, but once they have committed to the fixed cost, it's uneconomical to do the daily life the carless way.
All this is in the context of the necessary reduction of private car usage, considering the need recuce CO2 emissions. (Electric cars only cut car lifetime emissions to half, which is still unsustainable).
I already choose Uber-like cars over owning my own car, since I only need it occasionally - I commute by bus or bicycle. If driverless cars mean Uber is much cheaper, then my choice will make sense for even more people living in cities.
There are two trends in the car rental world that are happening today I think we'll see a lot more of in the next three years.
First, the Care By Volvo program on their new 2019 XC40, which is a set monthly fee that includes the lease cost + insurance + all maintenance + 15k miles per year, and which allows you to upgrade cars every year. It starts at $500/mo for a $35k car.
Second, the Porsche Passport program, which is a flat subscription fee for truly unlimited access to any Porsche. $2000/mo for their cheaper models, $3000/mo for any model. No mileage restrictions, 30 day commitment, everything included, and you can swap cars anytime (they deliver and pick up for you).
These are both expensive, but that's primarily because we're talking about luxury cars. Once manufacturers like Honda and Toyota enter this game the prices will come down. I really think we'll see this model, especially the Volvo one, replace traditional leases.
You're right that the capital cost of new cars is becoming too much for the average person to bear, but the rise of these new "all inclusive" lease-like programs means more people will use them. And all of these new leased cars enter the used market at some point, where prices come down and high-mileage drivers will buy them.
The key factor in favor of driverless cars is the time saved from driving. For many working professionals, their time is worth more than the monetary costs of owning and operating a car, electric or gasoline. Even if one already owns a normal electric car, it would be economically compelling to switch to a driverless one and use the time saved to create more value for work and career.
You’d think if using commute time productively was a big draw, everybody would be riding the bus already, right? I think that the folks who work in occupations amenable to telecommuting and who could be actually productive during a long enough commute probably already are. Self driving cars might add value for the folks whose schedule doesn’t work for park’n ride but otherwise fit the above params.
I mostly work remotely anyway these days, but taking a bus/train has rarely been an option for me during my career. And that's very common for suburban/exurban to suburban/exurban commutes.
There is a prisoners dilemma with car vs bus. If everyone would take a bus to go somewhere, in total people would be quicker home as there are no traffic jams. But if the traffic jams disappeared then people would selfishly take the car for a speed improvement.
> You’d think if using commute time productively was a big draw, everybody would be riding the bus already, right?
No, because the buses themselves are not a reasonable working environment. Every single person I know who can take the train to work does, many of them able to work (on trains that allow working - that's not, e.g. subways anywhere I've seen).
Not a single one of those will use a bus, even if it is overall faster (better route, etc.).
Driverless cars are likely to provide a reasonable working environment.
So electric cars cost more up front but less per mile, and you're maybe going to keep them longer.
There's an opposing dynamic: a taxi drives a lot more miles per year than a personal vehicle, so taxis (driverless or not) are likely to be the biggest early adopters of electric cars.
Maybe if autonomy develops slowly relative to electric cars, we'll mostly end up driving our own electrics, but if autonony develops relatively quickly, we'll have a more sudden transition to autonomous ride-sharing and hardly anyone will drive.
Absolutely. High-milage fleets turn over every three years. The Toyota Prius became popular with taxi drivers very rapidly. As soon as the economic advantage is real and large enough, the switchover will be quick.
There will then be a long tail of lower-milage users swapping over gradually, and a small chunk of resisters and nostalgists driving manual petrol vehicles because they like it or have the same "beater" for 20 years and don't want to change.
48 comments
[ 3.4 ms ] story [ 101 ms ] threadNobody will ever choose driverless Uber-like cars for a long time. Paying by the drink (ie paying the $0.60-0.80 + margin that it costs to operate a car) will never be a good deal for the consumer. People will rent because they need to.
The current situation with Uber and Lyft is an exception. Uber has revealed itself to be a criminal enterprise and both companies rely on VC subsidy and pushing liability to operators to survive. That’s not a model that will stand the test of time.
I would switch to a per-mile-rent if it was affordable. Electric, self driving cars will be cheaper than people.
The economics of this are the opposite. Having cars sit unused in driveways and parking spots, and be maintained by non-experts, is highly inefficient. Just like cloud infrastructure, it is more efficient to centralize and rent out capacity as needed.
That said, a lot of people (such as people who lose money driving for Uber) don't really properly account for their per-mile costs. I expect a lot of people would think twice about their driving habits if they had to feed a meter at 50 cents per mile every time they got in the car. (And, yes, that number includes some amortization of fixed costs, will be lower some day with electrics, and is probably lower today with an econobox--but the point stands..)
It’s still cheaper.
That's the cost I'm talking about. The maintenance cost of raw sitting time (rust, freeze-thaw cycles, etc.) are additional.
Per mechanic friends, most better cars go until about 250k miles with modest maintenance issues, and after that things like transmissions fail and are not a good value.
The problem with the pooling is that any outlier blows the financial model. Whatever interest I would make would be wiped out by a single event where I’d need to rent a car.
There’s a place for things like Uber (or traditional taxis) and zip car. But unless you barely use a vehicle or cannot afford one, it’s a pain in the ass to share.
- Insurance. Some policies have surcharges or discounts based on mileage but it's mostly an annual cost for a typical driver. (Plus excise tax and state inspection.)
- In northern states, corrosion due to salt in the winter happens even if you don't drive a car many miles. My Honda dealer is always amazed when I bring in my old '97 Del Sol with minimal rust; I've just garaged it in the winter for about the past decade.
- Some fluid, belt, etc. maintenance is time-based
- If you want to have a relatively late model car then of course that's a time-based cost. But, if you hold onto a car until it becomes unreliable/expensive to operate then you're back to auto costs being mostly a result of mileage.
AAA publishes car ownership cost numbers. But the big challenge of quantifying this in general is that the relatively fixed component is a lot more for someone who gets a new leased vehicle every three years vs. someone who owns a car for 10-15 years and 150K miles.
http://newsroom.aaa.com/tag/driving-cost-per-mile/
They don't salt the roads in the winter in Silly-Valley so maintenance costs are primarily mileage based.
In somewhere like NY it's going to be more time based because the vehicle will corrode at a fairly steady rate and salt solutions will counteract lubricants in places that can't be cleaned out (e.g. wheel bearings and ball joints) at a fairly fixed rate.
Cloud infrastructure is probably a bad example. It essentially got started at the same time as x64 virtualization was becoming a thing, so you had people comparing the cost of operating 25 physical servers to what Amazon charges for 25 virtual machines and then the virtual machines are obviously a better deal.
But one VM host can run all 25 VMs, and if you operate it locally then the latency is much lower (<1ms vs. ~40ms), there is much more local bandwidth (symmetric gigabit or 10G vs. 50Mbps/20Mbps or whatever your site's internet link is), you don't have to pay anyone for those bytes, local servers are still accessible when Comcast is down, your data stays on site instead of being stored who-knows-where accessible to who-knows-who, etc.
There are use cases for cloud hosting, high volume static web hosting being the major one because then the data is public, the users aren't local and (this is the main reason) last mile ISPs charge uncompetitive prices for fast WAN connections. But people massively overuse it for other things. Companies that put the local file/print server in The Cloud are doing it wrong.
If it were more cost effective to rent cars, we’d be going to Hertz every month. By the time you build in the overhead, profit, risk, etc, you always pay more. A middle of the road car costs <$0.60/mile to operate, insure, fuel and maintain. Uber is easily 10x that, even with Uber losing money on almost every ride.
I’ve been in the business of centralizing technology services for twenty years now. It’s a clear cost savings when you’re offering something that is a commodity that can obtained on the open market. It’s clearly cheaper to rent email. If you have the volume and capital, it’s clearly cheaper to own the fiber between facilities.
Outsourcing business process is only cheaper as long as your use profile is static — the change orders kill you. Transportation and people are similar to that. Taking the bus to work is often a money-saver. But paying a $30 uber tax or overpaying for Amazon groceries is just a drag. A family of 4 probably saves enough in paper towels alone to make 1 car payment.
1. Driverless Uber means that you don't have to pay for an expensive human driver 2. I expect that the number of driverless taxis will be far more than the current number of taxis+Ubers/Lyfts, because suppliers are limited only by the capital costs of the cars. 2. As a passenger relying on driverless taxis, you no longer have to pay car insurance. 3. You no longer have to pay for parking. 4. You no longer have to find parking. 5. Businesses will no longer have to pay for parking. 6. You don't have to pay Uber for the entire trip -- you can get a taxi from home to a bus stop or light rail station.
The benefit of renting a driverless taxi over owning a car is that most commuters are only using their cars for an hour or two a day. This turns car renting into a by-the-hour spot market, which is far more granular than renting a Hertz car for a month.
But I think that freedom from parking will be the really big incentive that pushes driverless cars forward. Cars are really big, land in cities is really scarce, and car storage is a really wasteful use of land. We currently have homes and malls and offices surrounded by miles of parking[1]. I expect most of that parking to go away within a decade or two of driverless cars becoming popular, as owners reclaim it for more profitable uses of the land. And without abundant free parking, driverless cars & taxis will become more popular.
1. https://d3i6fh83elv35t.cloudfront.net/newshour/app/uploads/2...
If you Uber everywhere, you need to pay someone else to drive you around all the time, that wipes out any possible cost savings.
Get rid of the cost of the person driving you and it will almost certainly be cheaper to rent than buy unless you have a very high utilization rate for the car.
Add in surge pricing, surcharges for dead-ending routes, etc and you have a cab. Useful, but inferior to controlling the car.
I think people are also over-focusing on monetary costs; there are lots of other considerations like convenience, cleanliness, predictability, reliability, and finally social status and taste.
In the limit of a high utilization rate (i.e., the increased fraction of the time the car is driving, which is probably 5-10), the capital costs are roughly given by the interest rate compounded over the lifetime of the car. Even at just 2% interest, this means a 21% savings if you would otherwise buy a new car every decade.
> I think people are also over-focusing on monetary costs;
I definitely wasn't suggesting capital costs will be the most important effects, I was just pointing out that an effective 20% discount on car purchase price is non-negligible.
https://news.ycombinator.com/item?id=15940795
I described a procedure for a rideshare co-op in which the members provide their own capital, which, from a capital cost perspective, is a clear win (>20%) compared to the status quo of individually owned cars. The members may in addition decide it's even better for outside investors to provide the capital in exchange for the members paying a monthly fee, but this means the total benefit to members is even larger than the member-capitalized co-op scenario.
The guy who owns the car, the guy who provides the software to drive it and the money guy all need their piece of the action.
Many people currently drive daily just because the marginal costs are so low (and aren't very environmentally conscious).
You just answered my own question.
For a fixed cost of $400/year, insurance of $1000 year, I can be at any point on the east coast in <24 hours for like $0.25/mile + time.
To interpret this charitably, I guess you may be trying to say that it is still a good deal for people who drive a lot in any case and don't mind the fixed cost? I agree, but I think these people are going to be a shrinking segment, in contrast to your claim that it will "never" be a good deal for the consumer.
To reiterate, many people don't want to invest this fixed cost and drive everywhere, but once they have committed to the fixed cost, it's uneconomical to do the daily life the carless way.
All this is in the context of the necessary reduction of private car usage, considering the need recuce CO2 emissions. (Electric cars only cut car lifetime emissions to half, which is still unsustainable).
First, the Care By Volvo program on their new 2019 XC40, which is a set monthly fee that includes the lease cost + insurance + all maintenance + 15k miles per year, and which allows you to upgrade cars every year. It starts at $500/mo for a $35k car.
Second, the Porsche Passport program, which is a flat subscription fee for truly unlimited access to any Porsche. $2000/mo for their cheaper models, $3000/mo for any model. No mileage restrictions, 30 day commitment, everything included, and you can swap cars anytime (they deliver and pick up for you).
These are both expensive, but that's primarily because we're talking about luxury cars. Once manufacturers like Honda and Toyota enter this game the prices will come down. I really think we'll see this model, especially the Volvo one, replace traditional leases.
You're right that the capital cost of new cars is becoming too much for the average person to bear, but the rise of these new "all inclusive" lease-like programs means more people will use them. And all of these new leased cars enter the used market at some point, where prices come down and high-mileage drivers will buy them.
The difference is that cars are reliable enough for the model to work without the implicit tax subsidy.
It also reinforces to argument that paying Uber/Lyft/etc by the drink is a bad deal :)
Over 80% of new cars are financed so I think we're already past that point.
https://perspicacity.xyz/2017/05/24/this-is-how-big-oil-will...
https://equilibriabook.com/
No, because the buses themselves are not a reasonable working environment. Every single person I know who can take the train to work does, many of them able to work (on trains that allow working - that's not, e.g. subways anywhere I've seen).
Not a single one of those will use a bus, even if it is overall faster (better route, etc.).
Driverless cars are likely to provide a reasonable working environment.
There's an opposing dynamic: a taxi drives a lot more miles per year than a personal vehicle, so taxis (driverless or not) are likely to be the biggest early adopters of electric cars.
Maybe if autonomy develops slowly relative to electric cars, we'll mostly end up driving our own electrics, but if autonony develops relatively quickly, we'll have a more sudden transition to autonomous ride-sharing and hardly anyone will drive.
Here's an interesting article from this perspective: https://perspicacity.xyz/2017/05/24/this-is-how-big-oil-will...
There will then be a long tail of lower-milage users swapping over gradually, and a small chunk of resisters and nostalgists driving manual petrol vehicles because they like it or have the same "beater" for 20 years and don't want to change.