Snark versus Trains

by Henry on October 16, 2013

shark-vs-train-cover

“Notorious technophobe Luddite”:http://en.wikipedia.org/wiki/Ethan_Zuckerman “Ethan Zuckerman”:http://www.ethanzuckerman.com/blog/2013/10/15/google-cars-versus-public-transit-the-uss-problem-with-public-goods/

“I don’t want a Google car,” I tell her. “I want a train.” … There’s something very odd about a world in which it’s easier to imagine a futuristic technology that doesn’t exist outside of lab tests than to envision expansion of a technology that’s in wide use around the world. How did we reach a state in America where highly speculative technologies, backed by private companies, are seen as a plausible future while routine, ordinary technologies backed by governments are seen as unrealistic and impossible?

… My student Rodrigo Davies has been writing about civic crowdfunding, looking at cases where people join together online and raise money for projects we’d expect a government to otherwise provide. On the one hand, this is an exciting development, allowing neighbors to raise money and turn a vacant lot into a community garden quickly and efficiently. But we’re also starting to see cases where civic crowdfunding challenges services we expect governments to provide, like security. Three comparatively wealthy neighborhoods in Oakland have used crowdfunding to raise money for private security patrols to respond to concerns about crime in their communities. …

… On the one hand, I appreciate the innovation of crowdfunding, and think it’s done remarkable things for some artists and designers. On the other hand, looking towards crowdfunding to solve civic problems seems like a woefully unimaginative solution to an interesting set of problems. It’s the sort of solution we’d expect at a moment where we’ve given up on the ability to influence our government and demand creative, large-scale solutions to pressing problems, where we look to new technologies for solutions or pool our funds to hire someone to do the work we once expected our governments to do.

{ 95 comments }

1

Phil 10.16.13 at 1:29 pm

In response to the first paragraph, Debord vous l’avait dit before I was born:

“A mistake made by all city planners is to consider the private automobile (and its by-products, such as the motorcycle) as essentially a means of transportation. In reality, it is the most notable material symbol of the notion of happiness that developed capitalism tends to spread throughout society. The automobile is at the heart of this general propaganda, both as supreme good of an alienated life and as essential product of the capitalist market”

Shorter: trains are no fun! You don’t sit on a train past the Stop’N’Shop with the radio on, now do you?

2

Barry Freed 10.16.13 at 1:44 pm

It is, however, the only legal way to go a hundred mile per hour.

3

Tom Slee 10.16.13 at 1:51 pm

Much needed points, well made.

Innovative community provision of necessary services has a long tradition of course, sexual assault centres being just one example that comes to mind. But this thing of looking exclusively to markets to “scale up” community-based innovation is walking a road to hell paved with mercenary intentions.

The Google car initiative is not crowdfunding of course.

4

William Timberman 10.16.13 at 2:28 pm

See the remarks about government and IT projects from the preceding thread. On the one hand, you’ve got the Right’s government-is-bloated-and-inefficient, and on the other, the Left’s privatization-is-economic-warlordism-and-ultimately-anarchy. In the middle, there’s a herd of bewildered but still selfish consumers, whose panic increases as its economic wherewithal shrinks.

There’s some room for optimism, I suppose, in that a lot of people are attempting a serious re-think, but most are constrained by circumstances — chiefly, but not exclusively economic — to the parochialism of limited possibility. Does this all come together at some point? Maybe, but we don’t have a lot of time.

5

Wonks Anonymous 10.16.13 at 2:29 pm

Car vs train and public vs private seem like somewhat different issues. Roads are (usually) publicly provided, busses (similar to a car in most relevant ways) are often provided publicly. Many trains are private (Berkshire Hathaway owns some lines). The fact that we’ve already got a lot of paved roads means it’s easier to extend that system using self-driving vehicles.

6

Brett 10.16.13 at 2:56 pm

Trains seem to run into more issues with local control/NIMBYism than highways, for reasons I can’t really explain except possibly that more people drive cars. Even if 90% of the people along a proposed train route think it’s the way to go, there’s the remaining 10% who don’t want it running through their neighborhood, so you either have to override their desires or spend the money to detour the route around them (and unlike the late 19th century wave, railroads aren’t getting paid money based off of the length of track they lay, so this is a money-loser). That usually adds some serious legal expenses in court, not to mention the years spent waiting for these things to get settled.

@Wonks Anonymous

Many trains are private (Berkshire Hathaway owns some lines).

It seems to be working well with freight rail after the removal of many regulations after 1980. I’m not sure about passenger rail, though – if I remember right, passenger traffic usually only covered the operating costs at best even back when trains were fastest and cheapest way to move people over land.

Personally, I’d be fine with privatizing Amtrak, while also allowing them to collect a subsidy and/or tax credit to maintain their rail lines (which is only fair when you consider that roads are publicly paid for in this country).

7

ajay 10.16.13 at 3:09 pm

Trains seem to run into more issues with local control/NIMBYism than highways, for reasons I can’t really explain except possibly that more people drive cars.

Just speculating here: maybe because it’s rarer to build a completely new highway than it is to lay a completely new railway – highways maybe tend to follow the routes of existing roads? Certainly building new bypasses etc attracts plenty of protest over here (UK).

8

Scott P. 10.16.13 at 3:21 pm

In the U.S., new highways get built all the time, granted usually not in dense urban areas…

9

Brendan Taylor 10.16.13 at 3:29 pm

The excerpted bit runs parallel to a recent piece by Evgeny Morozov:

“[T]echnology” is just a slick, depoliticized euphemism for the neoliberal regime itself.

http://blogs.sueddeutsche.de/feuilletonist/2013/10/10/ghosts-in-the-machines/

10

Kenny Easwaran 10.16.13 at 3:45 pm

Obviously the US needs some high speed rail, and in general, better public infrastructure, and especially for transportation. But Albany to Boston is not one of the highest priority lines, and judging by the map he links, it’s just too crooked a right-of-way for trains to be able to manage speeds much higher than freeway driving (especially since I-90 is a much straighter right-of-way than this proposed train route).

The thing that is most likely to actually speed his commute is a bus, but buses aren’t considered “sexy” like trains. Fortunately, buses can benefit from Google’s self-driving technology – all of a sudden, the biggest expense in public transportation will be gone, which will allow for huge expansion, especially for urban buses in off-peak hours.

11

Josh G. 10.16.13 at 3:51 pm

Call me crazy, but if I was designing a blueprint for America’s future, it would include both a robust rail network and self-driving cars. I don’t see why these things have to be mutually exclusive.

Zuckerman is right that the Reagan/Thatcher neoliberal consensus has made it far harder than it should be to provide public infrastructure. But the American left is not completely free from blame, either. David Frum had an excellent chapter in How We Got Here regarding the way our policies on infrastructure have evolved. During the mid-century period, the majority got to build stuff, and the minority that didn’t like it (or was in the way of construction) was out of luck. Around the 1970s, things changed so that NIMBYs and environmentalists could much more easily throw sand in the gears. Projects that once would have taken a few years at most and cost a couple of million dollars would now stretch out for decades and cost billions. If we really want a good public transit system in the U.S., we will have to ditch environmental impact statements (or at least substantially streamline them) and ram projects through quickly over the objections of local residents. The reason why European and Asian countries do better on these fronts isn’t just because they tend to have higher population densities, it’s because they operate in a manner more similar to the mid-century U.S.: visible, accountable public officials can make decisions that stick, without a bunch of crappy local veto points.

12

mpowell 10.16.13 at 4:33 pm

I found the story a little odd. A self driving car can take you the whole way. A train will not get you door to door unless you are very lucky. Both have their advantages, but I’d focus first on not having a severely long commute.

A big part of the difficulty with infrastructure projects in the US is that they have gotten too expensive. And this applies to both roads and rail. I don’t think that is really a left/right issue either.

13

Alex 10.16.13 at 5:01 pm

I don’t see why these things have to be mutually exclusive.

Think of the quote about the 1960s MLF project being the first time in history a fleet of imaginary ships sank another fleet of imaginary ships. Self-driving cars as an idea exist to prevent public transport. See also, projects for an airport in the Thames Estuary, which have historically been proposed by people who want to prevent airports elsewhere rather than particularly wanting an airport there.

14

chris 10.16.13 at 5:48 pm

busses (similar to a car in most relevant ways)

I’d say they’re more similar to trains. At least for people who are inclined to strongly prefer cars to trains, the attractions include:
– Your car is your castle — you don’t have to share it with anyone you didn’t specifically invite
– No need to adhere to a prearranged schedule
– Never have to transfer between routes to reach a destination
– If not door-to-door, at least parking-lot-to-parking-lot service
– No fixed service area; ability to take long trips is limited only by your time and fuel
– Can change destinations on a whim
– You can drive your car at any time of the day or night, any day of the week, even on holidays

Now of course trains and buses have other advantages and how to balance that against the advantages of cars is somewhat of a judgment call (especially, the availability of both bus/train/subway stops and parking lots where you are going has a big impact on which is likely to be more convenient). But I think in most of the important respects as a user, buses are more like trains than they are like cars.

Self-driving cars as an idea exist to prevent public transport.

This seems hard to believe — even if the government handed one out to every citizen, which seems unlikely, I still don’t see how they would solve the traffic congestion and parking problems of dense cities. Cars may be convenient in sprawl hell, but they just don’t scale well compared to genuine mass transit.

15

Igor Belanov 10.16.13 at 6:18 pm

@14

That’s the whole point- it’s the idea rather than the practicality of ‘self-driving cars’ that is doing the work of sabotaging public transport alternatives.

Dr Beeching and his ilk did great work in the UK in persuading people that trains were hopelessly old-fashioned. This was in the same period that the Japanese Bullet Train and the TGV were under development.

16

mpowell 10.16.13 at 6:33 pm

Igor @ 15: This narrative just sounds absurd to me. People have been talking seriously about self driving cars for what, two years now? And they are mostly being used to sabotage public transport alternatives. It’s Google that put self driving cars on the map. They’re not doing it to kill public transportation. They’re doing it because they want to (and are) building self driving cars.

17

Chatham 10.16.13 at 6:42 pm

The importance of self-driving cars gets exaggerated. It seems the benefits would be comparable to living in places where cabs and delivery services are cheap. While it’s convenient, it doesn’t change your life.

18

Josh G. 10.16.13 at 7:59 pm

Alex @ 13: “Self-driving cars as an idea exist to prevent public transport.

This strikes me as absurd on its face, a conspiracy theory with nothing to back it up. First of all, self-driving cars are being developed and advocated by Google more so than by any particular car company. What interest would Google have in fighting against public transit?

Secondly, anyone familiar with the geography and demographics of the U.S. understands that public transit is not and cannot be a panacea. To be sure, we have plenty of towns and cities with enough population density to support passenger rail networks within, and between, them. We are way underdeveloped in this arena (there should probably be a high-speed rail network running north-south along the entire East Coast, and that’s just for starters). But we also have a lot of people living in distant suburbs, exurbs, and rural areas where public transit simply isn’t practical. For these, self-driving cars are going to be the best option for the forseeable future. Self-driving vehicles will be more convenient, far safer, and probably more fuel-efficient than manually driven cars.

19

mud man 10.16.13 at 8:02 pm

Self-driving cars seem to have done the work of sabotaging discussion of crowd-sourced civic infrastructure.

What I want to ask is, isn’t crowd sourcing sorta like how local government is supposed to work? Consent of the governed, and all that …

20

Igor Belanov 10.16.13 at 8:04 pm

I was referring to the whole ideology of the motor car, of which the ‘self-driving car’ is just one step further. It has yet to prove the panacea it was supposed to be, so needs constant propaganda and the promise of technological progress to rubbish the notion that there might be alternatives.

21

Chatham 10.16.13 at 8:35 pm

For these, self-driving cars are going to be the best option for the forseeable future.

Manual cars will be the best option for the foreseeable future. Self-driving cars might prove a bit more convenient at some point in the future, but a lot depends on how they actually handle in the real world (google glasses seemed great before they were tested by people outside of Google) when the price will come down to a reasonable level.

22

chris 10.16.13 at 8:44 pm

What I want to ask is, isn’t crowd sourcing sorta like how local government is supposed to work? Consent of the governed, and all that …

Free riding is the difference between democracy and crowdsourcing. Once a democracy passes a law, even those who disagree with it must obey it. “Consent of the governed” was never intended to mean that individuals could choose which laws not to consent to; that would lead to thieves refusing to consent to laws against theft, etc.

23

Tom Slee 10.16.13 at 9:06 pm

Side question on self-driving cars. What happens when someone steps out in front of one and the options are to hit the person or swerve into another lane? Or when a deer runs in front of one? [Genuine question, not snark]

24

hix 10.16.13 at 9:31 pm

The biggest downside of individual motoriced transport is the high accident risk. So, yeah, self driving cars would fundamentally shift the balance against public transport. There is a broad spectrum between self drivintg and entirely driver controled. Electronic driving aides aviable in luxury cars today do improve safety already. The trend at the moment is pretty much anti car. Its a good time to be young – smartphones have replaced cigarets and cars as the cool thing. The share of the below 26 year olds with a drivers license went down a lot in the last decade.

25

Metatone 10.16.13 at 9:40 pm

@Tom Slee

We’re in Asimov territory. Lots of interesting stories to be written.
The practical answer however is that either self-driving cars will get accident-exemptions (particularly regarding insurance and liability) or they will be DOA for general use. It will be the mother of all lobbying battles, one suspects.

26

Metatone 10.16.13 at 9:45 pm

Stepping aside from the question of particular train lines in New England, I’ve been banging on about the general point for a while now. We live with the legacy of Hayek, which is a special kind of learned helplessness, where we’re incapable of conceiving infrastructure solutions to problems, instead we invest all our faith in local solutions (because that’s where the knowledge is!) and expect them to scale.

Only for many things it just doesn’t work, because power and information are both distributed unevenly and while “Teh Internets!!1!” have altered some of the distribution of power and information, there’s little evidence to suggest that it has actually made the distribution any more even.

27

guthrie 10.16.13 at 10:33 pm

Tom Slee #23 – I expect the car to hit the person or deer. But you’d have to speak to the programmers to find out what they want it to do. Other options include bonnet mounted airbags.
I’d also expect there to be cameras recording everything that happened, so that if someone does take things to court, the car’s reactions (probably faster than a humans) would prove that there was nothing else to be done. It’s certainly not an insurmountable problem.

28

PJW 10.17.13 at 12:13 am

New trolley problems loom. Interesting thread.

29

bxg 10.17.13 at 3:15 am

> New trolley problems loom. Interesting thread

But wouldn’t someone have said exactly this (i.e. trolley problems, yay!) considering a move from a world with no/few manual cars to one where everyone drives? Thinking about this shift, one might be terribly concerned with asking: “If someone jumps in front of you while driving, but to avoid them you swerve put X other lives at risk, don’t you need to know when that is acceptable?” Shouldn’t guidelines be written into the vehicle code? Shouldn’t there be extensively established case law to help you out with such dilemmas: crisp enough we can convey the general principles at driver’s ed?

And yet somehow, although we (rightly) throw vast legal resources at automotive-related issues, we muddle through with no great (widely appreciated) clarity on how a real human driver, today, should tackle trolley problems.

Maybe these issues don’t come up so much, or many they are so situation specific that it isn’t that important: perhaps we are ok if a driver behaves broadly rationally, broadly non-negligibly, and it’s either not possible or not worthwhile to pin things down more precisely.

In short: why hase trolley problem philosophy had so little impact on human driver education or legislation or behavior today? Whatever your answer is, why would the situation be so very different for autonomous drivers?

30

RSA 10.17.13 at 4:03 am

@Tom Slee:

What happens when someone steps out in front of one and the options are to hit the person or swerve into another lane?

Very interesting question. Computer scientists have been thinking about risks for decades; for example, Peter G. Neumann and others have written a column called Inside Risks for the Communications of the ACM, the professional organization for computer scientists, since 1990 (and maybe earlier). Topics have included security, privacy, voting, computer addiction, device controllers… you name it.

Self-driving cars aren’t really a new kind of problem, but I do think it will be tough to address for reasons beyond the purely technical. Here are a few refinements on your question: If the self-driving car is utility-based agent, which utilities should be associated with hitting a pedestrian and with swerving, all else being equal? (It would eventually have to come down to numbers.) If the car gets in an avoidable accident, how should fault be apportioned between the person in the nominal driver’s seat, the buyer, the car manufacturer, the software company, an algorithm designer, a programmer, and so forth? (The conventional End User Licensing Agreement won’t fly here.) Would hacking make the roads more dangerous than they are already?

It’s a huge can of worms in part because (a) we don’t have a lot of experience putting systems in the hands of end users that can make decisions that could result in deaths, (b) people tend to underestimate the time and cost of developing good software, and (c) the relationship between software and the law is currently a mess.

31

Zamfir 10.17.13 at 5:30 am

Unlike hix, I would expect self-driving cars to be somewhat good for public transport (though I don’t expect to see fully self driving cars for quite some time).

For one, renting a self-driving car would be an attractive option, making it easier to go without car and rent one when convenient. There’s a car rental company around here that parks cars all kver town, so there’s always one free near you. But that free one still takes time to get to, and it implies that its fleet goes almost as underused as private cars.

A self-driving rental should be able to offer lower prices and more choice in vehicles. That would still be more attractive for public transport to compete with, compared to private cars where people have already sunk most of the cost.

And second, I bet that self driving cars would take some of the magic of cars away, the magic that tempts many people to buy the nicest car they can. That goes double for affluent (and influential) parts of society, whose cars are typically much nicer than public transport can offer as alternative. I know quite some people who just don’t get the point of high speed trains, because it covers exactly the trips where they can enjoy their BMWs most.

32

js. 10.17.13 at 5:41 am

we muddle through with no great (widely appreciated) clarity on how a real human driver, today, should tackle trolley problems.

If someone (or some deer) jumps out in front of your car, you don’t really have the time to make a rationally considered choice. So the trolley problem doesn’t arise (if it ever does).

(Also, somewhat OT, but am I the only one who finds the idea of self-driving cars to be the worst of all possible worlds: you’re still going to have me travel in an isolated little box, but you’re going to take away the one genuine upside of traveling in an isolated little box, which is you know the actual driving.)

33

ajay 10.17.13 at 8:47 am

you’re still going to have me travel in an isolated little box, but you’re going to take away the one genuine upside of traveling in an isolated little box, which is you know the actual driving.

Not everybody actually enjoys driving all the time. That’s why you have to pay people to do it, and why one of the perks of being rich is having a driver.

If someone (or some deer) jumps out in front of your car, you don’t really have the time to make a rationally considered choice. So the trolley problem doesn’t arise

Well, this is actually a really interesting point. If you’re making a choice without conscious thought, are you still making a moral choice? If a child jumps in front of my car, I’ll stop instinctively, even if this means a crash stop that risks injuring myself. If a pigeon flutters in front of it, I won’t. Now, I don’t have the time to make a considered moral choice about the correct value of the life of a child vs. that of a pigeon, but none the less I am still making a snap decision that reflects the conclusion I’d reach if I had time to think about it. And if I then went away and got argued into believing that actually a pigeon’s life is just as valuable as a child’s, I daresay I’d start crash stopping for pigeons.

34

Mao Cheng Ji 10.17.13 at 8:50 am

Don’t be silly, Google will implant a microchip in every deer.

35

Niall McAuley 10.17.13 at 8:53 am

Tom Slee writes: What happens when someone steps out in front of one and the options are to hit the person or swerve into another lane? Or when a deer runs in front of one?

Firstly, the self driving car will be morre likely to be aware of the person/deer before it jumps out than a human driver would, since it’ll have 360 degree always-on sensors.

Secondly, it’ll react a lot faster than a human would.

Third, it’ll always do the same thing, not panic.

As to what that thing is, I don’t know exactly . It’ll probably brake a lot harder than humans do, a lot earlier. It’ll probably hit the deer, and try and miss the human.

36

Alex 10.17.13 at 9:03 am

You have all read Bruno Latour’s Aramis or the love of technology, I hope…in many ways, self-driving cars as public transport is another way of saying “a Personal Rapid Transit system”, something which is famously near-impossible to make work for a variety of interesting sociological reasons.

The genesis of PRT was all about the urban crisis! and the 1970s! and there are kids on my lawn! black ones! and we need to find a way of getting downtown to the AT&T Tower without sitting in traffic for hours two or being…on a bus…with those people!

I suspect its rebranding as self-driving car is an epiphenomenon of the general rightward march of the Overton window – it’s not just enough to have public transport that doesn’t require you to interact with the public, it’s necessary to deny that it is public transport at all, because public transport is definitionally used by those people!

The Aramis engineers originally aimed to have a 2-person car, but for various reasons they ended up having bigger, shared ones, with the result that test passengers pointed out that you were still faced with those people! but just stuck in a capsule rather than in a roomier metro carriage.

Anyway, the whole idea reflects out of date suburbanist/urban decline assumptions. Nobody buys this stuff any more, in the most literal sense; mcmansions and SUVs don’t sell, big road projects have been brutally opposed for 30+ years now, city centre populations are rising.

It’s very telling that the backers all live in Silicon Valley, where you can be 20 years in the future technologically (insert robot here) and 40 years in the past sociologically (it really is all white men in cars) whenever you want:-)

37

Pete 10.17.13 at 10:14 am

@Alex: ‘Aramis …’ is an excellent read.

highly speculative technologies, backed by private companies, are seen as a plausible future while routine, ordinary technologies backed by governments are seen as unrealistic and impossible

Let me rephrase that to “highly speculative technologies, backed by private companies, are seen as potentially profitable while routine, ordinary technologies backed by governments are known to be unprofitable. There was a time when trains were the speculative new technology, with potential huge revenues from all the trade they would enable.

Self-driving cars will start as a gimmick, but there are plenty of non-gimmick legitimate use cases for the technology: mobility for the disabled and old. Taxis with zero risk of sexual assault by the driver. And long-distance frieght that isn’t limited by the requirement to let the driver sleep every 8 hours. Obviously they’re not testing the technology with articulated container lorries yet, but there’s no reason for it to be too different. Tricky reversing or sections at the start and end can be handled by handing it over to external control by a banksman.

Public transport seems to be one of those things that can be done well or badly depending on whether those responsible actually care about making it work or not. There’s also a tipping point effect of density; it becomes just too much hassle and expense to get into the city and park, so middleclass people get on the public transport and demand that it works.

38

Zamfir 10.17.13 at 10:22 am

Alex, are you sure it’s just sociological reasons? The great attraction here is driverless point to point transport, which is at least partially a technological problem. After all, taxis exist, car rentals exists. What are the sociological barriers facing a working self-driving car rental, that are not faced by those?

I am skeptical that the current state of the art has actually solved that problem as well as some claim, when let loose on a real world with ordinary people. But at the same time it’s a lot, lot better than in 1973. Unlike some AI directions, this one shows genuine progress.

39

Phil 10.17.13 at 10:24 am

As to what that thing is, I don’t know exactly

That is rather the point. What we’ve got at the moment is driven by an unpredictable combination of un- and semi-conscious calculation and estimation, rational forward planning on a variety of timescales, muscle memory, learned reflexes (slamming on the brake), instinctual reflexes (flinching & hence swerving), knowledge of particular road conditions, knowledge of the legal system, varying degrees of trust in other drivers, a sense of what you as a driver owe to other drivers, and a constantly-updated reflective sense of what’s at stake in any given situation – including what the driver will be accountable for if anything does go wrong. And if an accident happens despite all of that, we roll all of those factors up into the assumption that you as a person were a competent driver and are therefore responsible for what happened to this other person, unless you or your legal representative can prove otherwise. I’m deeply sceptical about the feasibility of replicating all of these factors in software, and rather concerned about what would happen when things went wrong – who would be held responsible?

40

Alex 10.17.13 at 10:54 am

Obviously they’re not testing the technology with articulated container lorries yet

You know, there’s this really awesome technology for long-range, huge-capacity intermodal…it’s called a train, and Warren Buffett makes a ton of money from them.

But at the same time it’s a lot, lot better than in 1973.

Well, the one at Heathrow T5 actually opened. Months late, but it opened, which is indeed vastly better than all other PRT projects. Mind you, I’ve never had the chance to use it because it goes to the car parks and I get to the airport…on the tube.

41

Pete 10.17.13 at 11:00 am

“who would be held responsible”

I suspect the situation would differ little from the current one: hardly anyone held responsible, unless it’s really obvious that they did the wrong thing, and maybe the victim’s family get an insurance payout.

A common complaint in the UK is that deaths of cyclists are not taken seriously. This seems to happen in the US too: http://road.cc/content/news/46285-price-cyclists-life-42-washington-state

(I think people are assuming that self-driving cars will be held to the “chemicals and toys” level of product liability, not the “cars and guns” level)

42

Pete 10.17.13 at 11:04 am

You know, there’s this really awesome technology for long-range, huge-capacity intermodal…it’s called a train

So why are the roads around me clogged with container lorries? Why is UK freight 90% by road? ( http://www.imeche.org/knowledge/themes/transport/freight/rail-freight ) Why is the UK rail network so abominably expensive?

43

Zamfir 10.17.13 at 11:40 am

A prt to car parks is not point to point? That’s the promise of Google cars and the like. A PRT from a parking lot to a terminal is just a bus for people who cant stand to share their bus with strangers. One that picks you up at home has value beyond that.

44

ajay 10.17.13 at 1:11 pm

So why are the roads around me clogged with container lorries? Why is UK freight 90% by road?

Not actually evidence that rail isn’t awesome for long-range high-capacity freight; just suggests that a lot of UK freight is short-range. Tonne-kilometre comparison isn’t the best here. You’d want to look at market share over the same sector. Might as well ask “if jet airliners are so good at taking people from London to New York, how come everyone around me seems to get around by bus?”

45

sanbikinoraion 10.17.13 at 1:43 pm

Eventually, you won’t be allowed to drive on the self-drive-only highways, because they’ll be proven to be so much more reliable, fast and safe compared to meat-driven cars. Governments can, will, and should tax the bejeezus out of the fewer and fewer dinosaurs who continue to insist on driving their own vehicles.

46

Wonks Anonymous 10.17.13 at 2:49 pm

Pete, from what I’ve heard Europe’s rail system is more devoted to passengers, shifting freight to trucks. America is the reverse.

47

ajay 10.17.13 at 3:01 pm

46: and also shifting a lot of freight to shipping. You can ship stuff around Europe in a way that you can’t ship it across America, because, you know, sea.

48

chris 10.17.13 at 3:49 pm

Governments can, will, and should tax the bejeezus out of the fewer and fewer dinosaurs who continue to insist on driving their own vehicles.

There’s no need for such a tax if the autocar is cheaper than insuring a manual driver.

It may take a while for the technology to reach that level though.

49

mud man 10.17.13 at 3:51 pm

@Chris 22

No free-riding in a Democracy? Leaving aside practical issues with the society we see around us, as you say democracy is structured for the purpose that the majority can impose their will/costs on the whole, that is to say, the minority; cheap rides, if you like.

If you belong to something like a homeowner’s association or a monastic order, you don’t get to pick and choose at the level of regulation/”laws”; you have chosen the governing authority you submit to. In these homogenous days groups like that have had their teeth extracted, but I don’t know why all actual government should be monopolistic and geographically territorial.

I am sad nobody wants to talk about this. Some other day, maybe ….

50

Mao Cheng Ji 10.17.13 at 5:16 pm

‘Crowdsourcing’ doesn’t sound like a democracy; more like anarchy. Democracy is a political system for maintaining a bourgeois society.

51

DaveL 10.17.13 at 9:08 pm

Josh.G @18 We are way underdeveloped in this arena (there should probably be a high-speed rail network running north-south along the entire East Coast, and that’s just for starters).

Indeed there should, but there isn’t for several quite good reasons, the foremost of which is the cost of the land takings to build it. The Acela (and Amtrak in general) shares the tracks of privately owned railroads, and actually has lower priority on them than the companies’ trains. However, to build a completely new track that could support true high-speed rail between Boston and DC (ignoring DC to Miami) would cost hundreds of billions, maybe more. Unless we adopt Chinese methods of land-taking, it just isn’t going to happen.

On the other hand there is a thriving and relatively new market for bus rides among Boston/NYC/DC with many companies providing very inexpensive service at varying levels of comfort. Buses just aren’t as sexy as trains.

Igor @15.

No, no, no, no. You are plugged into the wrong conspiracy. The point of self-driving cars is that once we are all using them, the government can take control of them and prevent us from going where we want to. (They tried it with mass transit and it didn’t work, now they are going to try it with Google cars.)

52

John Quiggin 10.17.13 at 11:16 pm

I suspect there’s a generational divide here. Cars ceased to be fun about the time seat belt laws and radar came in (1970s in Australia). People my age have never really adjusted to this, but those under 30 have grown up with driving as a chore, and have much less interest in it.

53

Bus Nut 10.17.13 at 11:58 pm

DaveL,

I’d like to correct a misconception in your post about Amtrak’s status vis a vis the freight railroads.

Under the terms of the formation of the National Passenger Rail Corporation, the freight railroads who joined the compact were relieved of their requirement to run passenger service in return for giving Amtrak rights and priority over their track. Amtrak does not have to pay to operate service, unlike commuter rail services which must be separately negotiated. However, in practice railroads have some bargaining power to demand capital infusions to upgrade their track and signals when and if Amtrak attempts to add trains on an existing line or change their slots.

Amtrak trains can be under some circumstances severely delayed on the host railroads, but railroads face monetary damages (and Congressional scrutiny) when they deliberately sabotage Amtrak’s timekeeping. More typical are delays caused by trackwork or weather problems (flooding, mudslides, avalanches, heat orders). Infrequently, major delays are caused by mechanical problems or passenger problems (such as a medical emergency) that result in a train being stopped in a station and the train losing its slot. When it loses its slot it often ends up in between slower freight trains and the delays often cascade from there. If the train is delayed too much the crew may run out of hours and then a relief crew has to be brought in. In very rural areas with infrequent service replacing this may compound the delay.

54

js. 10.18.13 at 4:00 am

Alex @36:

I don’t know the Latour, but it sounds great; will check it out. Cheers. Also, this:

The Aramis engineers originally aimed to have a 2-person car, but for various reasons they ended up having bigger, shared ones, with the result that test passengers pointed out that you were still faced with those people!

made me realize that we could just as well be talking about self-driving auto-rickshaws instead. Which, somehow, sounds way more awesome!

55

js. 10.18.13 at 4:05 am

Also, too, whatever other advantages self-driving cars may have, I don’t get this bizarre idea that they would be faster. I have pretty no doubt that I can drive myself from point A to point B faster than any self-driving car could, at least as long as it remains recognizably a car. (Or what, they don’t stop at red lights? They’re self-driving ferraris?)

56

Matt 10.18.13 at 5:11 am

Also, too, whatever other advantages self-driving cars may have, I don’t get this bizarre idea that they would be faster. I have pretty no doubt that I can drive myself from point A to point B faster than any self-driving car could, at least as long as it remains recognizably a car. (Or what, they don’t stop at red lights? They’re self-driving ferraris?)

Machines have faster reaction times than humans, so they may be able to drive safely at higher speeds than currently permitted on highways. More importantly for heavily trafficked regions, traffic jams are often caused by slow reactions and over-reactions by human drivers in response to ordinary but unanticipated events.

“When you tap your brake, the traffic may come to a full stand-still several miles behind you. It really matters how hard you brake – a slight braking from a driver who has identified a problem early will allow the traffic flow to remain smooth. Heavier braking, usually caused by a driver reacting late to a problem, can affect traffic flow for many miles.”

57

Phil 10.18.13 at 7:04 am

Machines have faster reaction times than humans, so they may be able to drive safely at higher speeds than currently permitted on highways.

Although for that very reason introducing self-driving cars into ordinary roads would be a recipe for trouble.

I’d favour isolating self-driving cars on a highway of their own. Between one town and the next they could all drive at the optimum speed, which in turn would mean that most of the highway would only need to be single-lane – it’s not a queue if everyone’s going as fast as they possibly can. Also, with a dedicated highway we could save on guidance systems by laying a guide into the highway surface, perhaps in the form of two parallel ‘rails’. Obviously it needs a bit of refinement, but I think it could be a winner.

58

ajay 10.18.13 at 9:16 am

I’d favour isolating self-driving cars on a highway of their own. Between one town and the next they could all drive at the optimum speed, which in turn would mean that most of the highway would only need to be single-lane – it’s not a queue if everyone’s going as fast as they possibly can. Also, with a dedicated highway we could save on guidance systems by laying a guide into the highway surface, perhaps in the form of two parallel ‘rails’. Obviously it needs a bit of refinement, but I think it could be a winner.

This is something that’s been proposed quite widely as an interim stage: you drive your car from your house to the highway and then hand over to the autopilot, which slipstreams you on to the back of a passing convoy of autocars all going exactly at the speed limit, bumper to bumper, all talking to each other to synchronise braking. Faster and more fuel efficient than manual driving. When you get near your destination, it slows you, pulls you out of the convoy, and then hands you back manual control. You don’t even need an isolated lane – essentially the entire convoy is acting as a single very long vehicle. (In some models, the lead vehicle is manually driven.)
Also something the US army’s looked at for logistic convoys: one manned vehicle and the rest following its lead.

59

Alex 10.18.13 at 10:05 am

More importantly for heavily trafficked regions, traffic jams are often caused by slow reactions and over-reactions by human drivers in response to ordinary but unanticipated events

The notion that getting rid of this sort of complex feedback/control problem is simple is absolutely hilarious.

60

Chris williams 10.18.13 at 11:04 am

Yr Heathrow PRT relies on lead acid batteries charged through surface contact at stops. The track isn’t live. Not a lot of speculative tech in the motive power, then… Though its 1880s technology is still about 200 years in advance of the people mover at Birmingham airport, which is cable-hauled. They took the maglev out, you see: couldn’t get the parts. The moral of this story is that that future remains unevenly distributed.

As for trains, before thinking anything about trains, look at a paper with the words “what does this have to do with land values?’ written on it in big pen.

61

Pete 10.18.13 at 11:09 am

@Phil: that sounds like a guided busway. We have one of those, it combines the unreliability of a human driver with the inflexibility of a rail system.

http://www.noguidedbus.com/

62

Zamfir 10.18.13 at 11:23 am

Those convoys are not on the near horizon, but cars that drive themselves on the highway are basically market ready. S-class currently, but this might see wide adoption within the next decade.

Not sure if this would have much effect on public transport, in any direction.

63

NomadUK 10.18.13 at 11:27 am

Isn’t the future exciting? I can hardly wait to ditch my flying car so that I can get a self-driving one.

64

chris 10.18.13 at 12:19 pm

a passing convoy of autocars all going exactly at the speed limit, bumper to bumper, all talking to each other to synchronise braking.

This seems like a recipe for an instant 20-car pileup at the first deer or mechanical problem. The reflexes of a robot are a fine thing, but they don’t repeal laws of physics. To say nothing of some meatbag in the next lane losing control or attention, or what it’s going to be like trying to merge into that lane when you have to match speed while still in a regular lane.

When you get near your destination, it slows you, pulls you out of the convoy, and then hands you back manual control.

I assume you mean “pulls you out of the convoy and then slows you” — otherwise it has to slow the whole convoy, since there’s no space to have them travel at different speeds even briefly. Do you build a second autocar-only lane solely for merging?

Also, if you’re going the speed limit, that’s slower than manual driving on most US highways. Speeding by small amounts is normal because there’s only enough police to go after the most egregious speeders, and slight speeders know this.

65

David J. Littleboy 10.18.13 at 1:26 pm

“The biggest downside of individual motoriced transport is the high accident risk.”

They’re also expensive, stinky, slower than public transportation (if you live in a civilized city such as Boston or Tokyo), a horrific waste of time, and expensive. Did I mention expensive? I’ve bought a lot of cameras and guitars on money I didn’t spend on cars. (It’d be nice if I were making better use of the time I’ve saved, though.)

I don’t get cars. They seem real dumb to me. They’re not all that dangerous , though: medical malpractice kills more Americans than car accidents.

Out of curiosity, is there anyone else here who has made the decision to not own a car?

66

Josh G. 10.18.13 at 1:34 pm

DaveL @ 51: “However, to build a completely new track that could support true high-speed rail between Boston and DC (ignoring DC to Miami) would cost hundreds of billions, maybe more. Unless we adopt Chinese methods of land-taking, it just isn’t going to happen.

First of all, “hundreds of billions” isn’t an insurmountable obstacle for the federal government. We spent $670 billion on the military in FY2012, and $740 billion in FY2011. If our military was oriented around actual national defense rather than playing the Great Game of Empire, we could easily cut spending on it by at least 50%, and maybe as much as 75% or more. Even if a high-speed rail system cost a trillion dollars, a couple years of military spending cuts would pay for it. Is that politically impossible? Today it is, but Millennials tend to be much more skeptical of military adventurism than their parents and grandparents. Even the young conservatives tend to be more on the Ron Paul side of things, leaning toward an isolationist posture. In a couple of decades, when the Tea Party is six feet under and Millennials are entering their prime voting years, more spending on infrastructure and less on the military may well be a realistic political option.

Secondly, there is quite a bit of room for middle positions between “Chinese methods of land-taking” and the current convoluted U.S. system. In fact, most European countries fall in that middle ground. The U.S. Constitution requires just compensation for property taken via eminent domain. It does not require that the people whose land is taken be allowed to actually throw sand in the gears of the project. It would be totally constitutional for the government to pay compensation to the owners at fair market value and say “You have 6 months to get out.” Draw the line, pay compensation (heck, pay 25% above fair market value to reduce complaints), and bulldoze. If the evicted landowners want to appeal the amount of compensation, they can do that, but they shouldn’t be able to appeal the project itself.

Environmental impact reports are not required by the Constitution either; they are a failed legislative experiment that has made public infrastructure improvements all but impossibly expensive and lengthy. And they fail to see the forest for the trees; a high-speed rail line would provide so many environmental benefits in other areas that worrying about whether one obscure species of fish or insect might be hurt is ridiculous.

67

Chatham 10.18.13 at 2:21 pm

The cost estimates I’ve seen for North East corridor high-speed rail puts it at about $150 billion. Let’s double that and assume a lot of cost overruns, and put it at $300 billion. The cost making most of the Bush tax cuts permanent over the next decade (rather than letting them expire) is around $3.7 trillion. The cost issue is merely one about priorities, and yeah, the priorities in the US at the moment seem to be for cutting taxes and social security while letting infrastructure fall apart.

Still, I’m not sure that high speed rail would be the best use of that money. It seems like cost is much more of a barrier in the NE than speed (at least, it’s the complaint I hear much more often). DC – NYC is under three hours on the Acela Express, and Boston – DC is under seven. We might get much more bang for our buck developing local public transportation systems and commuter rail lines.

68

Trader Joe 10.18.13 at 2:51 pm

“DC – NYC is under three hours on the Acela Express, and Boston – DC is under seven. We might get much more bang for our buck developing local public transportation systems and commuter rail lines.”

That’s it exactly. The full fare business traveler doesn’t ride in a train for 7 hours when they can fly in 3 including the time spent on each end w/ security and to the airport. There are few travel routes that can justify the cost of implementation without full fare passengers willing to pay the actual cost of service. BOS-NY, DC_PHLY_NY and maybe LA_SF, most other pairs don’t have the combination of volume and willingness to pay premium money to make the investment work. Local public transport would have far more bang for the buck both environmentally and for benefitting the greatest number of people for the dollars spent.

69

ajay 10.18.13 at 3:39 pm

is there anyone else here who has made the decision to not own a car?

Yes.

This seems like a recipe for an instant 20-car pileup at the first deer or mechanical problem.

Mechanical problem yes, but the idea is that all vehicles brake simultaneously with the first one.

70

Wonks Anonymous 10.18.13 at 3:48 pm

@65 David J. Littleboy:
I’m originally from the suburbs and used to commute by car to another suburb. My car broke down later, and by that time I had started taking the train downtown and I figured I should just move. Now I get by on a bicycle and tend to look askance at job offers that would require me to buy a car, seeing as how don’t even have a parking spot at my current residence. I suppose that’s choosing in a more passive path-dependent way.

I know there’s been a big decline in vehicular fatalities (although there’s still plenty of room for improvement). Is that due to a decline in the number of accidents, the lethality of accidents, or both?

71

chris 10.18.13 at 5:24 pm

Mechanical problem yes, but the idea is that all vehicles brake simultaneously with the first one.

Brake yes, but if the first vehicle HITS the deer, then it may lose momentum more rapidly than the others can match by braking and/or more rapidly than it “expects” and can warn the others it’s going to. (And that’s assuming every vehicle has the exact same mass, acceleration and braking capacity, or only brakes as hard as the least capable vehicle in the convoy. Also, add lack of traction to the list of potentially pileup-causing events.)

Reducing the following distance is dangerous even with coordination between the autocars. Because danger is precisely about what happens when things don’t go smooth.

Oh, and if the manual lanes are actually substantially below the speed limit because of heavy traffic, construction, etc., then the autocar lane has to slow down too, otherwise it will be impossible to merge in and out, even for robots. And if it has no following distance to absorb the change in speed, then it has to slow down all the way back to the end of the line.

This means that there is not really a big potential improvement in road capacity without giving up much of the safety gain from a tireless and undistractable driver with lightspeed reflexes.

72

guthrie 10.18.13 at 6:10 pm

Regarding the potential deer caused mass pile up, here in the UK that probably isn’t so much of a problem; if it was I would expect to hear about it far more often on the news. WE’ve just shrugged off a multiple fatality accident on a bridge caused by careless drivers in fog; why would the deer- autocar interface end up any different?
Of course it’ll vary a bit in the USA, but I can’t help feeling that a lot of people bring up random out of control occaisions and expect automatically driven cars to be much better than humans, who are provably pretty rubbish as microsecond decision making in high speed car accidents, and therefore autonomous cars are rubbish and dangerous.

As far as I can tell, humans take tenths of seconds to notice things and then react, with it taking a second or more to hit the brakes. If computer driven cars can’t react ten times faster than that I’ll be really really surprised. Moreover the scenario on a motorway can be quite different – anything which isn’t a car and suddenly appears on the radar etc can be treated as a possible obstacle, which surely increases reaction time. Anyone know how fast computers can calculate that sort of thing? I seriously doubt they’ll be as slow as humans, although obviously there’ll need to be programming and testing first. Same goes for the car convoys – computers can communicate quite fast you know, so fast that you as passenger wouldn’t be able to tell the difference between the first one braking and the ones further back in the queue. I suspect the physical application of the brakes would be slower than the time it takes to pass the message back, once the decision has been made.

And there could be some advantages as well. Last year I hit a fox on the motorway. It was dark, and I had my headlights dipped because of cars on the other side of the road. I didn’t see it until it was at the outer edge of my headlights, which gave me well under a second in which to react, time enough to start braking and go from 70 to 50mph. But I didn’t really have time to think about swerving across the road, as well as being unwilling to see how well I could control a swerving car at 70mph.
A modern computer controlled car would, I hope, and I’m pretty sure it is possible, notice the fox as soon as it arrived on the road, whether by IR or radar or such, raising an instant “foreign body” flag, much quicker than I, with my human limitations, could see it.
The final result of the suicide attack on my car was a broken radiator, the bastard. But hopefully it hadn’t bred.

73

Alex 10.18.13 at 6:52 pm

The problem that sticks in my mind is just the old thinking distance/braking distance chart from the Highway Code. If you want to go faster, the braking distances go up very fast indeed. This is because braking is the conversion of kinetic energy to heat, and kinetic energy is half the mass times the velocity squared. So whatever capacity gain you get from not keeping-your-distance has to be spent if you want more speed.

74

js. 10.18.13 at 7:01 pm

I thought Phil @57 was really just describing a train. Which: yay!

75

Mao Cheng Ji 10.18.13 at 7:07 pm

@72, see, the smart car would probably have to detect and react to the fox well before it steps on the road. It would have to anticipate what the might do, and act accordingly, hit the brakes and/or change the direction. Even if 9 times out of 10 the fox would decide, in the end, against crossing the road.

Anyway, I remember my Indian colleague telling me that if you want to drive a fancy modern car in a large Indian city, the first thing you do is to disable all the safety functions, because all the blinking and beeping and flashing and automatic breaking makes driving impossible. See, computers are better at calculating, but human beings are better at anticipating the most likely moves of other actors on the road: human drivers, pedestrians, even dogs and cows perhaps. Dunno about the foxes.

76

chris 10.18.13 at 8:34 pm

Of course it’ll vary a bit in the USA, but I can’t help feeling that a lot of people bring up random out of control occaisions and expect automatically driven cars to be much better than humans, who are provably pretty rubbish as microsecond decision making in high speed car accidents, and therefore autonomous cars are rubbish and dangerous.

I was responding specifically to the idea (originally in comment 58) that the autocars could have much less following distance from each other while remaining safe. I agree that a properly programmed autocar maintaining a following distance that would be safe for a human at the same speed would usually be at least as safe, but if they maintain the same following distance then you can’t fit more of them on the road or have them drive faster. So they don’t solve any traffic problems relative to human-driven cars, and expecting them to replace already-existing mass transit systems would significantly *worsen* traffic compared to the status quo.

77

Matt 10.18.13 at 8:55 pm

Anyway, I remember my Indian colleague telling me that if you want to drive a fancy modern car in a large Indian city, the first thing you do is to disable all the safety functions, because all the blinking and beeping and flashing and automatic breaking makes driving impossible. See, computers are better at calculating, but human beings are better at anticipating the most likely moves of other actors on the road: human drivers, pedestrians, even dogs and cows perhaps. Dunno about the foxes.

This is currently true, and I don’t want to count technological eggs before they have hatched, but… it seems to me that historically when machines are set against humans at a task, there’s been a long experimentation period where they are just expensive jokes or curiosities, then a shorter period when they are minimally competent by comparison, then a never-ending period where they are clearly superior. It looks like self-driving vehicles are at least in the minimally competent phase now; they were jokes as recently as 2004 (the first DARPA Grand Challenge: the best entry got stuck on a rock after navigating 12 kilometers of a 240 km course).

78

Collin Street 10.18.13 at 9:07 pm

Just as a point, but paying “market value” in compulsory acquisition isn’t actually just compensation in the general case. Market value is the value to the marginal vendor, and the value to individual owners will vary, and not only through profiteering: if I only have one kidney I’m going to want rather more than “market prices” for it, no?

79

Mao Cheng Ji 10.18.13 at 10:21 pm

@77, two things:
First: I suspect, for machines to be superior you’ll have to remove all the human psychology from the process somehow. When you drive in some places (Italy, for example), you look into other drivers eyes to anticipate their reaction to your maneuver. And pedestrians do it too: you avoid turning your head to look at approaching cars, so that the drivers aren’t sure that you see them, and then they stop. Machines don’t understand this sort of thing. But of course it is possible to get rid of psychology; just a matter of time, like you said.

And second: I suspect most of the human car drivers operate mostly subconsciously these days, which means: very fast. There is an extremely fast computer inside your head, it’s your consciousness that slows it down.

80

Phil 10.18.13 at 10:39 pm

I thought Phil @57 was really just describing a train.

Bazinga!

81

Phil 10.18.13 at 10:57 pm

To unpack that thought process slightly: mixing robo-cars and human-driven cars is a recipe for all kinds of wackiness on the road itself, because the onboard computers wouldn’t think the way humans do and consequently wouldn’t drive the way people do. Unless you think recognising all the different kinds of pattern human drivers process routinely, balancing out all their (shifting) priorities and acting accordingly (making sure to slug down the resulting manoeuvres to something like human reaction speeds and muscle latencies so as not to freak out nearby drivers) is just a matter of writing some code. I may be proved wrong, but I suspect that the first generation of robo-cars that plays well with others will still be recognisably a robo-car, and I’m not too sure about the second, third or nth. There is also the matter of legal liability when something goes wrong, which could turn out to be rather intractable if the thing going wrong involved injury to a person. (Makes you wonder why they don’t avoid all these problems by hard-coding the Three Laws of Robotics.)

So let’s say we need to segregate robo-cars on their own road. Advantage: there’s no weaving in and out of (human) traffic required, and no need to react to other drivers drifting all over the road. In fact, there’s no need for more than a single lane, which can drive at the speed limit or (given the absence of any people to be put at risk) even faster. Put some guide rails down and none of the vehicles needs to steer at all, which cuts down on processing. The only problem is what happens when the lead car slows down suddenly: brakes need to be applied all down the line to ensure even deceleration, as there’s only a narrow space between each car and the car in front; also, there’s nothing in between any two cars but air. Really, it’d make more sense if you coupled them all together… and presto, we’ve reinvented the train.

82

Matt 10.18.13 at 11:37 pm

Unthinking human reflexes, e.g. the patellar reflex induced by the doctor hitting your knee area with that little hammer, have a latency on the order of 30 milliseconds. That’s pretty leisurely by the standards of machine reaction time. Reaction times that involve even the simplest decision, like “press a button when the light flashes,” take more than 100 ms even for attentive humans at the 90th percentile of ability.

I don’t think it’s necessary to eliminate psychology or give machines a robust theory of mind to produce capable machine drivers. “Orientation of pedestrian head relative to vehicle” just becomes one more input to a control system. Machines are already surprisingly good at statistical pattern recognition and classification in large unlabeled data sets. A machine driver can incorporate more “road experience” than any one human, since its control systems can be trained from thousands of years’ worth of data aggregated from many vehicles and locations. It can also incorporate many more kinds of senses and inputs than human drivers have available: continuous 360 degree camera observations, radar, infrared imaging, laser scanning, magnetic fields, radio communications… Self-driving machines don’t have to imitate humans any more than airplanes have to imitate birds.

A past example of this sort of advance: people have been trying to teach machines to recognize faces for decades, and humans have always been far better. Until 2006. That year the NIST Face Recognition Vendor Challenge identified several systems that could equal human accuracy in matching preexisting records of faces with faces imaged passing through a checkpoint. The biggest improvement over the previous challenge was using 3D images of faces rather than relying on 2D pictures. (The humans could look at the 3D data also, but they didn’t get as big of a boost.) If you can’t match human performance with human-equivalent senses, build better senses.

83

Matt 10.18.13 at 11:53 pm

To unpack that thought process slightly: mixing robo-cars and human-driven cars is a recipe for all kinds of wackiness on the road itself, because the onboard computers wouldn’t think the way humans do and consequently wouldn’t drive the way people do. Unless you think recognising all the different kinds of pattern human drivers process routinely, balancing out all their (shifting) priorities and acting accordingly (making sure to slug down the resulting manoeuvres to something like human reaction speeds and muscle latencies so as not to freak out nearby drivers) is just a matter of writing some code. I may be proved wrong, but I suspect that the first generation of robo-cars that plays well with others will still be recognisably a robo-car, and I’m not too sure about the second, third or nth.

Have you watched videos of the Google cars driving in urban California? Apart from the odd sensor packages on top of the car, they don’t stand out jarringly from human-driven cars. Google claims that they have logged 500,000 miles in self-driving cars on California roads. You should take self-promotional information like this with a grain of salt, but the vehicles are regularly observed operating in public.

84

JanieM 10.19.13 at 1:30 am

Really, it’d make more sense if you coupled them all together… and presto, we’ve reinvented the train.

Except that one of the things people love about cars is that they can be driven here, there, and everywhere, instead of just from one station to another. A lot of the attraction of a private car disappears if you can’t do that.

85

js. 10.19.13 at 2:30 am

the [robo-]vehicles are regularly observed operating in public.

And are they noticeably faster than cars with actual drivers?

86

Zamfir 10.19.13 at 5:01 am

I don’t get the insistence on speed. Why should robot cars need to be faster than other cars?

87

js. 10.19.13 at 5:52 am

It’s the defenders of robo-cars that keep insisting that robo-cars will be stronger, better, and faster. I’m merely expressing mild scepticism (and an implicit preference for public transport).

I can perfectly well imagine stronger, better, and slightly slower robo-cars, and I’ve no real problem with people who have difficulty walking preferring them. I’d still prefer a steering wheel and some manually changeable gears if you’re going to have me travel in a car.

88

Zamfir 10.19.13 at 6:24 am

I guess I just don’t see the conflict between (hypothetical) robocars, and public transport. We don’t own a car currently, we get by with public transport and cycling, which I am pretty haply with over all. But it has its limits, I can only keep this up because I can borrow or rent a car when convenient.

As it is, we’ll probably have to buy a car in a few years time, when we have kids. Once we have a car, one of us will probably use it for commuting, because a car plus two train commutes is expensive.

If there were trully auronomous robocars (which I do not see coming that soon, whatever the hype), then a public transport oriented life would be easier for us to maintain. Through robo taxis, because busses could be smaller and serve a denser, higer-frequency network.

89

js. 10.19.13 at 6:55 am

If there were trully auronomous robocars (which I do not see coming that soon, whatever the hype), then a public transport oriented life would be easier for us to maintain. Through robo taxis, because busses could be smaller and serve a denser, higer-frequency network.

A. I don’t know what “auronomous” means. B. If we lived in this magical world where tons of public resources could be thrown at things other than fancy toys that kill people, I’m all for building self-driving buses and cars and investing tons in other kinds of public transport too (trains!). Except that we live in a world where there are extraordinarily well-established entrenched and vested interests against the better development of public transportation and very much for individualized transport-boxes. (I live in the US.) In this world (or country), I’ll continue opposing self-driving tin-boxes and supporting public transport.

90

Zamfir 10.19.13 at 8:41 am

Autonomous. God knows what the spell checker had in mind :-) As in, it could reliably drive around without a driver. Whether that’s a box for 1, 2, 10 or 30 passengers strikes me as a secondary matter. I don’t mind sitting in a train or bus with many people, but it’s not much of an advantage either.

I don’t get the point about magic. As far as i can tell, the main obstscle to more investment in public transport, is that most people have a car anyway. And once they have those fixed costs, other modes of transport become less attractive. They have cars because they really are very convenient, even if you live (like I do) in a place which is pretty accessible for public transport and bikes.

So to me, the road to more (investment in) public transport requires that more people do not own cars. Which would be greatly helped, if there was a better option for those trips where cars have the greatest benefit over public transport.

I have lived in places where taxi drivers are cheap, compared to my own income. That makes car ownership a lot less enticing. It would be great if we could have that for everyone, instead of the small class of people for whom a taxi driver’s income is negligible.

91

Matt 10.19.13 at 8:51 pm

It’s the defenders of robo-cars that keep insisting that robo-cars will be stronger, better, and faster. I’m merely expressing mild scepticism (and an implicit preference for public transport).

To be clear, I think that the “and faster” aspects can only happen once robo-cars are much more common, at least (say) 10% of the passenger vehicle fleet. It is very doubtful that the first-generation cars are going to get special speed limit raises on the freeway — they will have to demonstrate superior safety for years first. And it is also doubtful that a few robo-cars sprinkled throughout a mile-long traffic jam can control the situation enough to restore free flow. The fast reaction speeds I mention above can eventually contribute to faster traffic flow, but I was also bringing them up to counter the notion that robo-cars would need separate roads or lanes due to inability to deal with human behavior in other vehicles or pedestrians.

I didn’t have a driver’s license until I was 22, didn’t have a car until I was 27. I still don’t drive much: I need to refuel only every 6 weeks. Driving is a tedious chore and I would prefer to not drive at all, but even my slight driving would be considerably more expensive if I replaced it with taxi service. I hope that robo-cars arrive not so that I can own a better car but so that I can own no car at all, and temporarily summon one that’s cheaper than a taxi when I need to go further or carry heavier cargo than my bike allows.

92

JanieM 10.19.13 at 9:12 pm

I hope that robo-cars arrive not so that I can own a better car but so that I can own no car at all, and temporarily summon one that’s cheaper than a taxi when I need to go further or carry heavier cargo than my bike allows.

Except for the “summoning” part, you don’t need robo-cars for that.

93

Matt 10.19.13 at 10:16 pm

Yes, if Zipcar expanded to where I am that could work, though it still doesn’t reduce the inconvenience of the act of driving. Zipcar currently has locations in only 15 states. I have never lived somewhere so remote that I could not get a taxi dispatched to me. If Zipcar expanded to my region and started stocking self-driving cars, that would be ample compensation for the 21st century’s failure-so-far to deliver on lunar hotels and arcologies. FWIW I like riding trains too — the most difficult part of a train trip for me is getting to/from the train station from/to my home and final destination.

94

Tom Slee 10.19.13 at 10:51 pm

Matt – If Zipcar expanded to my region and started stocking self-driving cars

More likely Uber than Zipcar, given Google’s $58M investment.

95

Zamfir 10.21.13 at 5:56 am

I use a Zipcar-like service, and one of their cars is parked in front of our building. It helps, but it has limits. In particular, it’s expensive.

For good coverage and availability, the city has to be littered with cars. As result these cars are unused most of the time (like private cars), and that shows in the price. A related effect is that you pay quite heavily for the hours you use them, apart from kilometers.

Taking the car to the station would be folly, as you end up paying for the time parked at the station. Taking the car for a weekend away is a serious expense, even if it’s only for the trip to and back.

Comments on this entry are closed.