Hyperloop: lots of hype for something that doesn’t yet exist

Hyperloop

The last few days have seen lots of pixels spent on Elon Musk’s Hyperloop concept – and I couldn’t resist chiming in. It’s a fascinating idea, but far from an actionable one.  Musk seems to have put a lot of thought into dealing with some of the technical hurdles of previous vac-train ideas, but rather than put these forward in the marketplace of ideas, he has taken to bashing California’s high-speed rail project instead.

Musk’s supporters take his endorsement seriously, with many openly hoping that the Hyperloop will kill off high speed rail, even though high speed rail is a proven technology operating around the world, while the Hyperloop exists nowhere but as a sketch on the back of a cocktail napkin. It’s a testament to the power of an idea, but it also shows how easily we can fall for bad ones.

A few base criticisms of the idea come to mind: the Hyperloop is not necessarily a superior technology for the problem it seeks to solve (travel between SF and LA); technology does not change the basic geometry of a transport system (and it must respect the basic tolerances of the human body); a fancy new technology is not necessary for innovation and improvement; and every strain of common sense indicates that the cost projections for this thing are pulled out of thin air (where else would they be pulled from?)

Solving SF to LA: Musk’s proposal does not actually serve either Los Angeles or San Francisco. The ‘last mile’ problem in any urban transportation system can be really challenging and really expensive. Musk simply avoids the problem by terminating in Sylmar and the East Bay. The hyperloop’s faster speed is irrelevant to the real question: travel time. Maximum speed alone tells you little about travel time, just as the Acela Express (as limited as it is) easily takes the majority of air/rail traffic between DC and New York, despite slower vehicles and longer trip times – the benefits of easy boarding (Penn Station be damned), downtown station locations, and relatively low security requirements make for a better overall value. 

Musk didn’t just pitch the hyperloop as a way to make evacuated tube trains feasible, he pitched it as a way to make SF-LA travel work better than the CHSRA can. His pitch is part technical concept and part policy proposal, and the policy elements fall short.

Technology does not change geometry: This is true for driverless cars and for hyperloops. The type of technology used doesn’t change the technology’s purpose – moving people from place to place. Since the hyperloop is essentially a transit service, it still must obey the same geometric rules as all other modes, the ones that govern capacity, headway, throughput, etc. Musk ups the speed for his concept, but his own proposals show a very low overall capacity – and even those estimates seem optimistic given his assumptions on safety margins and safe distances between pods. At GGW, Matt Johnson compares capacities of different modes of transport:

According to Musk, pods would depart LA and San Francisco every 30 seconds during peak periods. Each pod can carry 28 passengers. That means that under the maximum throughput, the Hyperloop is capable of carrying 3,360 passengers each hour in each direction.

For context, a freeway lane can carry 2,000 cars per hour. A subway running at 3 minute headways (like the WMATA Red Line) can carry 36,000 passengers per hour. The California High Speed Rail, which this project is supposed to replace, will have a capacity of 12,000 passengers per hour.

Technology also does not magically change the tolerances of the human body (save for science fiction inventions like inertial dampers). Musk is selling speed, and his assumptions on acceleration are more akin to a roller coaster than rapid transit. From The Verge:

According to Powell, that’s a problem: “In all our tests, we found people started to feel nauseous when you went above 0.2 lateral Gs.” The closest comparison would be roller coasters, which usually top out around half a G — but the Hyperloop wouldn’t just peak at 0.5; it would stay there for the duration of the curve. The result would be well short of blackout, which most studies peg around 4.7 lateral Gs, but it would make the Hyperloop challenging for the faint of stomach.

Others have noted the lack of bathrooms in the Hyperloop pods. It would seem that the roller-coaster analogy is apt, as roller-coaster operators don’t want you leaving your seat in the middle of the ride, either. It’s not safe.

Other opportunities for innovation: Much of the praise for the Hyperloop seems to be based solely because it’s something new and exciting (and people take Musk’s cost claims at face value); part of it also seems to be a lack of faith in high speed rail. The desire for something new ignores the reality that most innovation is incremental; it also ignores the power of transportation networks and the value of connecting to something that already exists.

Matt Yglesias looks at alternative transportation improvements that would seek to solve the same problem (decreasing LA-SF travel time) by tackling some low-hanging fruit, rather than inventing new technology. Yet, people don’t want boring improvements in processes. Molly Wood at CNet just wants to believe in technology, noting that practicality is for cynics:

I refuse to keep accepting that until our cynically imagined dystopian future comes to pass. As justone alternative to the essentially already-failed high speed rail project, we now have a detailed plan for a high-speed transit system that could cost as little as $6 billion to build and, by the way, would be solar powered and infinitely more environmentally friendly than the dirty, diesel-powered rail project. It seems obvious that Musk is unveiling this plan ahead of the ground-breaking for the rail project in what is hopefully a successful attempt to stop the monster from ever being born. So get over the sunk cost fallacyof the California “high speed rail” and move on to a better solution.

All we citizens of California, and the Internet, and the world, have to do is believe that this technology is possible. Then those of us with the lucky happenstance of representative government should use it like it’s supposed to be used, and demand better. Instead we tend to give up and talk about great ideas that will never happen — or worse, tear those ideas down as silly, unrealistic, or impossible.

Leaving aside Wood’s unquestioning acceptance of Musk’s cost estimates f0r a technology that doesn’t exist (even in prototype!), and the obvious mis-information about HSR’s power source, this kind of technological evangelism is fine for entrepreneurs (as Alon Levy’s post title argues), but it makes for bad public policy. If Wood had the same faith in HSR, and was willing to look over HSR’s faults with the same starry-eyed gaze, then HSR wouldn’t have the PR headaches that it does.

Slate’s Will Oremus is similarly infatuated with the concept, but at least he realizes the steps required for the Hyperloop to prove itself worthy:

Wise or not, California is unlikely to drop its plans just because one rich guy has a light bulb over his head. On the other hand, if Musk does build a demonstration line, and it’s faster, cheaper, more energy-efficient, and requires seizing less private property than laying down train tracks, a change of plans might start to sound pretty appealing. That’s a lot of ifs—but so is every big idea, in the beginning.

Indeed, that is quite a few ‘ifs.

The odd thing is, despite all of the references to Musk as a master innovator, it’s worth noting that all of Musk’s companies and products, as daring and inventive as they are, still are just incremental improvements over existing technology. Tesla did not invent the electric car and certianly did not build the massive network of auto-centric transport. SpaceX did not invent rockets. SolarCity did not invent solar power. Each company offer incremental (though meaningful) improvements on existing concepts and products.

At the risk of stating the obvious: Hyperloop is not an incremental improvement for an existing technology. Existing technologies have the benefit of linking into existing networks. Tesla’s cars can use regular roads and charge through regular outlets. High speed rail can use existing tracks and rights of way to get into city centers.

This is not a serious cost estimate: Musk is not just proposing a new technology, he is also offering an explicit critique of high speed rail. Plenty of observers have critiqued the CHSRA’s track record to date; comparisons to HSR best practices in planning, construction, and operation from around the world are not favorable. Nevertheless, this does not make the Hyperloop’s assumptions any more realistic. And, if the state were to buy the hype, the Hyperloop would likely see even wilder cost overruns – putting it on the same trajectory as Seattle’s failed monorail transit system.

Alon Levy takes a closer look at some of Musk’s cost estimates, and finds that most don’t even pass the smell test:

This alone suggests that the real cost of constructing civil infrastructure for Hyperloop is ten times as high as advertised, to say nothing of the Bay crossing. So it’s the same cost as standard HSR. It’s supposedly faster, but since it doesn’t go all the way to Downtown Los Angeles it doesn’t actually provide faster door-to-door trip times.

At a broader level, consider what Musk is claiming: that a system of precisely aligned and machined de-pressurized tubes could be built for a fraction of the cost of infrastructure with a similar footprint. Musk is proposing that the pods would clear the tube walls by fractions of an inch, compared to much larger machined tolerances for lower-speed modes of travel:

The biggest question mark is the tube itself, which has emerged as the most genuinely unprecedented part of the plan. By enclosing the track, the Hyperloop is able to sidestep worries about air friction and noise that usually limit the speed of trains to under 400mph, but the tube also presents a unique set of challenges. James Powell PhD, co-inventor of the maglev train, is particularly concerned about the smoothness of the inside of the tube. As Powell points out, the current design allows for just three hundredths of an inch between the tube wall and the skis encircling the pod. “Getting it that smooth won’t be easy,” says Powell, and might require a more expensive production process than the plans envision.

The small gap is crucial to the system’s overall design, allowing for a stable air cushion that keeps the pod hovering frictionlessly in the tube. But the small gap also requires great precision in tube construction. Powell thinks that a single bump, just three-quarters of a millimeter high, would trigger catastrophic damages, possibly even ripping the ski from the pod at top speed. Keeping the tubes straight can be done, but it won’t be cheap. “It’s going to be an arduous process,” says Vinod Badani, an engineering consultant at E2 Consulting. “Quality control and measurement have to be very accurate.” Musk’s plans envision a specially designed device to smooth out the inside of the tube, but it presents a serious engineering problem for anyone thinking of building a prototype.

Musk proposed using I-5’s right of way as a way to keep land costs down. However, I-5 has trucks. If one semi truck jack-knifes on the road, ramming into one of Musk’s cheaply-built pylons, how will his tube maintain that level of precision required for safe operation? Musk asserts his system is safer than HSR during earthquakes (nevermind the safe operation of Japanese HSR during major earthquakes) without any evidence, yet the basic physics of what he is proposing demand a high level of precision on a massive scale.

The real question should be if it’s even possible, not asserting that it will be cheap.

In the New Yorker, Tad Friend takes note of Musk’s propensity for exaggeration:

The bad news is that there’s no conceivable way that the system would cost just six billion dollars, or that one-way tickets would cost twenty dollars. Overpromise disease is endemic to Silicon Valley, but Musk has an aggravated case. When I wrote a Profile of him, in 2009, he told me that a third-generation Tesla would be selling for less than thirty thousand dollars in 2014, the same year that he expected SpaceX’s Falcon 9 to begin ferrying tourists around the moon. Well, no and hell no. More worrisomely, he promised that you could start driving the Model S in western California “at breakfast and be halfway across the country by dinnertime.” Musk is a lot better at math than I am, but he eventually acknowledged that by “dinnertime” he really meant “the following morning’s breakfast”—if, again, you didn’t stop to go to the bathroom.

This isn’t to argue that exploring these ideas shouldn’t happen. It is, however, an argument that a concept like the hyperloop shouldn’t be used to bring down high speed rail. If the Hyperloop is nothing more than a device to force better results out of the CHSRA, that would be a welcome result. However, if that is to happen, it won’t be because the Hyperloop is a realistic (or event a plausible) alternative.

Two weeks ago, Eric Jaffe editorialized that we should stop obsessing about the “next great thing” in urban transportation. There’s thinking big, and then there is fantasy. It’s worth noting that a project like California High Speed Rail is plenty ambitious – it’s certainly thinking ‘big.’ It’s also achievable, but is facing real-world constraints (economic, political, physical, institutional, procedural, regulatory, etc) and is in need of some practical planning.

The Hyperloop may seem like an attractive end-run around these constraints, but such benefits are illusory. The real benefit is in reforming the institutions to reduce the constraints.