On Elon Musk

Every time I turn around, there is Elon Musk. Most recently he had a long TED presentation. TED interviewer Chris Anderson comes off as credulous — I have a bridge, or a tunnel to sell him. Yet reporters can’t get enough of Musk, and I have been asked to do a number of interviews about various Elon Musk adventures, most notably Hyperloop, introduced in 2013 but also The Boring Company. If reporters love him, I assume the public must too.

To be clear, Elon Musk is, like a small share of the population, a genius. Furthermore, like a different small share of the population, he is charismatic. Third he has a talent for setting up businesses, especially those that can get government and venture capital support. Finally, he is audacious, dreaming the impossible dreams of a fifth-grader who has not yet been beaten down by the world. This is a dangerous combination.

Even if he were smarter than every single individual (it probably depends on how you define ‘smart’), could he be smarter than everyone put together? The ideas he talks about are ones that have been discussed for decades or centuries in many cases. It is quite possible that some ideas have not seen progress not due to a lack of intelligence in the rest of humanity, but the nature of the problem. Musk didn’t invent electric money, electric cars, solar panels, rockets, pneumatic tubes, magnetic levitation, or tunnels. But, he has helped push the first few of these things forward, and is a shining beacon of a better technological future in a world mired in political retrogression and tall poppy syndrome.


Musk’s foray into notoriety occurred with PayPal (co-founded with subsequent Silicon Valley superstars Max Levchin, Peter Thiel, Ken Howery, and Luke Nosek). When PayPal first launched on the PalmPilot, it was used as a way to pay people e-money directly (say to settle-up dinner charges or buy a product offline without cash). To induce usage, they gave away some money with each transaction. My friend Lev (who has subsequently earned a PhD in Biology) and his roommate beamed money back and forth on their Palm Pilots to collect this subsidy. I can’t imagine it was a good use of so much intelligence. At any rate, PayPal really hit the big time with its adoption by the eBay community, and eBay subsequently purchased it. I am not a huge fan of the company it has become, but Musk left years ago.


Fast electric cars is one element of Musk’s vision of the future, which to his credit he has delivered to a few of America’s wealthy. To date, the cars have sold abut 186,000, which is more plug-in or battery EVs than GM or Ford, though less than Nissan  (as of 2016 GM’s Chevy has sold 135,000 Volts, and Nissan has sold over 250,000 Leafs). Yet TSLA is valued at $53B as of this writing, more than Ford ($43B), GM ($50.7B) or Nissan ($39B). It is also losing money (losing $667M last year on sales of under $8B), while its elder competitors are at least nominally profitable. Yet he is paid a good salary, $99,744,920 last year, probably to make up for living on only $200,000 a month at his low.

Normally market valuation is the net expected value of future profits. This is sheer guesswork on a company that has never been profitable for more than quarter. Certainly the large capital investments made by Tesla in auto manufacturing and the Giga-factory for batteries may be recovered if sales improve with new models; lots of people have put down deposits on future cars. But announcing and designing a model is a far different thing than delivering it, and the experience of Tesla with mass production is limited, over budget, and behind schedule. Now it is not Elon Musk’s fault that Wall Street has overvalued his company, and like any rational capitalist he certainly should exploit that mania to further his aims, if not to personally enrich himself.

Tesla AutoPilot

Tesla with AutoPilot is pushing forward Automated Vehicles, a second element of Musk’s vision. I reviewed it briefly in 2015, and they are 2 years better now. Using cameras and machine learning to bring about vehicle automation is a plausible strategy, and Musk is promising a coast-to-coast hands-off automated vehicle trip (limited to freeways and charging stations as far as I know), which will be both an important test and a valuable publicity stunt. Thus far only one death in AutoPilot mode has been confirmed (a second death in China is not confirmed by the company), with irresponsible driver behaviour a contributing factor, and the safety statistics are reasonably good, better than humans, which is all we should ask before deployment (and future deployments should steadily be increasingly safe compared to older models). Whether it proceeds to expand road capacity awaits another day, let’s get it working first.


Disclosure. I am a former shareholder in SolarCity. I actually made some money, unlike many people.

SolarCity, a Musk family enterprise, was purchased by Tesla under the reasonable pretext that the solar roofs and battery storage and electric vehicles have synergies. Maybe he also bailed out his cousins. The vision of using distributed solar power to replace centralised fossil fuel power plants is noble and probably good for the environment. Coupling that with storage (PowerWall and the batteries inside Tesla cars) is a necessary element given the vagaries of cloud cover and that damned inconvenient rotation of the earth. The roofing material that embeds solar power collectors is clever, and far less ugly than a solar panel retrofit on an existing sloped roof, and I hope something like that becomes standard.


Rockets are cool. Launching satellites is useful. Given the effective graceful abandonment of the public US space program in the post-Apollo period, and especially the post-Shuttle period, it is good private firms are picking up at least some of the slack with private rocket launches. SpaceX picked up refugees from the late TRW, a space and defence contractor, like CTO of Propulsion and experienced rocket engineer Tom Mueller.

Landing a rocket  (sub-orbital launch vehicle) vertically so that it could be reused seems a hard problem, and saves some metal. I am not clear whether this is a net benefit in terms of energy or pollution, but so far it is still R&D so I won’t worry about that. The plans for private spaceflight carrying humans for exceed what has actually been accomplished, but this is a long process, and eventually a private firm, perhaps SpaceX, will carry a human into space and return them safely to earth.

Mars – Interplanetary Transport System

The most out-there vision of Musk is manned travel and settlement of Mars. To be clear, I thought the movie The Martian was great, and everyone should see it. I would argue the greatest thing our species has done to date is to land a man on the moon and bring them home. It’s sad we haven’t surpassed that achievement in more than 5 decades.

If we don’t destroy ourselves here on Earth quickly, I certainly hope and expect our species will travel and settle other worlds, if for no other reason than species survival insurance. I have forecast we should leave for Alpha Centauri by around 2301. After the Moon, Mars is next in line, and certainly precedes inter-stellar travel. Fifty years after the Apollo program, with far better computers, somewhat better science, and lots more experience in space, it’s well within human technical capacity, though it will be expensive, and certainly dangerous for the first travelers, as even our unmanned probes have faced a number of Gremlins. Far too many unmanned vehicles trying to go to Mars have ended in failure to blithely send humans there.

Can Musk and SpaceX do this by 2030? I hope so, but doubt it. SpaceX’s main claim to fame, (i.e. doing something NASA had not already done decades earlier) landing a sub-orbital launch vehicle so that it can be reused, might be a useful skill, but seems minor compared to the real problem at hand.

In the long run though I favour space elevators to get earth-bound stuff and people into orbit inexpensively. Rockets, which burn enormous amounts of fuel will just further degrade the earth’s environment that escape to Mars was supposed to dodge, especially if we plan to do this at scale, and not for a selection of astronauts. In principle, we could use biofuels in rockets, but that is also decades away, and space-elevators seem likely to be safer and higher capacity once deployed.


Do I really have to talk about Hyperloop? OK. Musk had this idea and then dropped it on a credulous public, and walked away, as well he should. Networks of pneumatic tubes are a great idea that were the cargo mode of the future more than a century ago in Edward Bellamy’s time traveling 2000-1887 sci-fi romance Looking Backward.  Somehow that never panned out aside from a few niche cases. In Pneumatic Philadelphia, Harry Kyriakodis writes:

But pneumatic tube systems gradually lost their appeal when they became too expensive to maintain. Moreover, it was clear that the systems could not keep pace with the movement of ever-changing business centers in municipalities. Beside this, the tubes were never particularly efficient. About 90 percent of the power generated for the air pressure was wasted in pushing the air through the tubes. Furthermore, motorized delivery trucks came into use and were found to be much more efficient for transporting mail in urban centers.

Magnetic levitation is a plausible mode of mobility that is actually in use in Shanghai and a few other sites. Putting maglevs into pneumatic tubes (or semi-pneumatic evacuated air tubes) helps reduce air resistance. But if you want to get the speeds and accelerations they claim, you really should test it with actual people before you propose deploying whole lines or  networks. The Wright Brothers didn’t posit a hub-and-spoke jet aviation system before Kitty Hawk.  And there are all sorts of other problems that might arise, some we cannot imagine without both non-human and human testing. While college student participation in Hyperloop design contests is wonderful and motivating, one gets the feeling the whole enterprise is being run by amateurs and confidence men, even more so than transport in general. That people who have taken the names Bibop Gresta or Brogan BamBrogan are (or were) involved in the two main companies should be a clue.

The companies can’t decide if they are for people or freight. If they are for freight, we need to ask how fast you really need that pizza? If they are for people, you might ask how fast you can disgorge that pizza you ate earlier. My view is summarised on this podcast.

The Boring Company

In Musk’s terms, congestion is “soul-crushing”, and indeed it is, which is why I walk to work. Still, if you want to gather many people together in a small space, there is inevitably crowding in access just as there is crowding at the destination.

Musk says a network in tunnel-based transport is feasible if a ten-fold reduction in tunnelling costs can be achieved. (Tunnels are about ten times as expensive as surface construction, so I am not sure ten-fold is enough.) US infrastructure costs too much, and this applies to tunnels as well. The reasons for this are many. In the TED video, Musk suggests several techniques for cost-reduction, including reduced tunnel sizes and the process of continuous and faster boring.  I assume the reduced tunnel sizes must somehow be offset by many more tunnels if this is to actually work, and the more tunnels, the more access points and the deeper they need to be to avoid interactions.

Cost and usefulness depend on how access is done (an elevator is suggested, with cars on “skates” so that cars don’t drive in the tunnels (and they are electric cars anyway, so ventilation costs are reduced.) While the travel in tunnels may be fast (like air travel or high speed trains), the door-to-door times are greatly increased because of access costs to get to and from the tunnels.  I don’t fly to the airport, for instance, greatly lowering end-to-end average speed, even if the airplane is very fast.

Is this feasible?

At the extreme, we can imagine that the entire surface street network is replaced with elevator access portals, so no long-distance horizontal vehicular movement occurs on the street system, and every on-street parking space is an elevator (as a point of reference, the City of Minneapolis has about 7000 metered on-street parking spaces), but we probably don’t need that many elevators. Alternatively we can imagine giant parking structures at the final destination, that just drops the person off. We also need to imagine we can find tunnel access points where utility location is not an issue.

The tunnels can intersect (in an underground but grade-separated way) so that anyone can go point-to-point. We can also assume not every car accesses the system simultaneously, and when a car enters the system from a portal, another car can exit at the same portal (parking space), where they discharge their passengers, hopefully near their final destination. It is not at clear why there are cars riding on skates at all, rather than just people in capsules, (or magical floos) but, ok, assume there is some transition between today and the final end-state of such a network.

How big do the parking structures need to be to allow everyone to be able to drive to work. This figure shows how large it would need to be for Minneapolis.

The 278 Story Parking Ramp Downtown Minneapolis Would Need Without Transit
The 278 Story Parking Ramp Downtown Minneapolis Would Need Without Transit https://streets.mn/2015/12/04/the-278-story-parking-ramp-minneapolis-would-need-without-transit/

Now people could share cars (not simultaneously, but in sequence), but if they are all using the space (downtown) at the same time, it is not clear how many runs can be done with a vehicle to get everyone downtown so they have a reasonable number of shared hours together (which is the whole point). Cars could also be single person, reducing space requirements, so let’s reduce this four-fold, two runs per vehicle and half-sized vehicles. Instead of 278 stories, it would ‘only’ need to be 70 stories. (or 2 structures of 35 stories, or 4 structures of 18 stories, you get the idea).

Of course if tunnelling is cheap, presumably excavation is cheap too, and these can all be underground, and people would need to take up to a 278, 72, 36, or 18 story elevator ride before reaching the surface and walking to their final destination.

Now Musk might argue with the assumptions about no surface travel, but he then has the problem of where to put all the tunnel access points, of which there must be a lot to make a dent in surface congestion. Each lane of traffic on a freeway can move 1800-2400 vehicles per hour today. Perhaps his system can move 3600 vehicles (1 second headways between vehicles). Getting a vehicle onto the skate onto the elevator will take more than a second (as shown in the video), so there need to be more skate portals than tunnelled lanes in the downtown area (leave aside exits). So for Minneapolis, with 160,000 workers, of which say 80,000 want to leave in the peak hour, at 10 seconds per vehicle per elevator (or optimistically 360 per elevator per hour) would need 80,000/360=222 elevators to serve everyone by ‘car on skate’ for his tunnel-vision. Since cars are exiting in 4 cardinal directions, this is about 55 per direction. If Minneapolis CBD were 100 square blocks (10 blocks long by 10 blocks wide, which is approximately true, depending on how you define downtown) there would need to be 5-6 elevators on each street, about the same number as lanes on each street.   So instead of exit ramps onto freeways, each road could terminate at an elevator, or you could put some elevators in parallel where normally you would have on street parking. However this would make entering the portal slower, so you would need more.

The system appears to be centrally controlled, some public utility owns the skates, while individuals own their cars which use the skates. It has been described on Twitter as public transport for cars. It is not at all clear why this is better than public transport to near the final destination, and cars (or bikes, or walking, or buses) for the ‘last mile’ (the park-and-ride model so popular with downtown commuters).

In any case, this is probably technically possible, if economically dubious. It would be much simpler to price roads, and uses the revenue to pay for the tunnels. But then fewer people would buy and drive Teslas, and there would be less need for tunnels.

That this is the latest idea of “the smartest man in the world” (really, click the Google link, there are many believers) worries me.


Overall, Elon Musk is interesting, in a techno-hucksterish sort of way, but is spread way thin and simply hand waves many real technical problems. If he just had a blog, that would be fine. But because he has actually done things, his crankish ideas carry far more weight than good, if more modest, mainstream, and plausible, ideas from the less famous. Even more than Steve Jobs, who delivered his vision of personal computing in far greater scale and has actually changed billions of people’s lives, the Reality Distortion Machine is strong with this one.




Acknowledgments: Thanks to Mark Palko for some comments and links on an earlier draft. Also some points from Brian Taylor. The opinions are those of the author.