Hacker News Reader

Comment by willvarfar | Hacker News Reader

Can poor people afford to buy a new car? If not, how old is the car when they buy it? In Sweden used cars are typically brought by their second buyer when they are 3 years old.

As software is so cheap to duplicate, why can't all cars be self-driving? Given the magnitude of traffic deaths, why can't all cars just be self-driving immediately? Why make it a premium thing that only the rich can afford?

Then within a few years, even poor people who buy second-hand cars will have self-driving models.

View on HN

> As software is so cheap to duplicate, why can't all cars be self-driving? Given the magnitude of traffic deaths, why can't all cars just be self-driving immediately?

I'm still feeling surprised by the number of intelligent people who are quite capable of thinking critically, quite uncritically deciding that self-driving cars are ready for the market now. The only thing which comes remotely close is Tesla's recent software upgrade, and Elon Musk was very careful to emphasise that it in no way replaces the driver. Yet still, the hype-train is running so fast that people decided what he really said was "self-driving cars are here now."

A moderately successful prototype from Google or anyone else does not mean the problems are solved. These innovations could still go the way of 90s-era virtual reality goggles, flying cars or the personal jet-pack. Don't get me wrong, it's exciting stuff but there are still significant technical issues, not to mention the legislative and social problems which will inevitably arise.

These links from a Google search for "Google self-driving car problems":

http://gizmodo.com/6-simple-things-googles-self-driving-car-...

http://www.nytimes.com/2015/09/02/technology/personaltech/go...

http://www.technologyreview.com/news/530276/hidden-obstacles...

There's much more, I can't dig out the links now. But don't be one of those guys in 10 or 15 years time being tedious about "Dude, where's my self-driving car?" It might not happen, and it's certainly not here yet.

And even if the technology was ready and on sale, the idea that we'd instantly dispose of our existing stock of automobiles and never feel the need to pay attention to the road again contrasts starkly with experiences in the aviation industry, with most control still being exercised by pilots (also the cause of many mistakes) despite viable autopilot systems being very well established.

Another issue is, would people board a fully-automated plane? Similarly, it seems many people see self-driving cars as a solution that other people should use, but they are, themselves, the competent drivers who don't need that kind of thing. I can foresee other issues with hacking firmware by drivers, people cracking the software as with the recent Jeep exploits, and actual software bugs leading to death or injury and the class-action lawsuits that might give rise to.

There are behavioural issues as well: people deliberately cutting in on autonomous vehicles because (rather than aggressive tail-gating) the AV will just back safely away, so that dangerous behaviour will become more commonplace. Before we get onto pedestrians just stepping into the road in the serene assurance that the AV will shudder to a halt for them, to the discomfort (possibly injury) of the vehicle's occupants. I can see people doing that kind of thing just for a lark, which would encourage drivers to go for the "aggressive" firmware upgrade to put those jaywalkers back in their place.

Personally I think (a) we'll get these vehicles sooner or later, but (b) they won't be a panacea, and (c) it's going to be an absolute minefield in technical, social and legal areas.

>> I'm still feeling surprised by the number of intelligent people who are quite capable of thinking critically, quite uncritically deciding that self-driving cars are ready for the market now.

I'm not. Not at all. Sadly.

Anecdotally, I'd say that something like 50% of the cars I see on the road are at least 10 years old. Probably 15% are 20 years old or more. I live in one of the poorer states in the US, so my experience may be somewhat skewed.

This also brings up one of my concerns with self driving cars. A vehicles (or and mechanical device) age, they become less reliable. The brake booster on my truck went out today (20 years old). I could still brake, but it took significant effort. I wonder what the failure modes look like on the self driving apparatus, and wether they fail gracefully and safely. I think it will be important that when we have 10 or 15 year old self driving vehicles on the road that they have engineered in safe failure processes.

I'm in Iowa. Growing up (in the 70's) cars were largely rusted-out hulks. Then new cars started appearing, so that almost every car on the road was shiny and new-looking by the 90's.

Now the rust is showing up again, and I don't think its the weather. Its the economy, and the general slide of the middle class.

Maybe we need a 'rust index' to measure mean spending/affluence?

The median car age in the US is just around 10 years.

There's a lot more to a self-diving car than just software, there is a ton of needed sensors and other hardware.

Let's say an expensive self-driving car is around $100,000 new, even buying that car used 3 years later would be out of the budget for many people.

You are conflating cost and price?

Self-driving cars will be expensive because they will be marketed as a premium product to maximize profit.

Sometimes public service must trump? Disclaimer: I live in Sweden, so my perspective may offend you ;)

> Self-driving cars will be expensive because they will be marketed as a premium product to maximize profit.

You have this backward. The reason entrepreneurs often start with premium products is because you can extract more money from rich people and use that to pay product development costs. Developing self-driving cars is extremely expensive, and it will be for quite a while. The costs will eventually fall, but as with personal computers, it will take a while to figure out how to do that.

If you really had cars that could self-drive and the technology was effectively free, then the profit-maximization strategy wouldn't be to sell Teslas, it would be to sell them to everybody. Tesla is selling maybe 10k cars per year. Even a $20k premium for self driving is only $200m. But the total US car market is 7 million or so. If you could get just an extra $1000, that would be $7,000m dollars in profit.

It's hard to forcibly lower the price of something without undermining your economic system. Software is expensive to make. Premium priced products exist because developers cost a lot of money and take a long time to get it right.

Practically speaking you could government subsidize the software which is a respectable policy in Europe, and that would foster competition and drive the prices down.

I expect that self-driving cars will -at least initially- be expensive -in large part- because the sensing and computational hardware on the car is expensive.

Once Mass production kicks in the cars will get cheaper and cheaper as the processes to create them does so.

When the automobile was first invented it would have seemed implausible for it to be accessible and (relatively) affordable to everyone. Look at where we are now.

Cars are still bloody expensive. Many many people make the decision not to buy a car when they live in a city with good public transport. It's only in places like Australia and most of the US where you practically need a car that ownership is ubiquitous. Elsewhere people find that there are better things to spend all of that money on...

> Once Mass production kicks in the cars will get cheaper and cheaper as the processes to create them does so.

Yes, but there is an expected near-to-medium-term price floor on the sorts of good sensing equipment that such cars will require, not to mention certified (and patent-encumbered) software to fuse that data and turn it into useful and safe action.

This is stuff that you don't want to cheap out on; it's how the car sees and operates safely!

Cost and price are the same thing. Maybe you mean cost/price and worth/value?

Isn't the definition of "profit" literally the difference between cost and price?

http://www.investopedia.com/ask/answers/101314/what-differen...

My point is the BoM of self-driving sensors and cpus for a car is not actually that great. The new Teslas and Volvos don't have fancy lidar scanners like the Google cars have.

The other day I was passenger in a automatic-breaking-at-intersections Volvo. We did not put that to the test. But its 'just' higher-precision parking sensors that are cheap and ubiquitous now.

Price is the amount the consumer pays, while cost is the price the business pays. It may cost me $5 to produce a product, which I could then sell at a price of $7, for a profit of $2.

I think with self driving cars the whole ownership model will become passe.

Think of it as a fleet of autonomous taxis operating at very low cost.

I don't see a great future there myself.

Imagine this: where you pay per the minute or hour for location to location driving, at a rate that is a multiple of owning it yourself. And for the poverty, low, and middle income classes, car ownership is unattainable.

Or if you are late on your car payment and the car drives back to the lot.

An anecdotal counterpoint: I experimented with rent-per-minute cars in my local city last year. I learned that almost all my driving was commuting to work or sport practice, so I figured "what would it actually cost to get rid of the car?"

It turns out I paid appreciably less than the self-owned car model by getting an unlimited public transportation pass and using self-driving cars when I didn't feel like riding a bus or tram. Something like 40% less, I don't remember exactly now. I'd previously been driving a 7 year old Ford, so it's not like I'm comparing it to something lux.

Also, you won't have traditional car payments to be late on if it's all Just in Time provisioning and service.

It's all about overlapping layers of service though. Private ownership will make sense for some. Never owning will as well. Robust bus networks, trams/trains in very dense established areas, self-driving (or at least shared economy) cars, and lastly private-owned vehicles all converge to provide a more accessible means of public transportation.

> Or if you are late on your car payment and the car drives back to the lot.

If automated repossession were effective, that would drive down car ownership costs (specifically, finance charges) in a competitive market, since it would reduce the risk of loss which is part of the cost of financing, particularly for people with the worst credit condition (for whom the risk of loss for the financing firm poses a higher portion of the financing costs than other borrowers.)

The barrier to entry into this market is exceedingly low. You only need a few self driving cars and there's your business. Competition will drive down prices.

The barrier of entry isn't the problem. The problem is that this segment is going to reduce options to those of the lower class and poverty class.

Now, it may be of the opinion that these people don't matter. And effectively, they have roughly a 0% economic impact.

Now, prices will be driven down. But so will wages and job losses due to more automation. These people in the lowest segments will be separated their transportation via the ways I mentioned.

Now, are less vehicles on the road good? Possibly. However if the public transportation isn't able to handle it, and these auto-Ubers are too expensive, then it's like having no transportation.

3 is still pretty new. I'm not poor by any stretch of the imagination, but I bought a 9 year old car. It is still going fine (now 12 years old).

Given that adaptive cruise control is still an expensive luxury, I can't see fully self-driving cars coming to the poor any time soon.

That said, if we ever do get totally autonomous cars, you'll probably be able to get a driverless taxi for very little cost without owning a car yourself.

3 year old used car would be off-lease or a retired fleet vehicle.

The average age of a car in the US is 10 years old.

You do realize it's not purely software, right? All cars don't come standard with Lidar, computer controlled braking, cameras, etc. There's a significant amount of hardware needing retrofitted to make it work assuming it's even practical at all.

>> As software is so cheap to duplicate,...

Wait, what?

You're not considering the cost of development.

Reply on news.ycombinator.com