1. When self-driving cars are common, people won't park their car from 9-5 while they're at work.
2. While the passenger is at work, the self-driving car will be giving rides to other people, or delivering packages and meals (not parking or driving around empty).
3. Greater car sharing means fewer parked cars, lower demand for parking and hence lower cost to park.
> While the passenger is at work, the self-driving car will be giving rides to other people, or delivering packages and meals (not parking or driving around empty).
Says who? I certainly don't want random people inside my car when I'm not around. Don't really feel like having "For a good time call 867-5309" and a drawing of a penis gouged into my center console, and having the seatbacks tagged with magic marker. ("We're sorry to hear that, sir. We have credited your account with a $25 credit for your trouble.")
130 million people drive to work every day in the USA. Something like 115 million cars. There are maybe 200,000 taxis in the USA. If we put 1 million cars to work, we've still got 114 million sitting idle, or rather, doing loops around the block because it's cheaper.
> Greater car sharing
Says who? You can already carpool and have ALL of the advantages of a self-driving car, it's just Larry driving the car instead of Waymo. But as everyone on HN will tell you, they don't want to carpool, they want to ride alone. Riding with others is what poor people do.
You also have the economics totally backward. Your assertion is that driving will be better in every way - easier personally, less congestion, cheaper parking - and as a result, there will be less driving? Utterly backwards. Make something cheaper easier or better, and we use more of it, not less.
> I certainly don't want random people inside my car when I'm not around
It's not your car, it's Waymo's car.
Waymo outfits the cars, provides continuous monitoring and support, operates a fleet of field techs for repair and cleaning, takes responsibility for insurance and regulatory compliance, provides dispatch and routing services, collects payments, etc.
The cost is justified by high utilization of the fleet. The economics don't work for a single car doing nothing 95% of the time.
Why would anyone want to own their own self-driver?
> Utterly backwards. Make something cheaper easier or better, and we use more of it, not less.
Yep, it's impossible to make anything cheaper, easier or better, because it instantly goes to shit when people want more of it.
> Why would anyone want to own their own self-driver?
Because they don't live in a large city for example. It's easier to cover a known average of rides with a million of users than when you have a town of 2000. You have a normal day where many projects walk to work, you get a local event where 200 people may turn to, you get forest fire where 2000 people need to get out.
It's 100% true that all the benefits of self-driving cars will be for naught if there's no congestion pricing implemented for driving alone. This will be either in the form of a toll or tax, or by making all but one of the freeway lanes shared-ride only.
Once you've implemented congestion pricing, riding with others is what 99% of people will be doing and the stigma is gone. If you don't want to ride in a car with graffiti etc. there'll be an option to order a better quality of ride.
Carsharing with autonomous cars will be way cheaper and 100% ubiquitous. I don’t even have Uber in my hometown, and I‘d love to get rid of my car. I‘m quite sure I could save €200 every month.
I generally agree with all of these, however demand is not uniform which results in some vehicles having to be parked while they're not needed.
As another commented, it will be a very interesting optimization problem to figure out how to efficiently use the underutilized time of the fleet. There are a lot of things which are done at peak times now only because the owner of the car has to physically drive it there that could easily be moved to non-peak.
The nice part about that problem is that it can be solved by some startup, which means that we don't have to be alarmist about it.
As for the uniform thing, you are right, but it can be restated in economic terms as the car being more expensive to use during certain times of the day -- so people will have an incentive to go to work in off-peak hours (which will be an option not available to everybody, but the more that do so, the less the premium cost will be) thus smoothing demand.
> demand is not uniform which results in some vehicles having to be parked while they're not needed
Exactly. The salient questions are whether, at the point of minimum demand during the day, there would be more unutilized self-driven vehicles than the current state of unutilized owner-driven vehicles, and if at the point of maximum demand, there would be more or less vehicles on the road than today.
Intuitively, the answer is "no" to the first question, since cars used for commuting in the morning and evening could be used for deliveries during the day.
I don't know about the second question, and this probably has more to do with shifting trips from public transit to self-driven cars (not cruising to avoid paying to park).
This research doesn't consider the reduction in parking demand, the reduced cost to park, the densification from reclaiming thousands of hectares of parking spaces, offsetting traffic of current delivery vehicles, etc.
The challenge I see here is mostly psychological. Will people be okay with having to wait for the arrival of their vehicle when the self-driving car is delayed? Will they be willing to accept the wear-and-tear their car suffers when someone who doens't own it uses it as a delivery vehicle?
The illusion of control is a powerful one, and I think many people will choose to own and park a personal car near to their work so that they can get in it and get stuck in traffic immediately, rather than waiting at work while the car returns through said traffic.
Another scenario is to have the car drop you off at a mass transit hub, then return home. When you get on the bus/train/whatever to return home, the car will meet you there. This may involve cars cruising around near the hub, though.
In urban settings, when self-driving cars are common, people (generally) won't own cars. Owning a car – an expensive depreciating asset – will be an enthusiast thing, like golf clubs or whatever; eventually, the self-driving truck will tow your non-self-driving car to the track...
Assuming the pricing works, most people will just use a car service.
I don't believe that for one second. Would it be cheaper for me to take a taxi to work every day than drive my own car? Absolutely. Do I want to? Absolutely not. The comfort of having my own vehicle that is set to my own preference and loaded with my stuff whenever I need it trumps the cost argument. Shared vehicles are just gross.
>Would it be cheaper for me to take a taxi to work every day than drive my own car? Absolutely.
I don't believe that for one second. Assume that the average car costs $523/mo[1] (and if you wanted a lower car payment it's totally possible to get a cheap car with a lower monthly payment). That means the daily cost of the car is $17.19 (523*12/365).
An uber (which is often less than a taxi and way more convenient) costs about $15 to get to work if you live nearby (it would be closer to $30-40 if you lived in the suburbs), and another $15 to get back home (possibly more because of surge pricing, depending on when you leave and local events such as sports games etc). This is assuming you never want to drive to get groceries, visit friends, etc... That's another $10-15 each way, every time.
Sure you don't have to pay for gas but there's no way the <$1/ride increase will account for that. I suppose you might have to pay for parking, but even then you could just buy a cheaper car ($400/mo car payments aren't exactly unheard of) and use the difference to pay for monthly parking.
I think using numbers with human drivers doesn’t help because
a) the uber driver has much higher marginal cost on each drive than a self driving car because of the human in the drivers seat and
b) the self driving car is likely to have a much higher fixed cost than the taxi with a human driver because of the expensive computer, software and new hardware like lidar.
So self driving cars are in my best guess going to have very high fixed costs with very low marginal costs, which means there is a large opportunity for a “renters” market
> the self driving car is likely to have a much higher fixed cost than the taxi with a human driver because of the expensive computer, software and new hardware like lidar.
These will become commodities when the technology matures.
I've owned quite a few cars in my life and not once was my payment > $400 even when putting very little down (usually my current car). I only buy gently used, reasonably priced cars though. My current truck is a 2007.
I think the only way the math would work out favorably is if you had an expensive car, a short commute, or preferably both. For a regular appliance car, say a Prius, used strictly for commuting (5 days/week) the TCO would be about $21/day. This is almost exactly what Uber would charge me to go one way 14 miles to my office downtown (I live on the outskirts of the city). So about twice as expensive to use Uber, and only if it's just for commuting.
But if I had something fancy, like a Tesla Model X, the daily cost (similar parameters as before) works out to $68/day, and then I could save money by using Uber.
Of course, that's not really a fair comparison because the Tesla is much nicer than anything Uber is likely to pick me up in, so the Prius is a much more honest comparison.
> The comfort of having my own vehicle that is set to my own preference and loaded with my stuff
From this statement it seems like you either don't work in a city, or live outside of a large metropolitan area. There's little chance you're leaving stuff in your car routinely in a large city and not having it stolen. This isn't meant as an insult, but I think it means you're not in the target demographic for a shared self-driving car service.
"Loaded with my stuff" doesn't mean all of my favorite consumer electronics left in the open in the passenger's seat.
My behavior barely changed between living a mile away from the nearest person to parking daily in SF. Yes if I have something valuable in the cabin I'll put it in the trunk when I leave the car, but that's it.
The cost difference is likely to be much more than between a present-day taxi and a car. The biggest cost for taxis is the labor of the driver, which usually dwarfs depreciation/maintenance/gas by an order of magnitude. If taxis are currently cheaper than car ownership for you, self-driving cars may be cheaper than taxis may be cheaper than that by a factor of 10x.
Would you pay 10x to avoid having to share seats, particularly for an activity that's still a fairly large chunk of the household budget?
I would be interested in knowing how you calculated your numbers. I did some digging and found an old (10 years ago) report from the Univ of Illinois on the operating costs of taxis in Chicago. In round numbers, the typical taxi driver brought in just under 60K/year in revenue and had 42K/year in operating costs. Assuming there hasn't been a major upset since then, this is a very long ways from 10x.
I'm figuring that a typical taxi driver might make $30/hour and in that hour drive about 20 miles. At 30 mpg and $3.00/gallon gasoline that's about $2 for gas. Figure that a typical car costs $20K and lasts for 200,000 miles, so 20 miles = 1/10,000 of the car = ~$2 in depreciation. Maybe another $1/hour in maintenance - that'd imply about $2K/year, which seems high, but this is higher mileage than a family car anyway. Total about $5/hour, which would be around 1/6 labor costs; not quite 10x, but around that ballpark.
Switch to electric vehicles (which are suddenly feasible with the car-service, never-have-to-park model!) and centralize maintenance across a whole fleet of cars and you can shave a bunch off gas & maintenance, though your depreciation would probably go up unless battery packs became dirt cheap.
It'll depend a lot on the difference in price for people - or another way of thinking about it is you can rent a better car than you can buy.
I think a lot of people will have much higher cost pressures than many of us on HN (at least, my impression is that we here are typically well above local median wages). That may push it a lot, and even more when you remove the upfront costs.
Also personally I'd love to not have to own a car. If it's even cheaper - great.
American buses really aren't that gross, they just don't go a lot of the places most people need to go. Plus they stop running at weird times (the one that stopped at the place I interned during the summer had the last pickup time at 6PM, so if you weren't done by then you needed an uber. Plus that office park was the most rural stop in the bus system so you had to live further down town if you needed to take it.
When self-driving cars are common, most people living in the city won't own cars. It will be cheaper to take a cab and if you need to go out you can rent a car. Skyrocketing insurance will be the real killer of urban free-driving.
The one way I can think it would get more expensive is that the pool of drivers with insurance would get smaller. That would naturally drive the rates and such up.
The best analogy would be insurance for historical or classic vehicles; if you've ever looked into it, not only do they require certain things normal car insurance doesn't (for instance, they will typically only insure you if your car is kept in an enclosed garage), but because the pool of drivers is much smaller, it is more expensive to get.
> The one way I can think it would get more expensive is that the pool of drivers with insurance would get smaller. That would naturally drive the rates and such up.
Why would it increase the rates?
Rates are mostly driven by statistical analysis taking risk of accidents and cost of repairs into consideration. It is not a product particularly fit for a volume discount. If anything, fewer drivers on the road should mean fewer accidents, if the area/volume of roads stays the same (because of the drop in density, which means a drop in probability to be in the same spot as another driver at a given time).
> The best analogy would be insurance for historical or classic vehicles; if you've ever looked into it, not only do they require certain things normal car insurance doesn't (for instance, they will typically only insure you if your car is kept in an enclosed garage), but because the pool of drivers is much smaller, it is more expensive to get.
I don't think it's due to the pool of drivers being smaller. I think it's because the incident cost is much higher than high volume cars, because both parts and labor are more expensive for low volume productions (custom/exotic). Source: my exotic car.
Classic car insurance is more expensive because they are more expensive to fix. Insurance for a piece of property is proportional to the risks of damage to or caused by the property, which shouldn't change much in a landscape with self driving cars in play
As to point 1. why do you assume that current patterns of ownership will persist? If self-driving cars can be summoned as needed why would you want the hassle of maintaining a private one especially as parking becomes astronomically expensive?
With regards to point #2, do you really think that multiple manufactures will be able to agree on a single vehicle to vehicle communication protocol that is robust enough to sort out shifting around a parking lot? Or would the parking lot have the infrastructure that communicates to the cars? It doesn't seem an insurmountable task either way.
those protocols were developed when the internet was much smaller and (mostly) in the control of researchers & technologists. now, we have multi-billion/trillion dollar entities jostling for control over a multi-billion/trillion dollar industry.
I would direct you to the numerous things that _are_ standard on automobiles. For example, CAN bus[1] which is used in almost every single car you'll encounter these days and is what allows their modern electronics to communicate with each other.
The auto industry is fully capable of standardizing on something when it's mutually beneficial for them to do so.
There's a solution for #2, and it's even better than having the car park itself in a traditional parking structure, which has a lot of unused wasted space dedicated to things other than the individual parking space the car occupies (the paths to the parking spaces, stairs, elevators, wider parking spaces so people can get in/out of the car, etc. I saw in Japan, China, Germany (and other places) they have circular parking structures that you just drive your car onto a platform, get out and pay. The automated, robotic system, lifts the car up to whatever level there is an open parking space, and just puts the car in the space. When you need it, it retrieves it and away you go. No stairs to climb, no elevators, no paths for cars to drive on. Very compact and efficient. I imagine it cuts down on quite a bit of theft, breakins, sexual assaults and security equipment (cameras, staff) as you really just need to make sure the entrance is secure. Even better, you can put them all underground. https://www.youtube.com/watch?v=rXWrpE1Khkk
This doesn't look that more space efficient than a normal parking garage. I'm sure better implementations could exist that would make it way better. Also... I'm imagining having to wait in a huge line during rush hour because everyone wants to get their car out, and they have to wait for the car retriever arm to travel to each car and back for all 20-30 people in line ahead of them.
There is a lot of space saving there. There are no "roads" for cars to have to drive on in the garage to get to the space individual spaces, no stairwells.
> I'm imagining having to wait in a huge line during rush hour because everyone wants to get their car out
That problem exists currently, except you wait an hour in line trying to exit the garage and nobody allowing you to back out because they want out first. I don't really see a solution to that particular problem if everybody is going to continue to have their own individual transports taking up space all day while not in use, and then everybody getting off work at the same time and rushing to get their cars at the same time, clogging the freeways at the same time. Well, except for things like Teslas, it would be nice to just summon your vehicle and have it unpark itself and drive to you to pick you up and give you an uber-like map and ETA on where it is. You could be doing more productive things while waiting for your car instead of standing (or sitting) in a line. 20 minutes before work ends, just summon your car so it's out front waiting, or circling the block or whatever.
(reading back over this comment, I might appear overly negative, but I would think all the issues I raise are solvable)
What's the throughput like on those systems? I could imagine at peak times it might take a while to get your car back. You'd also want to make sure you didn't leave anything in the car accidentally. Finally, imagine the trouble one of these breaking down would cause.
they address point #1 directly (still cheaper to drive around, at $0.50/hour). but really, they buried the lede:
> "But no one owns an autonomous vehicle now, so there's no constituency organized to oppose charging for the use of public streets. This is the time to establish the principle and use it to avoid the nightmarish scenario of total gridlock."
their thesis is that now is the right time from a policy perspective to implement congestion pricing on autonomous cars. the driving around aimlessly bit is literally to alarm politicians into action.
a pay-per-use model is economically rational but notoriously hard politically. only recently has technology advanced enough to do real-time pricing, which is arguably more fair than former systems (like flat rate pricing). ideally you'd pay per mile*minute (similar to taxis and ridehailing).
Not a bad idea, but there's the question of why we should single out autonomous cars, rather than having congestion pricing for everyone? Every car takes up space. If it's doing something useful then it's worth it.
you shouldn't limit it to autonomous vehcles, but as the article notes, no one wants to pay for something that was formerly free, so it's politically difficult. but the strategy of implementing it only on autonomous vehicles now provides a wedge for wider implementation later.
But we already have 'no-cruising' ordinances to harass human drivers endlessly circling around looking for fares, parking or prostitutes, artificially creating traffic and the hyperbolic 'nightmarish scenario of total gridlock.'
This is not a new problem, and focusing solutions on autonomous vehicles is pointless discrimination.
Part of the benefit (for me) of traveling by car is thst I can leave whenever I want. If I have to wait even 5 mins for my car to return to me, I'd consider that an inconvenience - if I could pay for short term parking near by, I'd take it.
It's reasonable that you'd pay a premium for the privilege of being able to leave the second you're ready to go, compared with say:
- Telling your car to do rideshare until 5pm, and then stand by until you're ready to go sometime between 5pm and 6pm.
- Telling your car to go park far away and you'll give it twenty minute's notice when you're ready to go so that it can come get you.
- Telling your car to park in a lot where it'll be boxed in and may take a few minutes of shuffling to get itself out.
That said, I definitely can see an issue here where everyone's cars all descend on the downtown trying to perform curbside pickups right at 5pm. You could imagine similar things happening in any parking lot environment (mall, movie theatre, amusement park, etc), where the prime kiss-and-ride location ends up being a giant snarl of self driving cars all trying to save their owners a five minute walk across some asphalt.
I don't disagree, but that one time you have a sudden errand to run or an urgent need to get him,e and your car is off being driven by others... Obvs I'm thinking very much in an ownership mentality here, and in tbe future I'd just grab another random car and head home in that, but this is the mindset that we will have to overcome.
(btw I never commute by car, only by bike ad public transport - so I'm just playing devil's advocate here)
I believe the assumption is that the car is owned by a ride-sharing company, not you. You are not letting random strangers into your car -- you are one of the random strangers.
If it halved the cost of ownership of your car and the service was guaranteed to reimburse you fully for damages (zero deductible and zero higher insurance rates) then a lot of people would, I think.
And you could probably keep personal things locked in your trunk or a high-security locked glove compartment.
> and the service was guaranteed to reimburse you fully for damages
That will be of zero comfort the moment you get inside your car, and find out a drunk has puked into the foot well, or somebody was incontinent in the back seat.
...and those are the nice scenarios - I could think of plenty of others that wouldn't be pleasant at all for the owner.
At that point, your only recourse would be to send your car off to be detailed/repaired/whatever, while you went to hail a ride service (and hope you don't get another car with the same issue).
Heck - can you imagine getting into your car, only to find that it was inhabited by a chain smoker for the entire day?
Even if it took full pictures of everything before and after the ride, and the identities of the people involved where validated?
I mean there is still the freak cases of them doing meth or otherwise not caring, but that seems likely to be something you can insure yourself out of for a relatively small cost.
Sure, all that is possible, but you'd still be deprived of your vehicle when it showed up all funkified, while you then had to scramble to hail another vehicle - potentially making you late to your next destination. If it's your personal car, then you're stuff using such ride services until you get your car back. With no guarantee that a similar thing will happen to you when one of them picks you up.
I would love it if it cost me only $0.50/hour to operate my car. Conservatively I'd go with more like $5/hour (assume 10mph average at $0.50 per mile).
According to [1], median daily unreserved parking in SF is $29/day, so at 6 hours you break even. You can likely get much cheaper than this by driving a bit further away. Even at the $38 for New York and $42 for Honolulu you still break even around 8 hours.
> "Even when you factor in electricity, depreciation, wear and tear, and maintenance, cruising costs about 50 cents an hour—that's cheaper than parking even in a small town," says Millard-Ball. "Unless it's free or cheaper than cruising, why would anyone use a remote lot?"
Doesn't take too long to prove it is too low. Lets say this fancy self-driving car cost $50,000. Lets say it is expected to go 250,000 miles in its lifetime. That is $.20 per mile! If it cruises at only 25 mph, that is $5 in wear and tear alone! Even if you slash the price and up double the life of the vehicle, you are paying more than $.50 an hour.
I don't know, that low of a cruising speed suggest a lot of stop and go driving. Even if true, the analysis was meant to be a simple refutation of the $.50 per hour claim, and I really doubt you can increase durability enough to tip the balance. And that doesn't even include regular maintenance- lubrication, tires, brakes. It doesn't include electricity. It doesn't include damage, insurance costs, etc.
When the traffic lights can coordinate with cars, they can avoid the need to stop, at most they may just need to slow down, which can be done without brakes.
An EV doesn't need (much) lubrication, for example, the Chevy Bolt EV maintenance schedule has only tire rotations for the first 150,000 miles:
Realistically, the average speed of any car navigating an urban grid with stop lights/signs is usually going to be a lot less than 25 miles per hour. Maybe more like 5-10, even if it's not deliberately trying to drive slow.
The point is that 50 cents an hour is not an accurate representation of the cost of continuously operating a car for an hour. The IRS allows something near that per mile as a standard mileage rate. If the car is driving more than a mile per hour on average, it's significantly exceeding 50 cents/hour.
Even if you assume a modest 20 mi/gallon at $3/gal, you're looking at about $0.15 a mile. A $40 oil change every 5000 miles, and a more expensive maintenance job every 10000 miles isn't going to push you past $0.25/mi.
The IRS comes up with a roughly $.50/mile figure. It takes into consideration the physical wear on the vehicle as well as consumables and fuel costs, among other things I think (averaged nationally).
Came here to say this. The article seems to have a basic lack of imagination about the types of new vlaue automated cars will create.
It also sounds like a crazy hard logistics/optimziation problem: "I have X number of rolling cars, how do I use them to maximize their value and reduce maintance and resource use?"
Maybe the author of the article has a basic understanding of physics of rock throwing? Those empty cruising cars will be sitting ducks for angry or even simply very bored and irresponsible people.
Right now we have parking lots full of cars without people in them. Seems like sitting ducks for bored people with rocks, right?
Why would anyone have windows in their house if someone could just throw a rock through it?
In reality, people are usually well behaved. The fact that self-driving cars have cameras and sensors like mad means that even fewer people would risk the jail-time they'd get by throwing rocks at them.
The idea that suddenly once cars are driving without people there will be a significant enough uptick in vandalism that it will deter cars from driving around is absolutely laughable.
If this were a realistic threat, we wouldn't have parking lots now.
Not sure it would be any different with people inside. If I'm using an Uber self-driver why would I care if someone tossed a rock at it (so long as the self driving sensors aren't damaged)?
Right! Each self-driving car replaces how many other cars on the road? Look at regions where only cabs are allowed - they don't park much either - and congestion goes down. There are solutions.
"Cars can go off and park somewhere further away. Cruising around the city isn't free (fuel, maintenance, risk of damage)"
They can, but then they are farther away when someone calls for a ride. If there's only a few self-driving car rental companies, then maybe it's not so bad, they can send most cars to a remote lot, with a few cars remaining in the city to address demand. But if hundreds of companies are going to have cars on the street, then they'll all want their cars to be crusing near potential riders.
Though a bigger problem is that self-driving cars will cause double the traffic of normally driven cars -- they'll drive into the city to drop off their passenger, then out of the city to a remote lot.
1. Cars can go off and park somewhere further away. Cruising around the city isn't free (fuel, maintenance, risk of damage)
2. We should be able to get more density out of our existing parking lots if cars can park each other in and shuffle around to let each other out.
3. Cars can go do something more useful (run errands, work for ride share).