I'm getting ready to go to Novacon (taking tomorrow and Thursday for the trip down), a round trip drive of around 650 miles, and I'm not looking forward to it. Yes, I have a co-driver, and yes, we're not doing it in one run; but the A1/A1(M)/M1 from Edinburgh to Nottingham and back via a family visit in Leeds isn't exactly enjoyable. At this time of year the light up here is fading — it gets dark by 5pm — and the weather is likely to involve clouds, and rain. There are stretches of single carriageway along the A1 north of Alnwick (yes, this is the main highway connecting Scotland's capital city to England in the 21st century: why do you ask?) where one gets stuck behind slow goods vehicles and/or tractors towing tanks of pig slurry. There are stretches further south, where the road rises to the status of motorway (two lanes plus hard shoulder in each direction until you reach Englandshire), and then the happy fun fifty kilometres of roadworks monitored by average speed cameras. I have eyeball-related medical issues that make driving an exercise in applied paranoia, and even when I go sit in the passenger seat I can't read a book — I get motion sickness if I read in any vehicle that weighs less than twenty tons.
The only reason I'm putting up with it is that the alternatives are worse ...
Catching a train from Edinburgh to Nottingham is expensive, limits the amount of baggage we can take, and involves at least one change of train (there are no direct services). Flying from Edinburgh to Nottingham is ... well, there are flights from EDI to East Midlands Airport, which is trying to rebrand itself as Nottingham's airport, causing mirth and hilarity among those who might actually want Nottingham to have a real airport rather than one that's nearer to Derby, but it's not so much use if you're carrying suitcases and looking to save time. Driving, despite the high cost of fuel (petrol currently costs around £1.17 per litre — that translates as US $7.11 per US gallon), is the cheapest way for more than one person to get from Edinburgh to Nottingham, albeit not quite the fastest. (Edinburgh to Leeds can be driven in just under four hours without breaking the speed limit [much]; Leeds to Nottingham is another 90 minutes — the train journey can be done in four and a half hours, including one change.)
Driving may be cheap, but it's fatiguing, blocks you from doing anything else with your time, and is dangerous — much more so in terms of deaths and injuries per passenger-Km travelled than rail or air travel. Why do we put up with it?
The automobile industry spends a lot of money advertising their products (US $31Bn in 2007 — that's $100 per person in the United States per year). As the automotive industry is a mature sector, car marketing is based on market segmentation, pushing different vehicles at specific target audiences by appealing to lifestyle choices and personal self-definition. A lot of advertising relies on the myth of autonomy and freedom; car ads don't show vehicles stuck in traffic jams on the motorway, or commuting to the office, but position them as the means of escape from such mundane aspects of life.
Meanwhile, the death toll mounts. Over 90% of injuries and fatal accidents on the roads are the result of human error, almost always driver error. If you drive, you've doubtless had run-ins with empty-headed idiots who had lost situational awareness, or murderous clowns who had mistaken the public highway for a race track. The accident rate for males suffering from testosterone poisoning (age range: 16-24) is an order of magnitude higher than for those who are old enough to have an appreciation of their own mortality. Here in a nation where we see fit to tightly regulate the ownership of firearms we nevertheless hand out driving licenses to anyone who wants one and can pass a proficiency test simple enough that most 17 year olds pass on the first attempt — even though these vehicles pack as much kinetic energy as an artillery shell (with inevitable consequences).
The death toll is not small. The UK, despite having an excellent road safety record, sees more people killed on the roads every year than died during the entire duration of the Troubles in Northern Ireland; "According to the World Health Organization motor vehicle collisions are the 6th most common cause of death in developed nations, with an average rate of 20.8 per 100,000 populations in the year 2000 (30.8 for males, 11.0 for females). African nations have the world's highest road traffic injury mortality rates." (Wikipedia on traffic-related death rates.) 1.2 million were killed and 50 million injured in road traffic accidents in 2004.
Want a yardstick to put that last figure in perspective?
During the first world war, there were 9.7 million military deaths, 6.8 million civilian deaths, and 21.2 million surviving casualties, over a 4.5 year period. That's an average of 2.15 million dead per year ... but today we have better medical care, and a much higher probability of surviving a bullet or a car crash. If you compare the total casualty rates, then we have one and a half times as many people injured in road traffic accidents every year than were wounded or killed during the entire first world war.
I am all in favour of sporting or recreational driving — as long as it's on a closed track separate from roads where pedestrians, cyclists, and other drivers are trying to get from A to B without injury, death, or delay. But it'd be a very good thing indeed if we could decouple the cultural associations of escape, romance, and autonomy from our transport vehicles. And it'd be an even better thing indeed if nobody in possession of a car was forced to operate the controls and pay attention to the roads if they didn't want to.
While the basic automobile is a mature technology, autonomous vehicles — specifically, self-driving cars — are not. However, they're clearly coming along by leaps and bounds. And unlike human drivers, computers don't generally suffer from lapses of attention, have heart attacks at the wheel, drive home from the pub after a couple of pints too many, or plough into cyclists while texting their girlfriends.
Shortly after (not if, but when) we see autopilots become standard equipment in cars, we can expect to see insurance premiums start to rise sharply for people who insist on driving themselves around on the public highways — especially for third-party insurance.
(Remember, it's not about you: it's about the guy in the pick-up behind you who's had six pints of beer, or the gal in the SUV bearing down on the pedestrian crossing who's paying more attention to the friend she's chatting to than the kids crossing the road. You could be that guy or that gal; or you could be scrupulously attentive the whole time. Your insurance company's computer can't tell until you have an accident ... that's the problem with Baye's Theorem.)
Longer term (I suspect a generation after that point) we'll begin to see pressure to ban humans from driving on the public roads. By this point, the cost of electronics required to upgrade a vehicle to self-driving capability will have fallen so much that it's ubiquitous, even in the developing world.
By around 2050, I'm fairly sure that the human-driven automobile will be a specialised race-track toy for gear-heads, much as horse-drawn carriages in the developed world are a quaint hobby or a deliberate affectation demanded by certain cultural groups (I'm thinking Amish here). Privately owned cars will exist, but will function more like a chauffeur-driven limo. They won't even need to be parked by your house; whistle and it'll come when you need it. Poor folks won't have their own car, they'll just have fractional reserve part-ownership of a vehicle — after all, even at peak rush hour, 95% of the UK vehicle fleet is parked up; we don't need one car per person, we just need available wheels whenever we want to go somewhere. By 2110, I figure driving a manually-controlled car around will be looked on the way we'd look on someone carrying a sword in public; at best it's a weird and archaic affectation, and at worst — call the police!
What are the (science fictional) consequences of assuming we get self-driving vehicles (but not, say, teleportation)?
First, the automobile industry will shrink somewhat. Right now it's still expanding, as the developing world demands mobility — China and India are rolling out vast tracts of concrete and their new middle-class populations are looking for wheels and places to go. But if we look past the climate change and energy crunch politics of the noughties, an end is in sight: a fully developed world and people who are used to having robot chauffeurs rather than doing it for themselves. Ultimately I expect the automobile industry to show the same kind of demographic overshoot as the human population, but to dwindle rather more steeply as it makes a parallel demographic transition into a different kind of technological expression of the human will to motion. I don't know what will power the automobiles of 2110, but if they are self-driving vehicles providing transport for a world of 9 billion humans with a roughly middle-class lifestyle, there will be a lot fewer of them (about 80-90% fewer) than a straight extrapolation from today would indicate.
Secondly ... over a decade, 12 million fewer deaths and 500 million fewer injuries. (We're getting into WW2 territory here.) If we see the shift to autonomous vehicles complete by 2050, then by 2110 there will be 70 million more people than would otherwise have survived, and 3 billion fewer injuries. If the financial cost of treating a road traffic injury is 's extrapolated from the cost of motor vehicle injuries in the contemporary USA ($42Bn for 268,000 survivable injuries in 2005; $158,000 per person) then we're looking at 400 trillion dollars, about ten years planetary GDP or double the value of the entire sunk infrastructure and real estate of the UK — even accounting for the vastly inflated price of medical treatment in the USA, eliminating (or reducing by two orders of magnitude) the damage caused by road traffic injuries would save unimaginably vast amounts of money by the end of the 21st century.
Thirdly: changes in driving culture. Right now, drink-driving is a crime; talking on a mobile phone while driving is a lesser offense in many jurisdictions: we expect drivers to possess a certain degree of health and visual ability, and to be of age. If cars are automatic, however, there should be no problem with driving home from the pub after a lengthy session — as long as the human in question is sitting in the back and the car is making its own way. Ditto, kids: why wait until age 16-18 and take driving lessons and a test, when the car can give anyone old enough to crawl and talk a ride to wherever their parents are willing to let them go?
Fourthly: changes to the structure of cities (which are where, by 2110, the vast majority of the human population will live). If we have autonomous cars, we don't need to park them next to our front door; they'll come when we call them. Much better to park the expensive asset on the outskirts of town, in multi-story facilities with guards to keep an eye on them and car wash, valeting, refuelling and maintenance bays on the premises. This makes life much nicer for those of us who live in cramped but walkable pre-automobile cities, and means new residential developments can be designed that have better local amenities and permit walking and use of shared public spaces while not forcing residents to give up their precious four-wheeled sense of autonomy.
This is going to have long-term social consequences: what happens if we re-acquire a local sense of community while retaining the distributed global communities we gained from the internet and the ability to travel point to point that the automobile brought along?