site banner

In Defense of The Boring Company

I'm going to talk about what the Boring Company is doing and why I think it is not only not a terrible idea, but actively a good idea!


This is a complicated idea with a lot of moving parts, both metaphorically and literally. You will have totally reasonable questions! Hopefully they will be answered by the time I reach the end, but keep reading until you get to the end; in written format I can only answer questions one at a time, and your specific question might take longer to get to.

In addition, this is describing the system that I think Elon Musk is working on. He hasn't announced that this is what he's working on - it's guesswork and theorycrafting by me - but there is some evidence to it.

A summary: Elon Musk is attempting to redesign urban and suburban transportation on a grand scale, so thoroughly that the majority of commuters choose to use this system because it's better. This is not a thing you accomplish by building a few tunnels under Las Vegas. The Loop is a prototype of a prototype of a prototype; the beginnings can be seen there, but claiming his plans are invalid because of Loop's problems is like criticizing the concept of trains based on Locomotion #1's terrible speed.

Elon Musk has a unique goal: to make a fast inexpensive public transportation system.

Uber and Lyft have a similar goal! They want to make a fast public transportation system, and they have succeeded! They don't care about inexpensive, and in fact they can't accomplish inexpensive, because drivers are expensive. They're working on self-driving vehicles, and this will help, but it won't solve the issue because Uber and Lyft need lots of roads, roads take up land, and land is also expensive. Note that land isn't just financially expensive, it's valuable - we only have so many square meters of sunlight surface on this planet, and it's a shame we're using it on transportation. This is opportunity-cost even if we don't normally count it as a cost of roads; it's kinda factored in right now because we don't have an alternative, but we could have an alternative and we should consider land usage as part of cost.

Car manufacturers also have a similar goal! They want to make a fast inexpensive transportation system, and they have succeeded! They've abandoned "public" by requiring people to buy into the system with a large upfront expenditure (specifically, "buying a car".) This allows them to get rid of that whole "pay for a driver" thing - the passenger is the driver. It's not as inexpensive as it could be, though, because cars need lots of roads, thus land, thus expense.

Public transport systems also have a similar goal! They want to make an inexpensive public transportation system, and they have succeeded! But it's not fast. In fact, it cannot be fast. Group transportation is intrinsically slow; putting more people on a vehicle either requires frequent stops which slows down everyone else on board, or it requires stops at junction nodes which implies transfers which also take a lot of time. Short-to-mid-distance buses, trains, and subways cannot match uncongested cars, and you can test this on Google Maps by going to a city of your choice, picking two positions, and twiddling with the "Start At" option until you find the fastest times for cars and the fastest times for public transportation; in almost all cases, cars are significantly faster, and I've never found a case where cars are more than a minute slower.

(Airplanes have the same problem, but they're fast enough that people put up with it; nevertheless, an airplane trip still involves an hour or two of bureaucracy and waiting on either side, and chances are good you're not landing at the exact time you'd prefer to. Long-distance trains also have the same problem and the same solution, specifically, "we put up with it because the speed makes it worth it". In both cases, avoiding all that added complexity would make them significantly better. If you can think of a way to accomplish that without a drastic price increase you will become extremely rich.)

tl;dr: Transportation has traditionally been "fast, inexpensive, public; pick two", and Elon Musk is trying to pick all three at the same time.

The Basic Idea

If you haven't heard of the Boring company or the Las Vegas Loop, here's the concept:

Elon Musk thinks tunnels can be built for much cheaper than they previously could be. He is building a large underground network under Las Vegas, with something like 45 stops (this number keeps increasing as they add more to the plan). You will walk up to a stop, request a car, and travel to any other stop in the network. You can do this today, although right now they only have 3 stops, but construction continues.

This is literally the basis of the plan; "let's make tunnels and drive cars through them". I acknowledge this sounds dumb, but it may actually be the best way to accomplish Fast, Inexpensive, and Public.

Let's tackle the easy ones first.

Boring Company tunnels are public because you don't need to buy in with a large investment to use them. You can just show up at a stop, pay a fare, and ride a vehicle to wherever you want to go.

Boring Company tunnels are fast . . . sort of . . . because it's point-to-point transportation. The vehicle is ideally already at the stop, or close by, when you request it, and it takes you directly to your destination, as long as your destination is on the system. This "on the system" limitation is a flaw! We'll get back to that, though.

Boring Company cars currently require drivers, which is expensive. They've said multiple times that this is a stopgap until they have self-driving working. I see no reason to doubt them and the rest of this post is going to take on faith that they'll get self-driving working. Again, prototype of a prototype of a prototype. If you're skeptical about self-driving in general, note that as of this writing there are multiple companies running public services in multiple cities; if you're skeptical about Tesla self-driving, well, me too, but they can always license it. I'm going to just accept this part as solved-in-the-next-decade-one-way-or-another.

Boring Company tunnels are inexpensive because oh god this is where the complicated part starts


Tunnels are, traditionally, very expensive.

There's a lot of reasons for this. Cost disease, in general, is one of the big ones, and if Boring Company gets hit by cost disease then this entire thing might be doomed. I think they're more resistant to this because they are not having cities come to them asking for services, they are going to cities to propose services, and if they're expensive, they won't get any contracts. Note that Boring Company has already turned down a contract because the company was going to waste a lot of money on things that weren't the tunnel, and they just didn't want to be a part of that. I'm going to just cross my fingers that this doesn't happen.

Tunnel size is another big one. Tunnels get much more expensive as they get larger. Train tunnels need to be surprisingly large; they need to hold a train that's big enough for people to stand up in and walk around in. They also need to hold some kind of emergency exit system. With trains, this traditionally hasn't been compatible with the train rails themselves; the cross-ties are a tripping hazard. If you have to run a second extra walkway next to your train then that makes your tunnels even larger. Finally, you need a lot of emergency equipment. The reason this is required is that stations are rather far apart; if stations were closer, the safety regulations let you basically say "look, there's an exit right there, just walk to the exit". Far-apart stations cause significant added tunnel expenses.

The biggest issue, surprisingly, is the underground stations. The most common way of making an underground station is as simple as it is costly:

  • Knock down all the buildings above the station

  • Dig a giant rectangular hole

  • Reinforce the top of the hole

  • Fill the top of the hole back in

  • Build new buildings on top

This isn't a lack of foresight on the part of the builders, this is actually how it tends to be done. Underground stations are horribly expensive, and this has consequences for the rest of the system. Remember how I kind of skimmed past "far-apart stations cause significant tunnel expenses"? Well, they do, but this is still cheaper than building more underground stations!

This is how the Boring Company is going to solve tunnel price:

  • Cars are much smaller than trains [citation needed] and don't require as much sheer size.

  • Cars travel on concrete, not rail, and this surface is perfectly suited for passenger exit, meaning that you don't need an extra passenger lane as long as there's enough room to get past the cars. (Note: in the current Loop tunnels, there is, even though it's not obvious in a lot of the videos that have been posted. It's not comfortable, but it's enough for emergency evac.)

  • We can reduce the necessary emergency equipment by having frequent stations. Trust me on this for now! I'll get back to this one very quickly.

All of this put together makes Boring Company tunnels a whole lot cheaper than train tunnels.


Twice, now, I've glossed past issues with stations. The Las Vegas Loop requires stations at every stop so people can get on and off; our emergency system also requires frequent stations. These can both be solved by having lots of stations.

but wait, I thought stations were expensive Nope! Stations are cheap. Underground stations are expensive. The solution is that you just put your stations above-ground. Any parking lot can become a station terminal, as can underground floors of already-constructed buildings.

This works for Boring Company cars because car station positioning is far more flexible than train station positioning is. Train stations have to be long because trains are long; cars are short and so car stations can have basically any layout. Trains run on rails, which have extremely low friction - this is good from an efficiency perspective, but means that trains cannot handle significant slopes without expensive equipment like cable cars. If trains can't handle slopes then above-ground stations for underground rails simply aren't possible. Meanwhile, the minimum footprint of a full-fledged aboveground car station connecting to an underground network is the same footprint as a small house; a tunnel up, a tunnel down, and a few parking spaces, done.

Now we have cheap stations! We can toss a station at every casino on the Las Vegas Loop and not think twice about it. Our tunnels become smaller because we don't need as much emergency equipment, and our trips are faster because you can enter and exit from the cars in more places.

This is a reasonable solution. But it's not a great solution. We still have to drop people off at stations and pick people up at stations; what if someone doesn't have a station nearby? What if someone wants a car from their house off in a suburb or true rural area? Do we need to build tunnels to every single neighborhood, and then require that people walk across half their neighborhood to get home? It's 108 degrees out right now, I'm not walking in that weather. Screw that. And worst, we still need significant land dedicated to this system for the parking-lot terminuses, and land, as I've mentioned, is expensive.

We can do better.

Stationless Point-To-Point

This is where I move into speculation territory. But I really do think this is the plan.

We have an underground network of self-driving vehicles. We have cheap entry and exit tunnels. This is all we need to finish the entire system.

We keep our entry and exit tunnels, and we put them everywhere (which also solves our emergency exit requirements.) However, we get rid of the stations. The tunnels are simply a way of transiting from the underground network to the aboveground road network. "The aboveground road network", you ask? Sure; we're going to co-opt the aboveground road network for part of this. We're not using it for long-distance travel, so we can get rid of the giant tangles of freeways and onramps. But we are using it for last-mile travel, because it's there.

When you request a vehicle, one shows up at your doorway. You get inside and it heads to the nearest convenient tunnel entrance. Most of your trip is spent underground, and then it pops back up into the sunlight to bring you straight to your destination.

No stations, low land usage, point-to-point congestionless travel.

That's the actual goal.

Common Objections

Moved to its own comment due to character count limitations.


The goal of the Boring Company is to make the first fast inexpensive public transportation system. Cars are fast and kinda inexpensive, but not public; Uber/Lyft are fast and public, but expensive; trains and buses are inexpensive and public, but not fast. Elon Musk is trying to get all three at once, and the decisions being made are in service to that. The thing being designed really could not exist before self-driving vehicles; it is a truly 21st-century transportation system and hopes to redesign the urban landscape on a level that we haven't seen in a century.

I have no idea if it will succeed.

Jump in the discussion.

No email address required.

Common Objections

But hold on, in this plan we still need roads.

Yeah, we do. But we need roads anyway.

We need roads for emergency vehicles. We need roads for utility vehicles. We need roads for cargo delivery and for oversize transport. I just replaced my roof; how are you planning to get a full pallet loaded with tons of shingles to my doorstep any other way?

Roads are useful and they're not going anywhere. But a lot of these things don't require big roads, they require some roads. In the eventual future, the plan would be a dramatic reduction in road width; two-lane residential roads become one-lane residential roads, four-lane arterials become two-lane roads, highways mostly vanish.

If they must exist, and they must, then we may as well make them do double-duty. Below a certain threshold of usage, roads cost money simply by existing because they require regular maintenance to protect them from the elements. Usage up to that threshold is free, and using them just for the last-mile trips is going to be very light usage compared to how they're used today.

Why not use rails for this?

Remember, we need the ability to have inexpensive entries and exits from the system. This is both so we can provide easy emergency exits, so we can reduce the amount of surface road required for the last-mile trips, and so we can spend less land for the entire project. Rails have low friction which makes significant slopes impossible; without significant slopes, we cannot have cheap connections between underground and surface; without those cheap connections, the entire system economically falls apart.

In addition, part of how the system works is that self-driving cars can travel along residential roads to bring you straight to your destination. Rails would require that we build rails into every road in residential areas. This is just not economically feasible.

Isn't this last-mile travel still going to cause problems in downtown areas?

While I've been describing surface entrances and exits as the normal way of getting in and out of the system, you don't have to use surface exits. Skyscrapers already have enormous basements, and you don't need more than a single floor to build a terminus. In this case you really could have stations, located in multiple building basements; the car delivers you straight to your workplace and you go straight from car to elevator.

Cars pollute, and this is still going to cause lots of pollution.

There's no way this is going to be based on internal combustion engines - that makes tunnels far more complicated to begin with. All electric, all the time.

Electric does not solve the problem entirely. A significant source of car pollution is actually particulates from tire wear. But a lot of the tire wear will be localized within tunnels, and can be swept up and disposed of much more efficiently.

The mass number of cars required isn't going to be cheap or great for the environment. But it may be better than subways. Your average subway car costs almost a hundred times more than a single car, and costs similar amounts more to maintain. Fleet cars gain huge advantages from massive mass-production and you simply cannot do that with subways. In addition, subways often run mostly empty, and the cars in this system will rarely be running empty; a car sitting waiting for a passenger is not polluting, while a six-car subway train carrying a total of three people is polluting quite a bit.

Don't you still need big parking garages in downtown areas?

Maybe! The actual numbers are currently unclear.

But this isn't as big a problem as it sounds. If we have cheap tunnels, then we can make dedicated parking tunnels. Fleet cars are entirely homogenous, and we never need to access any specific car, we can just cram a bunch of cars in a tunnel and do first-in-first-out retrieval as they're requested.

We don't even need dedicated parking tunnels. Because these are all computer-controlled, we can dynamically reprovision transport tunnels as they're not needed. Maybe we need four parallel tunnels in one area for peak hours; well, once the peak is done, we can turn that into one travel tunnel and three storage tunnels. As it gets closer to peak, clear out a storage tunnel and turn it back into a travel tunnel.

A lot of this is very much hypothetical and requires a lot of software development. But it's all software, not hardware; this can be improved as the system grows.

Wouldn't you need to build the entire thing all at once, just to find out if it works?

Not at all!

The Las Vegas Loop is a good example of this. You can build one tunnel, then connect another tunnel to it, and just keep on going. Cars don't require complicated switching systems like trains do, and computer-controlled cars don't require complicated signage. In practice I expect this will generally start with the backbone of a city, then get expanded as it gets used, in order to move more traffic off streets and into the underground.

How expensive does this actually end up being?

Nobody knows!

Seriously, nobody knows. This is all quite experimental. I made an argument above that Boring Company tunnels will be a lot cheaper to build than train tunnels, which is, I think, true. But you'll need more Boring Company tunnels, because each one is incapable of holding as many people as a train tunnel, and we expect more people to use them, and we expect people to travel longer distances in them too. It adds up fast.

Traditionally, this is handled by trying to buy property, then eminent-domaining it if necessary, usually with some detours through poor areas to cut down on price and voter dislike. But this is the old model where building a station requires knocking down a building. If we're going completely underground without needing that aboveground work, this should be cheaper.

Subsurface rights are complicated; some properties technically own all the land beneath them, some properties (especially in Texas) own only a few feet down and the rest is owned under "mineral rights". Some cities explicitly reserve the right to run underground utilities under your property without consulting you. Even if you do own the rights, eminent domain can still apply, and how many people are really going to object to getting a few thousand bucks in return for a slice of rock a hundred feet down?

Often train tunnels get run under roads as much as possible, specifically to avoid the ownership issues. Low-speed car tunnels are more agile than train tunnels, able to make tighter curves and hug the road network better than trains. High-speed car tunnels aren't, but they can be run deeper, which is likely to cause fewer property arguments.

I'm reasonably confident that property costs won't turn out to be a big problem, assuming the Las Vegas Loop turns out well and other cities start salivating at the idea of their own Loops; minor things like "unused property rights" tend to fold quickly in the face of government interest. Practically, though, we won't know until the lawyers are done with it.

Isn't this the same thing as that vacuum-tunnel Hyperloop idea?

Nope! But I can't blame you for being confused.

The Las Vegas Loop has, officially, nothing in common with Hyperloop, besides the word "loop" and the fact that both of them are intended to move people. Elon Musk has mentioned that he wants to put some effort into building Hyperloop tunnels with Boring Company but so far he's done nothing more concrete than speculation.

I think, assuming Hyperloop can be made to work, it could turn into the long-distance leg of a country-spanning network using Loop as the short-distance part. Don't hold your breath on that one though; in a post that is almost entirely speculation, that is, like, speculation-squared.

In this model, you can't own your own car, and people won't like that.

Nothing stops you from owning your own car! Boring Company has said that they would allow privately-owned cars in with the appropriate self-driving software. I suspect most people wouldn't own their own cars, but if you want to spend the money on it, you're welcome to do so.

In addition, the ground road network does still exist! Likely cut down quite a bit, but only because of a lack of demand; if you want to use it, you're welcome to.

Is there anything you couldn't find a good place for but want to mention anyway?

Yeah: Weather!

In some areas of the world, traditional roads are very expensive to keep clear and have constant maintenance burdens due to freeze/thaw cycles. Putting the roads underground reduces the impact of weather and may cut down on maintenance costs considerably long-term. This also protects cars from weather, improving driving speeds during rain and snow.

(Comedy option: install air curtains at every entrance and exit, then heat the entire underground section from waste heat in order to reduce per-vehicle vehicle heating costs. I have no idea if this would be cost-effective.)

There's potentially a flooding problem during extremely heavy rains and, obviously, on coastal flood-prone cities. Similar to how aboveground roads have to shut down during heavy snow, the underground network would probably just have to shut down during flood conditions.

In general I suspect this ends up being a net improvement, although it obviously isn't a panacea.

Are you getting paid for this?

I own Tesla stock, and I would own Boring Company stock if I could. But besides me buying some of Tesla's stock, I have no further relationship (that I'm aware of, at least!) with any Elon Musk company.