2016-06-20

Autonomous Cars and Autonomous Ownership

I was originally going to do a Linux distribution review this month. However, when I tried a couple of distributions that I wanted to test, none of them would properly boot from a live USB, so I gave up on those. Instead, I wanted to use this space to ramble a bit on what the near-future of self-driving cars might look like. It comes from some conversations I had with my family last weekend while visiting California, after having seen the limited self-driving capabilities of a Tesla Model S (namely, its ability to autonomously pull in and out of a parking space). Moreover, as some of you who know me personally would know, I have a disability that prevents me from driving, so the sight of even minimally-autonomous cars as a present reality excites me, and I'm keeping an eye on current developments in that field/market. Given this, if you'll indulge me, then follow the jump to (not exhaustively) explore some possibilities for self-driving cars.

There are a few potential developments to consider with respect to autonomous cars. One is that cars could be networked with each other to a much greater degree than is seen today. Another is that cars could be connected to their manufacturers to a greater degree than is seen today. Finally, models of car ownership could change quite a bit from the current model of individual private ownership, which still drives (pun slightly intended) the vast majority of car purchases, notwithstanding private taxi and more recent ride-sharing services. (There are plenty of other new issues to reckon with as well; again, this is not an exhaustive list, but merely a small sampling of the issues I imagine we could face once autonomous cars become better-developed.)

What could happen with respect to networking between cars? I imagine that initially, autonomous cars will have to be built with the assumption that they are the only such autonomous cars on the road, and all of the other vehicles and road features are as they are today. This will require a ton of sensors and processing power to deal with unexpected road closures, bad [human] drivers, and so on. However, farther down the line, as autonomous cars become more common, I hope that standards will be developed for cars from different manufacturers to be able to reliably communicate with each other, sharing computational power among clusters of nearby cars on a road in a networked fashion and thereby decreasing the general sensing and processing loads on any given car; although I am a little pessimistic about the ability of different car companies to cooperate and develop truly open standards while remaining competitive in the marketplace (i.e. as opposed to the market devolving into a monopoly characterized by proprietary standards, which would be but one symptom of the large barriers to entry), we do have things like USB and PDF as open standards, so I do hold out hope that this may happen. Even farther down the line, it is possible that governments at the state and local level could augment roads with devices that communicate with autonomous cars in order to further reduce the sensing and computational loads on cars themselves and provide even more accurate real-time information. This doesn't have to necessarily be done on every side street, but could be done first on interstate highways, US routes, and busy local highways, before coming to more minor roads; that said, given the near-total lack of political willpower to fund even basic upgrades for serious problems with roads and bridges that everyone on either side of the political aisle agrees exist, I'm not holding out hope for further road augmentation like this for a while.

What could happen with respect to communication between a car and its manufacturer? Given how large and influential these companies can be, and given that product differentiation precludes perfect competition even when many companies are present in the market, this may be the biggest problem to reckon with. Already, companies like John Deere and General Motors have been abusing the anti-circumvention provisions of the DMCA (credit to this article by Mike Masnick on TechDirt) to broadly claim control over vehicles that people buy, thanks to the inclusion of computer software on said vehicles developed by the corresponding companies. In particular, these companies claim that they are the actual owners of the vehicles because they own the copyright on the software integral to those vehicles' functioning, and that the supposed owners are merely licensed to use the cars indefinitely, but this license can be taken away by these companies at any point if the companies don't like what owners, who should really be called "users" at this point, are doing with the vehicles. (As an aside, I just realized that this reminded me of the issue in the Harry Potter series of novels in which goblins believe that the original manufacturer of an object is its true owner in perpetuity, so humans are merely borrowing goblin-made objects from the manufacturers/"true owners"; perhaps J. K. Rowling was inspired to put that detail into the books in part thanks to laws similar to the DMCA in the UK.)

So what would happen when cars become autonomous? I fear that the above issue will only get worse, especially if cars continue to be privately owned primarily by individuals and families. I'm reminded of this XKCD comic (credit to Randall Munroe), whose mouseover text imagines a future where an autonomous car will dutifully take its human passengers to most places, but which will lock its human occupants inside until they capitulate (or worse, die) if they try to take it to a dealership or other place for it to be sold. I am hopeful that lawmakers will eventually be able to pass laws forbidding companies from programming their cars to allow this to happen, though I'm not necessarily hopeful that lawmakers will have the foresight to act on this before an actual incident like this happens. That said, although artificial intelligence/machine learning is a somewhat separate issue from autonomous driving (as this comic envisions the dystopian hostage situation as being due to a car's artificial intelligence), I could still imagine it happening just by virtue of the car communicating with its manufacturer. More broadly, manufacturers would have an almost unprecedented degree of control over what the users of its cars can do, if the future of car ownership is the same as it is today (namely, dominated by private ownership by individuals or families) and if the car market is the same as it is today (rather differentiated, with not enough companies to allow for perfect competition that would allow a company to enter the market and succeed by promising to not screw over customers); I figure it would be like the General Motors or John Deere situations on steroids.

For example, when my family and I visited north India 6 years ago, we had tour guides who were generally pretty good but were fairly pushy when it came to us going to particular restaurants and other eateries, because they got commissions only from certain restaurants. Usually, going to their preferred restaurants worked out fine for us, but even on the one or two occasions when we didn't want to go, we were able to go to the restaurant we wanted to for that meal and then go to their preferred restaurant for the following meal or two. Sure, they'd simmer in frustration about it at that point, but what could they do? Those tour guides were individuals, and there were several of us traveling together; plus, at the end of the day, those tour guides were otherwise generally reasonable people, so we wouldn't hold something like this against each other. What would happen when car passengers have to deal with corporations though? I could imagine that car companies would start making deals with corporate eateries as well as smaller restaurants to guarantee them business in exchange for a cut of money, and if cars are owned by individuals or families, it would be much harder, if not impossible (or even illegal, thanks to the DMCA anti-circumvention clauses), to override the programming of the car and make it go to an eatery that does not have a partnership with that car company. This could be more broadly applied to any business, and more particularly to repair shops (as has already been demonstrated by John Deere or General Motors). It almost seems like the closed app ecosystem development model come to life: it wouldn't just be the case anymore that people are restricted in the apps that people can get and use on their smart devices or in the content that people can access through certain platforms (e.g. how not all movies are available on Netflix, but those that aren't may be available on lesser-known/lesser-used platforms).

What could happen with ownership? It's possible that individuals and families could still buy one or two cars for a household just as they do today. As I've detailed above, though, the level of control that car companies could have over autonomous cars would be terrible for consumer rights if ownership were to remain individualized. Beyond those issues, though, I also don't see this as a trend that is necessarily in line with current trends in housing. Cities are becoming easier to live in every day, and even many formerly sprawling suburbs are becoming denser to attract young people and their families. Such dense developments also push to be easier for walking and biking, and often pursue public transportation options more seriously; for whatever options are pursued, it is typically the case that cities and dense suburbs now eschew large amounts of parking, making it hard to keep a car there, and thereby making it harder to justify owning a car as an individual (or family) when the vast majority of its time will be spent sitting idly in a lot/garage/driveway.

What else could happen with respect to ownership then? It seems more likely to me that the model of car ownership would change quite a bit with the advent of autonomous cars, and there are three thoughts that I have on this in particular (that are, again, by no means exhaustive). One is that most private car ownership would shift toward models based on ride-sharing services like Uber or Lyft. I don't know if this would lead to increased competition or not with respect to ride-sharing services, but if ride-sharing services still do exist, I think they would be beneficial by pooling the market power of the consumers to ensure that car companies can't do some of the things that I suggested above (like restrict where consumers can go, or even lock consumers in the car). Moreover, because consumers won't generally own these cars, issues like repair costs will become moot points, and consumers will be freer to live in places where storing a car is difficult in the long term. The one big issue I could see is that families with young kids would require specialized equipment like child seats, so it would be a huge drag to have to insert, remove, and carry around such equipment if the car isn't the family's own. The workaround that I can think of is that a ride-sharing service could stock up on sufficient child seats, and a family can then order a car to come with child seats when it comes to pick them up (and people, or robots for that matter, at a certain location whence these cars are dispatched can insert or remove child seats and other such equipment as needed).

The second thought is that private ownership could shift toward ride-sharing, but cutting out the middleman, so that car companies are directly in control of the cars going around, although individuals or families won't generally own these vehicles. If every company services every residence, the competition between companies in a nearly-free market would ensure that car companies can't arbitrarily restrict their customers from certain destinations. The more likely scenario as I see it, though, is that car companies will restrict themselves to serving only subsets of residential areas, much as how ISPs have local monopolies and generally don't compete with each other for the same residences. In such a case, it would probably come to the government to enforce rules akin to net neutrality ensuring that autonomous cars from any company treat all (literal, not digital) traffic, sources, and destinations on equal footing.

With either of the above two options, ride-sharing would be operated almost exclusively privately, and this would probably seriously eat into (if not entirely supplant) the ridership of public transit. Either poor people would be forced into using these ride-sharing services, which may be too costly for them, or tax money would go into continuing to fund public transit, which would make ride-sharing customers unhappy as they aren't really being served by public transit (unless somehow autonomous cars do not significantly mitigate traffic issues, in which case public transit, especially in rail form, would continue to ease the burden on roads). The third option then, which doesn't necessarily have to exclude the above two options, is that government could essentially augment public transit (or, from another view, compete in a very specific niche of the ride-sharing market) and thereby solve the last-mile issue. (For those who haven't heard of it, the last-mile issue in public transit is the problem of how to quickly and effectively get people from their homes and places of business to public transit routes, because public transit tends to run on fixed routes; this is an even bigger problem with regard to rail, because while buses have the flexibility of being able to run on any road, trains can only run on rails, which cannot feasibly be present everywhere.) In particular, the local government could operate fleets of autonomous cars to efficiently and quickly get people from their homes or places of business to rail or bus routes. I don't envision that the local government would get rid of its traditional public transit systems (bus or rail, though those would likely become self-driving as well), even if its operation of autonomous cars is essentially like a ride-sharing service; given how much more efficiently buses and trains can transport large numbers of people, I think it would not be in the government's interest to have its autonomous cars make full journeys that could be covered by bus or rail service, as that would further clog the streets (as roads cannot have arbitrarily high speed limits by their construction, they also then cannot have arbitrarily high throughput). Of course, this would also require revising bus and rail routes and timings to more accurately reflect the entries and exits of people from transit stations, given that different people might be walking, biking, or taking autonomous cars to get to these stations, but those in cars probably won't have to deal with traffic problems as much as is the case today.

I haven't even touched on the other major issues with autonomous cars, such as security/hacking. I don't know what the solutions to these issues will be, but I do think it will be important to at least grapple with these issues before it is too late, rather than "letting the market sort it out"; this development isn't occurring in a vacuum, and major car and tech companies are already shaping autonomous cars to suit their existing business models, so I don't think it's realistic to assume that everything will be hunky-dory with car companies, ride-sharing services, and consumers all engaging in perfect competition to yield the best possible experience for consumers. Anyway, that's all I have to say about this issue for right now; I'd be interested to hear from you, the reader, so please feel free to share any of your thoughts on the matter in a comment below!