Over the weekend, in No. 13 from the Ulysses-scale saga* of California's plan to build a north-south High-Speed Rail (HSR) system, a reader from the Silicon Valley tech industry said that his state should just forget about railroads—normal, high-speed, maglev, or whatever. Instead it should embrace the future represented by self-driving cars. (For previous episodes see No. 1, No. 2, No. 3, No. 4,No. 5, No. 6, No. 7, No. 8, No. 9, No. 10, No. 11, No. 12, and No. 13. Today's is No. 14.)

Almost no one who's written in since then agrees. Here is a sampling of the case made, in dozens of messages, against self-driving cars as a realistic transport hope. Let's start with reader D.G. in the Maryland suburbs near Washington D.C.:

I have been reading your HSR series with great interest, even though I live on the wrong coast, since, as you point out, the decision California makes here will strongly influence the direction the rest of the country takes. I have no expertise in this area, only a strong interest in planning issues, as they relate to quality of life, climate change, safety, and other public policy concerns.  

I am moved to respond to the message from your info-tech reader regarding the relative virtues of self-driving car technology and HSR.

First, I observe that the reader's view embodies two infuriating and too-common tendencies when considering any public policy issue.

The first is view the issue so as to make the perfect the enemy of the good, and the good the enemy of the good-enough.

The second is to fetishize the highest, newest, "coolest" technology over "old" technology (defined roughly as anything that has already existed in the market for any length of time), and in so doing to accept uncritically speculations about the efficiencies and other virtues of the highest high tech and to dismiss anything else (e.g., rail) as somehow inherently obsolete, inferior, undesirable, and just generally icky.

The problem is that both tendencies inevitably reflect, and reinforce, lazy thinking.

The reader's thesis is that HSR "will be obsolete for most of its lifetime." It's worth noting that nothing the reader goes on to say actually proves, or even serves persuasively to support, this conclusion, or even defines what "obsolete" means here, or shows why this is bad.

Rather, s/he [JF note: it's a he, and I've changed future references accordingly] provides a series of tantalizing speculations about self-driving tech; if all of them are true—a conclusion that can be reached only by engaging in multiple levels of speculation—it still says nothing about the ability of HSR to do two things substantially better than they are done today: 1) move people efficiently, safely and at reasonable cost (yes, with major capital investment, but this is true of self-driving tech as well), and 2) substantially reduce the carbon emissions involved in doing so.

That HSR will do both things has been established, in my view, beyond reasonable dispute. Whether an as-yet nascent technology (self-driving cars) will at some later time—and with the development of tremendous tech infrastructure not yet planned, much less executed—be able to do both things and do them better, is not.

Some particular observations about the reader's mode of argument [JF note: statements in italics are from the previous pro-self-driving-car argument]:

-Self-driving cars cover ALL highways, not just one station-to-station route.

This is presumably true, but HSR will nevertheless still be substantially cleaner and more efficacious than the status quo, an important point the reader ignores.

-Self-driving cars will be safer and more efficient than current driving because they coordinate with each other.

Again, presumably true, but no safer and more efficient (perhaps less so, in the aggregate?) than HSR.

-Self-driving cars can be faster on highways because they can caravan. For the same reason, they can be more energy-efficient.  Because they are point-to-point instead of station-to-station, they get you from source to destination faster and with less hassle.

This appears to be a comparison between SDC and current driving, not HSR and current driving, so even if true it proves nothing with respect to HSR.

   -Self-driving cars will create productive time because the driver can attend to other things.

Just like HSR. I ride the Acela mainly for this reason.

  -Self-driving cars avoid a single point of failure (track disruption) because the road system and vehicles are distributed.

So SDC technology will not be subject to disruptions? There will be nothing that would, for instance, block SDC traffic on I-5 for hours at a time? Why on earth would one simply assume this? Has this person never had his network go down? Again, this is uncritical acceptance of "new" tech qua new as better than "old" tech. Fetishization -- not science.

- For these reasons, by 2030 or 2040 when HSR is done, the best physical investment in getting between San Francisco and Los Angeles will be to double the width of Interstate 5—much cheaper than a whole new train system.

I don't see anything in the reader's argument, or elsewhere, supporting this conclusion. How do we know it would be cheaper? And can we afford to wait and hope? Here the reader would make the good the enemy of the good-enough. We can be reasonably confident that HSR will make the improvements we seek. We cannot be reasonably confident that other technologies (e.g., SDC) will do so. At the current inflection point, then, why wouldn't we proceed with HSR?

- Rail is terrific for some purposes but it represents old technology.

Here we get to the crux of the reader's non-argument: newness for newness' sake.

My response: So what?

We want the right technology; it doesn't matter whether it's the newest. Carrying the reader's logic to its logical extreme, rail was "old" fifty or even a hundred years ago and ought to have been abandoned then.

In sum, the reader is just making another—particular unpersuasive—argument that we should say no to HSR, do nothing, and hope.

No sale.

Another from the 1950s, the Curtiss-Wright Gem 2500 Air Car built for the military (Army Transportation Museum)

Next, from a reader in San Diego:

As a 5th generation native Californian (a post-war baby boomer) I've seen the best of the Golden State. I've appreciated the superhighways and freshwater capture and delivery systems that have made the state "work", and I'm keenly interested in the development of HSR because our state desperately needs the benefits the proposed passenger rail system will bring to it. Currently we're at, or past, the point of sustainability in our transportation and freshwater systems.

The point of your 13th article was to highlight other high-tech solutions to the HSR solution. However, I see a few major objections to these.

As to "self-driving cars":

(1) Yes, theoretically they are technologically feasible, however it was also theoretically possible to develop an effective ICBM missile defense system (but it hasn't worked out so well, has it?). We need to be cautious in thinking that what our minds can conceive is as easily doable. The current self-driving system is only beginning to prove itself, and isn't up to the task of safely moving cars at high speeds and heavy traffic volumes. It would probably also need to have all vehicles on a traffic communication network for optimal operation, rather than simply radar avoidance and roadway tracking. A control system failure in one vehicle could cause havoc in a heavy traffic stream, so high reliability and redundancy are essential.

An effective car control system could pack the cars in tighter, reducing the space needed per vehicle (and thus reduce the demand for new freeway lanes), but this also reduces the safety margin per vehicle. This is not a comforting compromise, and only works when all the vehicles on the road are on the same control system (and with a backup). This also means that the traffic stream operational parameters are limited by the acceleration and braking performance of the worst vehicle on that stretch of roadway. These are, in short, some of the operational parameters to be used in the Caltrain C-BOSS traffic control system on that commuter corridor, but used by trains instead. This is still new, advanced technology, and used only to control train spacing rather than vehicle intra-lane tracking, inter-lane movements, and route computing. In short, there are some significant technical issues which need resolution before automatic car control on a heavily-used high speed highway could be deemed practicable.

The car control units would also need to be mass produced in order to make them not too expensive, but the total cost of all of them is still likely to be very sizable. In essense, they would be another tax imposed on drivers, hidden from the total system cost. Would all cars be required to have them? How safe is the system when many vehicles on the freeway don't?

(2) It also is very expensive to add more lanes to the freeways in case the car-control system can't handle all the traffic on the existing lanes. In many cases new lanes need land seizures as well as re-doing exchanges and overpasses, which greatly increases the expense and time required.

(3) The timeframe for implementation of such a technologically advanced and widely distributed system can't be easily estimated, and should not be counted on soon.

(4) One of the major reasons for the use of HSR and commuter systems together is to reduce pollution. Simply still relying on cars does NOTHING to address that long term issue.

SUMMARY: The costs of doing #1 + #2 would be expensive however you look at it, and cannot be assumed to be cheaper than the better understood cost of the passenger rail system. The technological unknowns are great, and it would be foolish to base a whole transportation system on something that is not proven even in a test framework yet. Lastly, this solution only makes pollution worse. In short, to add a more powerful control system to our current vehicles may be desirable to do, but it is not a substitute for the passenger rail system upgrade.

The reader then added an analysis of maglev trains as an alternative, which I am skipping for the moment. He sums up this way:

Finally, my conclusion: In the 50s and 60s we built our nuclear deterrence on the basic principle of using a "tripod" of different technologies for more reliability.

In an analogous way, our transportation system is built on using different technologies: roadways, railways, airways, and waterways ... each with their own range of abilities and potentials. In other places in the world, railways using HSR have been reborn, providing a cost-effective and useful service. The US has not made significant technological investments in rail, and so our railways have fallen behind. However, simply letting passenger rail wither away in the US would in the end prove a very short-sighted decision. In the end, I believe the California HSR system will be shown to be a workable compromise that does what it set out to do. That is,

    • link the major economic and population centers of the state together for business and social betterment,

    • provide a high speed, economical, statewide rail backbone that ties into local commuter systems,

    • reduce suburban sprawl, and promote a rewarding lifestyle for Californians without need for an auto,

    • reduce long-distance auto & airplane travel within the state, and the associated traffic congestion and air pollution they generate, and

    • utilize as many existing resources as possible to reduce costs and make the system affordable.

The Gem 2500 in action, riding on a cushion of air (Army Military Museum)

Next, this from reader Andrew Stokols, who is a generation-plus younger than the previous Boomer-era reader (or than me). He writes:

As someone who is: A. a California native, B. lived in China for several years, and C. has traveled on HSR in Europe, Japan, and China and D. A journalist who has written on urbanization in China and Asia, I should admit up front that I am very supportive of CA's HSR project. Remembering how many times I waited for delayed flights to take me from Socal where I grew up, to the Bay Area where I went to college, I, like many other Californians would have benefited greatly from a high speed rail system long ago, and still would.

While I am certainly open to hearing arguments of alternative "more advanced" technologies like maglev or the hyperloop that some people say should leapfrog a high speed rail system, I have to call out these kind of arguments for what I think they often are: a thick-headed rather unexplainable aversion to rail travel that has more to do with libertarian political bias against big government projects than it does with the merits of the technology, even if the people making these arguments present themselves as scientific, technocratic minds just interested in the best solution.

First lets look at the alternatives:

1. The Hyperloop: it sounds great, but its farfetched, the technology doesn't exist yet, and even if it does get developed, we're looking at several decades before something like this is able to be even tested, let alone put into commercial application. Saying we should abandon a technology like HSR which is proven around the world, reliable, that we can start building now, for something that of now exists only as a sketch and whimsical diagrams is ludicrous.

2. Self-driving cars/expanding highway surface: As one of your readers, the student from Caltech/MIT makes the claim, CA should invest in expanding I-5 instead of a HSR system. Apparently he is unaware of something in transportation planning called "induced demand", whereby expanding roads and lanes leads to more traffic, not less.

Anyone who has driven on stretches of 10-lane wide Socal highways like I-405 in Irvine (where I'm from) would know what I'm talking about. The whole point of HSR is to take demand off highways, reduce pollution, and get people out of cars. Sure, self-driving car technology is being developed by companies like Google (who certainly have an economic interest in supporting alternatives to rail travel). But that's far from being implemented and proven, and I-5 is not going to be a self-driving highway for some time, probably closer to decades than years. In a state increasingly dominated by tech industries, it makes sense that some of these people (Elon Musk included) think they can solve any problem better than gov't, but frankly look at SF. [He goes on to describe the contrast between private shuttle systems for tech-industry employees and a decaying public transport system for everyone else.]

3. Maglev: I've ridden on Shanghai's maglev, and I know Japan is developing a maglev alternative to the Shinkansen route. This alternative seems to be the most likely in the near term, to conventional HSR. But from what I know the costs are still higher, for the initial investment and development. It is not widely used around the world, and like you pointed out, the Shanghai track is a small demonstration line that runs for seven minutes. Simply put, Maglevs are not proven, and for a country where infrastructure is already hard to invest in, putting in a system that has not been proven is going to extend the cost, time until completion, and possibly safety risk for the system in the short term.

Finally: We can continue with a project and a technology that is reliable, demonstrated around the world in its safety and efficiency. Or we can wait on technologies that are still in development, postponing when California residents are able to make the journey from SF to LA in 2.5 hours. Already, we're looking at the late 2020's or longer for completion of HSR. I think its pretty astounding that even scientists and experts knowledgeable about the pros and cons of technologies maintain an ingrained aversion to HSR, when it's a reality in most other developed, and even developing countries around the world. It's really a hollow argument, though: "Would you prefer a system where you can be instantly teleported from SF to LA?"  "Of course." But, that doesn't mean its going to happen soon.

Thanks for reading my defense of HSR. But I think its important to point out that when people say we should leapfrog HSR for newer, sexier technologies, they aren't being realistic, practical, or "fiscally conservative" for that matter.

Model of GM's 1950s Firebird I, which surprisingly enough never made it into mass production (Wikimedia Commons)

And, as a bonus, here is one more, on some of the other complications in the self-driving car approach to California's challenges.

I see you've forwarded a note from a self-driving car enthusiast, which is of a piece with many arguments against building a flagship HSR system. These people are bad at separating their enculturation (inelegant use of the word, sorry) of car travel from actual rational planning. I'd like to toss out some calculations to stir the pot here:

What's the footprint of an average sedan, 25 square feet? That's five people max, but as we all know right around 90 percent of this country's commuters travel alone by car. Shinkansen's—which I'll lazily use as a real-world stand-in for CA HSR—wiki page cites a 384-seat capacity for six cars. Interstate lanes are 12 feet across, just bigger than the 11-foot width of the Shinkansen, so we can sloppily compare a single rail line with a single highway lane.

A mile of highway can fit 10 Shinkansen trains at six cars (well, almost 10.5 but we'll round down), or it can fit around 585 9-foot sedans (again, rounding down a bit from 587 and change). Ten Shinkansen trains total 3840 people, and 585 mid-size cars fit 2925 people, if each car has five butts in seats. If, however, self-driving cars do not materially affect the distribution of people choosing to drive alone (and if they're competing with people choosing to buy cars in order to drive alone, why would they?), that means that 89 percent of those cars only have one person, and the remaining 11 percent have an average of, oh, let's be generous and say three passengers per car. That's only 717 people per lane per highway mile, assuming the cars are driving end-to-end (they won't be).

And what about speed? The Autobahn recommends a top speed of about 80 mph on unrestricted highways, and a couple years ago Texas opened an 85 mph highway (those rebels). Self-driving cars will likely be able to handle higher speeds, but how much can the roadway conditions (and the gas mileage) tolerate? I ask because I don't know, but let's assume our driverless supercars can cruise along at 100 mph without much hardship.

Assuming that CA HSR trains will mirror the Shinkansen's average speed, 160 miles in an hour 20 minutes, (it's my understanding the plan is they'll actually be faster, so that's 75 percent of max operating speed for those who want to do further math), they'll make a depot-to-depot trip at around 120 mph—faster than the car, if the car is traveling depot-to-depot at the 100 mph max the whole time.

Napkin math says that, at the state level, investing in cars as people-movers is a terrible idea. I have no idea how to gauge the share of drivers who aren't going to a major transit destination, nor do I have much luck deciphering interstate origin/destination studies to figure that out, but my suspicion is that if you're not going somewhere the train would be, you're in the minority, which for one thing means an investment in you is relatively wasteful. And that doesn't even go into externalities like total land use or fuel expenditure!

I'm not saying self-driving cars are a bad idea, on the contrary they'll make it much easier and safer to travel to that suburb or rural town from which Mom refuses to move (or at least move the family reunion). But in an efficient transit model, cars are a specialty tool, and you just don't buy a spiralizer before a santoku knife, especially if that spiralizer is going to pump carbon into the air much longer than the knife will.

Coming soon: the end of the saga, and What It All Means, at least from my point of view. Then, Californians, get out and vote on November 4, since the rail plan is a significant issue in the gubernatorial race.