Hacker News Reader

> You assert this like it's a fact. But it seems like something that could be proven wrong. I don't see any real fundamental obstacles here.

I think it's just a safe conclusion based on some reasonable assumptions. Just to pick one as an example, Wiki tells me a NY subway train is usually 8 to 10 cars. I'll use 10 to make the math easier. They're between 15m to 23m long so I'll go with 20 and that's 200m. Capacity is around 200 people each which gives us 2000 people per train.

A model Y is 4.7m long and can seat up to 4 people. To get to 2000 people capacity you then need 500 model Y.

500 Model Y, one after the other in a straight line is 2350m so already to get to the same capacity you need 10x the space without even factoring in the fact you need to add space in beteen each and every one of those 500 cars as a safety measure.

At a 2 seconds clearance between vehicles, the Tesla system at peak capacity of 4 passengers will move 2 people per second. So it's equivalent with a subway train at crush capacity running every 1000 seconds or 16 minutes. That kind of frequency is typical for some longer haul train links, airport trains with underground sections like the RER etc.

Of course, there are other factors that preclude running Loop at 2s frequency, like station capacity and ensuing bottlenecks on ramps etc., but bear in mind there is no comparison between a train filled at crush capacity, with most people standing and smelling someone else's armpit, that you have to wait for a quarter of an hour, vs a leather seated car that takes you point to point to your destination without stopping at any intermediary station.

The point is, they are very different systems that don't compete against each other. If you need a subway and have the money for a subway, by all means, build a subway. Loop competes with light rail, and it says it can do better than most of the light rail systems in operation, at least in the US. Both its claimed pphpd and per mile cost are sufficient to be financially competitive, to the point that it can attract private capital and generate profits.

And the pphpd they demonstrated, of nearly 5000, was achieved with Tesla cars. If they manage to introduce 8-12 seaters and cracks self driving in the tunnels (a much simpler problem then full on road self driving) than it could compete with many of the less congested subway networks.

It seems very irrational not to let them demonstrate what they claim, as long as they are using their own money. It's transport capacity that does not compete with other transit options and does not capture significant urban real estate.

>At a 2 seconds clearance between vehicles, the Tesla system at peak capacity of 4 passengers will move 2 people per second. So it's equivalent with a subway train at crush capacity running every 1000 seconds or 16 minutes. That kind of frequency is typical for some longer haul train links, airport trains with underground sections like the RER etc.

Typical where? Any well functioning subway system I’ve been on has a much higher frequency than that. Every 3-5 minutes.

" ...typical for some longer haul train links, airport trains with underground sections like the RER etc."

I am certainly aware of the frequency most subways run at, that's why I clearly defined a comparable system, and later in the post underlined that Loop is not a direct competitor with subways but with above ground light rail.

I have no idea why people get triggered so easily and jump to type out a response without finishing reading the post - or, in this case, the sentence - they are replying to. It must have something to do with the compressed attention spans that come with smartphones.

> So it's equivalent with a subway train at crush capacity running every 1000 seconds or 16 minutes.

The Berlin subway typically runs 4-minute and 5-minute intervals (except at the periphery) and they're seeking to upgrade it. Some more advanced systems in the world run at 2-minute or even 90-second intervals. So, by your calculation, the tunnel has the capacity of 1/4 of an old subway and 1/10 of a cutting-edge subway.

> So, by your calculation, the tunnel has the capacity of 1/4 of an old subway and 1/10 of a cutting-edge subway.

Yes, if you use it with 4 passengers Model Ys spaced 2 seconds apart. There is nothing in principle that precludes using 8-16 minibuses (2-4x increase in raw tunnel capacity) or even linking multiple such cars to form a train, building essentially a rubber tired subway.

The point is, the tunnel diameter is not limiting throughput. It's a flexible system where you can mix multiple types of vehicles with various density, comfort, price points, number of stops to destination etc. In the extreme, you can have multiple station designs, so that a limited number of larger stations can handle the multi-cart trains, on a fixed schedule and stopping in every such large station, while a small number of lower capacity stations and spurs can handle the luxury point to point cab service in Model Ys or similar that will take you from your hotel to you casino or airport.

Essentially, Loop has its own private transport infrastructure and they will search a mix of vehicle types and station designs and locations that will maximize revenue and infrastructure utilization. It's mind boggling that some call it a failure when it's only beginning to create the system and unlock the network effects.

The metro in my city runs automated trains every 2-3 min at rush hour (and I think it can run them every 99s if needed) there is zero possibly the joke of a system using Tesla could keep up

And sure enough 5000 ppl is a far cry from 25,000 it’s rated for with mark 2 trains

> the Tesla system at peak capacity of 4 passengers will move 2 people per second.

Your analysis is incomplete. A moving sidewalk on the airport can move far more than 2 people per second, and needs no driver. Where is the call to revive that form of travel descried in Heinlein's "The Roads Must Roll" and Asimov's "The Caves of Steel"?

You are missing speed, and acceleration, and load time.

> So it's equivalent with a subway train at crush capacity running every 1000 seconds or 16 minutes.

Well-frequented London lines run with something like 2-3 minute intervals, not 16 minute intervals. And you can easily carry luggage with you, push a stroller, or roll your wheelchair on in.

> and smelling someone else's armpit

Yes, it appeals to rich people who want to use their money to avoid even minor irritations, and don't want to pay taxes to support mass transit that helps everyone. (From the link, "some press and discussion around the Loop figures it as more of a luxury option: something that casinos can comp for high rollers, that will spare people dealing with the general disaster of getting around the strip".)

But that means having an alternative for us poor stinking masses. The Loop is not that solution.

By the time the subway system gets to the crowd level you describe, a Loop-type system would be well over capacity. You'll be stuck in a line waiting for the next Tesla to arrive, surrounded by stinky people.

> was achieved with Tesla cars. If they manage to introduce 8-12 seaters

Why are they using Teslas? As the article notes, "due to their larger seating capacity and faster boarding/deboarding, the Loop would likely achieve a higher capacity if they just shifted operations entirely to the Club Cars".

With "four rows of seats", it's likely something like https://www.clubcar.com/en-us/commercial/transport/villager-... which says "A long time classic on resorts, the Villager 8 moves up to eight passengers in a single trip and can replace expensive vans. Yet it also travels narrow areas not accessible to conventional vehicles."

It seems they could easily increase capacity now by simply changing vehicle type.

Also, since the capacity numbers were achieved with Tesla cars, that means it excluded the GEM carts which the article says is uses for people with wheelchairs. That suggests the "nearly 5000" number does not include people with mobility issues, which would need to be supported under ADA requirements.

At peak capacity, even a small disturbance can greatly reduce flow.

> It seems very irrational not to let them demonstrate what they claim

The page says their claim is that the fully expanded system will target only 90k passengers per day. It seems very irrational to not draw conclusions from that.

The idea to use moving walkways is completely fantastical and has no bearing on this discussion. An automated walkway is just a (poor, slow and very expensive to build and maintain) transportation device, comparable to an electric bus. But electrics buses are not, by themselves, a solution to traffic because they must share the congested grid with all other above ground transport modes.

Even if you would have the money required to build a mile of bidirectional walkways and the necessary enclosure (without which they would break down at the first rain), you still don't have the required right of way. And if you do get it, you would be completely harebrained to employ it for this project to double the walking speed of travelers, instead of, say, a BRT system, a bike or electric scooter lane etc. that are in every way superior solutions.

Loop is comparable with subways because in each case we are talking about dedicated infrastructure that does not compete in the fixed sum struggle for street space. And it's primarily a tunnel network that has very high per tunnel theoretical throughput, see for example the London subways. The exact vehicle mix that's revenue maximizing for that infrastucture is still to be determined. It could remain a luxury taxi service or it could well evolve into a rubber wheeled metro. Probably they will do both at the same time.

> Why are they using Teslas?

Because it's the minimal viable product that allow them to meet city milestones and continue to evolve the system. Developing a completely new and custom vehicle for passenger transport is a massive risk, money and time sink that will be justified only later. Golf carts won't cut it for a variety of reasons.

It has bearing because it highlights how your analysis, based on "will move 2 people per second" is incomplete, as that same analysis would conclude that moving sidewalks are much better.

You do realize I'm referring to an electric walkway in a dedicated tunnel, right? https://en.wikipedia.org/wiki/Moving_walkway lists quite a few of those, including in the several-hundred meter range, for place where "a BRT system, a bike or electric scooter lane etc." are not superior solutions.

> Developing a completely new and custom vehicle for passenger transport

Very unlikely. As the article points out, the current system already supports golf cart systems, likely as an ADA requirement.

Why do they need a new and custom vehicle? The expense of special-purpose designs is one of the reasons monorails aren't popular.

What does a Tesla demonstrate - in terms of mass transit goals, not "cool, it's a Tesla" - over a golf cart system, or golf cart train, with higher capacity and better accessibility?

> an electric walkway in a dedicated tunnel

It's highly unlikely they would work in tunnels, where real estate is even more constrained than at street level, and where the capital investment is substantial. Walkways are a compromise between ingress delay and wait time (at which they kick ass) vs total speed, they only make sense for relatively short runs; you just wouldn't spend tens of millions per mile to double or treble walking speed.

But it any case, once you build the Loop network, there is nothing in principle stopping you to retrofit a moving walkway into the tunnels. Your initial retort was more of a reduction ad absurdum "if we don't do this with moving walkways... then surely using Tesla is worse idea". That line of argument simply doesn't work, a system using Tesla can hit two relevant metrics (pphpd and average speed) while the walkway system can hit only one (pphdd), and the good performance on that metric for that does not make up for the latter, especially considering the eye watering costs.

The club cars are used for tunnel evacuation, they are not disabled friendly; disabled passengers are handled with a few specialty minivans. So club cars are an unsafe and low autonomy solution that can get the most people out of the tunnel in a few trips, without worrying about speed, crash rating, if the doors open etc. (they have no doors). They would certainly be unacceptable as daily drivers at 40Mph+

Travel speed matters too. At non-peak times, when queues are short, going 100 kph will give much shorter commute times than moving sidewalks (5 kph) or club cars (25 kph).

Yes, travel speed is the first part of 'You are missing speed, and acceleration, and load time.' :)

In those SF stories, moving sidewalks were more then 5kph. "The Caves of Steel" had parallel strips, each at a faster speed. Quoting https://archive.org/details/robotnovels0000isaa/page/10/mode... :

"He stepped from strip to strip with the ease of a lifetime’s practice. Children learned to “hop the strips” as soon as they learned to walk. Baley scarcely felt the jerk of acceleration as his velocity increased with each step. He was not even aware that he leaned forward against the force. In thirty seconds he had reached the final sixty mile-an-hour strip and could step aboard the railed and glassed-in moving platform that was the expressway.'"

60 mph is 100 kph.

"The Roads Must Roll" had a similar design. The first version (quoting https://archive.org/details/pastthroughtomor0000robe/page/42... ) was:

"based on the ore belt conveyors of ten years earlier. The fastest strip moved only thirty miles per hour, and Was quite narrow, for no one had thought of the possibility of locating retail trade on the strips themselves. Nevertheless, it was a prototype of social pattern which was to dominate the American scene within the next two decades—neither rural, nor urban, but partaking equally of both, and based on rapid, safe, cheap, convenient transportation."

The first high-speed road went 80 mph. https://archive.org/details/pastthroughtomor0000robe/page/44... .

If we look to the moving sidewalks of now to understand the future dreams of the 1950s, then we should of course also look at the PRT systems of now to understand the problems with the Las Vegas loop.

The only reason it's interesting to share farts and bad breath with that many people is that subway tunnels in NY are really expensive to make. They cost many billions to dig and even minor extensions to existing tunnels seem to to turn into multi billion$ projects. And then you need these massive subway stations that also cost billions. Doing things at that scale seems really expensive and ripe for disruption.

Your assumption is that there would be just 1 tunnel instead of many. Probably because you are assuming tunnels are super expensive and that that is something that cannot improved. The premise of the Boring company was that they would be able to cut cost by some orders of magnitude by doing things at scale. They haven't yet, obviously. But that doesn't prove the basic premise wrong.

Cheap tunnels mean you could have many more tunnels between many more destinations; not all of which need to be super dense cities like New York. Which means you don't have to squeeze millions of people through a single one of them. The cheaper the tunnels, the less the capacity matters. Just add tunnels to scale if things get congested.

And of course the model Y is hardly the final answer in pod design. There's no good reason why we can't have a mix of smaller and bigger ones. In fact, using a big lump of an SUV designed for road traffic for this seems not very optimal.

Note that 2350m is 6 minute headway at a 17mph average subway speed, which isn't at all crazy for a subway.

A single-tunnel Boring Loop isn't competing against the 10th busiest subway, though. Even if Boring Loop were competing against New York City Subway, at $2.5B/mi you could build something like 50 parallel lanes at the same cost, and 2350m/50 is only ~50m. Space them out a bit so it's actually driveable and you're probably still ahead with Boring Loop, even assuming the trains are running full (which is not default).

Now, if you did invest enough for 50 parallel lanes you clearly wouldn't want to do 50 parallel lanes; they pack awkwardly, don't make sense to centralize that much if you design doesn't force you to, and you go from having a regulatory advantage of being able to build under city roads to a regulatory nightmare of asking to build 50 parallel lanes underground. It makes much more sense to build point-to-point connections out like a road network, and bring in more traditional mass transit where you have particularly busy connections that can't be handled with a reasonable number of lanes.

2350 is just the length of the cars one next to each other without factoring in any distance you obviously need to have between each car for safety.

Stopping distance at 40mph is apparently around 20m (without human reaction time)

With reaction time included is about double. So you need at least another extra 40m between each car.

There’s 499 gaps between those 500 cars so that gives you an extra 20km.

So your “train” of Model Y need roughly 23.5km worth of tunnel to move 2000 people.

Doesn’t look very efficient if you ask me.

Fascinating to think about if and when FSD is perfected (or heaven forbid before) what a world with cars moving about with near zero reaction time would look like. That is to say, completely removing reaction time out of the equation at scale

It would cut some space for sure but you still need space in between no matter what. You also need to factor in space because these cars have to stop, people have to move in and out of them and that takes time.

Yes I mentioned spacing them out, it doesn't change the point.

Let’s take the Elizabeth Line (Crossrail), where they run 24 trains per hour peak (every 2.4 minutes!) with up to 1500 people per train. That’s a theoretical maximum capacity of 36K passengers per hour.

Let’s say our four passenger Teslas can be dispatched every 10 seconds into the tunnel. That’s 360 vehicles per hour, four passengers per vehicle, or 1440 passengers per hour.

Up the pod size to 20 pax and you still only have 7,200 pax per hour. Halve the safety margin to 5 seconds and you still only have 14.4K per hour, 40% of the capacity of the conventional train. And filling and dispatching 20 people per pod every 5 seconds is a pretty tricky logistics problem!

So, increase the number of tunnels and remove bottlenecks where people are piling up by creating tunnels to where they actually want to go.

What would the London underground look like with 100x more tunnels and stations? A lot less busy in each tunnel probably. Plenty of room under London. The issue is the cost of tunneling and building these massively huge stations. Smaller could be a lot cheaper.

That would be ludicrously expensive - again, they had that idea at the Boring Company, and it hinged on finding cheaper ways of building tunnels (beyond just making them too small to fit emergency egress). As we've already seen, that never came to be, tunnelling is pretty optimised as it is.

With your 100 tunnels you could probably just reach the capacity of the Elizabeth Line (again, just one example) but at 75x the cost...

> But it's not a particularly hard problem. Don't crash into the thing in front of you

It turns out that this is a particularly hard problem.

Oh, and if you get it wrong, people will die. In some cases, dozens of people.

It’s a solved problem for real metros. Automated systems exist the world round.

My city has had and automated metro for nearly 40 years.

It’s rather easy with rail.

> It’s a solved problem for real metros.

This isn't a real metro.

> My city has had and automated metro for nearly 40 years.

Every year, in every country, there are failures that lead to fatalities. They are, thanks to technology, decreasing.

> It’s rather easy with rail.

It takes hundreds of engineers years to safely design and implement a train control system, and it branches out with complexity. Every year, cascading failures cause serious issues.

It sounds easy. It's not.

In 40 years there has not been a single failure of automation I am aware of leading to a death here in a city of millions. This is a busy metro not some tiny thing with a couple stops

40. Years.

So yea it is a solved problem from the perspective of living in an area with a large metro that has been fully automated for as long as I’ve been alive with very few failures and zero of these “serious issues” you think happen.

They were able to do it here with 40 year old tech, it should be easier and cheaper to do now

And it is easy when you are on rails in a controlled grade separate environment compared to automating cars/whatever Tesla is failing to do with cameras alone

And you say it’s not metro but people here are acting like the Tesla loop is “metro” and can compare/be better/cheaper then real metro by “just have multiple tunnels”

Hey, I also live in a large metro with a great safety record! Yes, because of some serious engineering. Tons of effort with continual upgrades and maintenance.

First of all, you seem to be assuming that I’m saying it’s not doable. It obviously is doable.

But you know, maybe go and read up some of the postmortems of train control systems failing.

And easy _compared to_ vision based systems? Sure! But that’s not even close to the same as “it’s easy”

> And you say it’s not metro but people here are acting like the Tesla loop is “metro” and can compare/be better/cheaper then real metro by “just have multiple tunnels”

I do not know what point you are trying to make here and I’m not sure you do either.

The system here still ran on 80s computers as of a few years ago and has had minimal upgrades.

Your missing the point: this was a solved problem 40 years ago and Tesla has decided to reinvent what works, for less capacity, with a far more difficult problem to solve and it’s somehow better?

You called it not metro, other people in this thread are calling it a metro and it “will be better then a metro

> Your missing the point: this was a solved problem 40 years ago and Tesla has decided to reinvent what works, for less capacity, with a far more difficult problem to solve and it’s somehow better?

I never said it was or wasn't unsolved. I also agree with that, and I do not understand how you could confuse what I've said as some sort of approval for Tesla. Like, I didn't say it was unsolved, just that it's a difficult problem. I didn't say their approach was good. And I sure didn't say it's better!

> You called it not metro, other people in this thread are calling it a metro and it “will be better then a metro

What? Okay? What's that got to do with me?

Like for real, you're just making stuff up in your head and getting mad at me over it, and I don't see a point of continuing this conversation if YOU'RE going to keep doing it.

But that’s the big safety advantage of pods over railcars. Only a small number of people can die in pods. /s

lol yeah although tbf American car fatality rates are pretty bad too

Listen to "Well There's your Problem" podcast on this. The maths don't math compared to literally any other form of mass underground transit.

> The premise of the Boring company is very simple: digging smaller tunnels at scale is going to have economies of scale associated that means you can have vastly more of them for the same money.

...so instead of 1 tunnel with a train in it, we could achieve the same throughput with 10 tunnels with single occupancy vehicles. Brilliant! Cost savings achieved! I'm sure the math will just somehow work out that 10 tunnels is cheaper than 1.

Economies of scale and learning effects are a pretty well studied phenomenon. So, yes, these are quite reasonable assumptions to make.

I'm more talking about replacing 1 massive tunnel between a handful of destinations with hundreds of tunnels going to many destinations. Why stop at 10?

I seriously doubt you can leverage economy of scale in something as labor intensive as building tunnels.

Why? They literally throw away machines that dig tunnels because they are built just for one tunnel. There's decades in between tunnels being dug for most big cities operating underground railways. A company that would be tunneling non stop could make a lot of improvements probably and get really good at it.

There is no basis whatsoever that learning effects don't apply to the business of tunneling.

Making improvements != economy of scale. Digging 10 tunnels in parallel using 10 machines and 10x the people I bet won't save you much in the grand scheme of things.

And the reason why they "throw away machines" is because most of the time those are custom built based on the project and the wear and tear is substantial.

The trick to cost saving: stack the tunnels! Exploit the 3D feature of below ground!

/s

You know, I was thinking about how that would actually work, and I just can't imagine it being anything remotely close to efficient.

For example, if you were driving on tunnel level 26 (you had to go down there because it's rush hour and the higher levels are jammed up), wouldn't it be insanely annoying to get back to a higher level? Do we have massive spiral ramps connecting each tunnel level? What about interchanges onto other directions, do you need to be on a particular level, or do you need to spiral your way up to Level 15 West Tunnel Expressway Interconnection Point to transfer to Northeast Extended Tunnel Expressway Business Loop 2A?

Why would every level have easy and uniform connections to every other level? They are transport connections, not destinations. And 26 levels? Why would you ever need that much capacity?

You need 26 stacked tunnels to meet the capacity of a subway line with 60 second interval, remember? (I don't think they've achieved intervals that low, but give it time)

And why would you ever stack them like a fence? Go ~2 deep, ~2 wide, and cover ~7x as many routes. Depth is more for avoiding intersections between cross-traffic than for lack of space.

What nightmare would that station look like and be to navigate

Loop has smaller but more numerous stations, because it can have smaller and more numerous stations. With trains every station along a route adds considerable delay for every passenger, exacerbated by the slow stopping speeds of rail. This is not so for pod-based transport where the stations are split off the main line. Stations are also cheaper because they can scale down to only a small subset of the though-traffic.

So.. roads. With taxies. You’ve reinvented roads without parking lots in taxies lol.

Except for all the significant differences, yes.

It’s roads in tunnels with small cars/vans going mostly point to point.

It’s just a separate road system with pods driven by a person or computer.

What are the significant differences?

Traditional surface roads in cities:

* Take surface space that can be better used for walkable densification, or are already taken by roads.

* Produce significant atmospheric and noise pollution, even from EVs.

* Conflict at cross-traffic, especially so for dense road networks like grids.

* Have to navigate around building geography, so for example can't support high speed turns.

* Conflict with pedestrians so are limited in speed and have to support frequent stopping points.

* Have no central routing and control for journeys so suffer unbounded traffic, even where additional vehicles lower capacity.

* Are hard to add to existing developments that need additional capacity.

* Don't have out-of-line stations for on- and off-boarding.

Add to this the differences you've already mentioned, like uniform taxies rather than mixed private cars, a comparative ease of automation, and being additive to existing network capacity rather than conflicting with it.

Most of that is just “tunnel vs surface” or “private vs public” ownership

It’s still just roads in tunnels with taxies.

Yes, I agreed already. Putting aside all the important differences it's the same thing.

Haha, you spent a lot more thought than me!

I don't know why you're getting downvoted.

I used to work for a PRT (podcar) company and for anything up to a large commuter train or major New York subway, compared seat-mile for seat-mile the networks of small vehicles are just as efficient as most urban transit. And then as a bonus you get no-transfer, direct point-to-point commutes. Even the loading and unloading is comparable between them for stadium event traffic.