3 ms·

At this stage of a project, claims like that should be regarded as pure speculation. Elon Musk is not known for rigorously meeting deadlines. But even assuming

At this stage of a project, claims like that should be regarded as pure speculation. Elon Musk is not known for rigorously meeting deadlines. But even assuming that all the claims are accurate, what good does it do to build it more quickly if you have to kill >95% of the project value?

erikpukinskis 8y ago
No, smaller tunnels are objectively cheaper.
- michaelt 8y ago
  Smaller diameter tunnels mean lower cost of boring, certainly. But some of the costs of satisfying external stakeholders, like buying land; getting approval from regulators, politicians and the local community; and so on would be independent of tunnel size. IMHO the main reason not to install a conventional metro line is Musk wants to demo something going 150mph, and I've never seen a metro line going that fast.
  - jcranmer 8y ago
    > IMHO the main reason not to install a conventional metro line is Musk wants to demo something going 150mph, and I've never seen a metro line going that fast. It may seem counterintuitive, but higher speeds mean less throughput. Most metro systems are concentrated on trying to move high numbers of people, so bumping train max speeds to 150mph just makes no sense. The sweet spot for speed appears to be around 50-70mph, which most metro trains will do when they're cruising between stations. But anyways, anyone who is complaining that local mass transit ought to be going 150mph is someone who hasn't looked at why local mass transit can't go 150mph or is someone who doesn't want the "mass" in "mass transit".
JoshCole 8y ago
Taking the concession as a given as you requested: It's like the difference between running everything on one very beefy server versus having a horizontally scaling architecture on commodity hardware. Each have pros and cons, so its not as simple as supposing that just because one is less beefy than the other, the total value of running on a supercomputer versus a cluster is >95% value for the supercomputer relative to the cluster. If you have the goal of being able to scale to an arbitrary amount of traffic, building one very fast and expensive computer doesn't meet the traffic needs as well as building a lot of computers. A lot of things start leading to this conclusion. Economies of scale which come in when doing tons of a particular thing rather than just doing it once. Benefits of automation starting to factor in heavily for repeated projects rather than one off projects. Amortized costs of resources for repeated projects rather than one off projects. Another way you can approach understanding the differences in benefit of this project compared to a subway is by thinking about above ground transportation. Why are roads good, since we already have highways? Why are lifts at ski resorts good, when we have airports? What benefit is there to a boat, when we have cargo ships? Why have cars, when we have trains? All of these questions are intended to point toward there being some value in a cheaper solution relative to a much more expensive but less practically extensive solution. Yet another way you can understand the potential benefit relative to other transport costs is to imagine trying to optimize sign up for a site. If you reduce the number of fields in a sign-up form with the goal of making it easier to register, it is alike to reducing the cost of adding additional transport. In one case, the probability of a register increases. In the other case, the probability of a project which increases transport capacity increases. This analogy is imperfect, but I'm trying to give a analogy that technology people will understand through their experience: drive down cost, increase usage. Finally and I think this is one of the key things that needs to be understood about the proof of concept: its intended to be a first and not a last. The idea in driving down costs in tunneling is that there is an arbitrary amount of untapped transport capacity going unused beneath us. Making use of the tiniest fraction of it will only have a tiny benefit, much like making use of a single commodity computer will only have a tiny benefit. Making use of a hundred thousand tiny fractions of it will have a much more pronounced impact, much like making use of thousands of computers can have substantial benefits. You need to keep in mind the end goal, because a journey isn't its first step: one of the superhuman evaluations of a position supposes that the value of a position is the value of the expectation of where you end up, not where you are
SideburnsOfDoom 8y ago
> At this stage of a project, claims like that should be regarded as pure speculation. Sure. At some point it's worthwhile to actually built it in order to test the projections. And it looks like that point is now.

erikpukinskis 8y ago

No, smaller tunnels are objectively cheaper.

michaelt 8y ago
Smaller diameter tunnels mean lower cost of boring, certainly. But some of the costs of satisfying external stakeholders, like buying land; getting approval from regulators, politicians and the local community; and so on would be independent of tunnel size. IMHO the main reason not to install a conventional metro line is Musk wants to demo something going 150mph, and I've never seen a metro line going that fast.
- jcranmer 8y ago
  > IMHO the main reason not to install a conventional metro line is Musk wants to demo something going 150mph, and I've never seen a metro line going that fast. It may seem counterintuitive, but higher speeds mean less throughput. Most metro systems are concentrated on trying to move high numbers of people, so bumping train max speeds to 150mph just makes no sense. The sweet spot for speed appears to be around 50-70mph, which most metro trains will do when they're cruising between stations. But anyways, anyone who is complaining that local mass transit ought to be going 150mph is someone who hasn't looked at why local mass transit can't go 150mph or is someone who doesn't want the "mass" in "mass transit".

JoshCole 8y ago

Taking the concession as a given as you requested: It's like the difference between running everything on one very beefy server versus having a horizontally scaling architecture on commodity hardware. Each have pros and cons, so its not as simple as supposing that just because one is less beefy than the other, the total value of running on a supercomputer versus a cluster is >95% value for the supercomputer relative to the cluster. If you have the goal of being able to scale to an arbitrary amount of traffic, building one very fast and expensive computer doesn't meet the traffic needs as well as building a lot of computers. A lot of things start leading to this conclusion. Economies of scale which come in when doing tons of a particular thing rather than just doing it once. Benefits of automation starting to factor in heavily for repeated projects rather than one off projects. Amortized costs of resources for repeated projects rather than one off projects. Another way you can approach understanding the differences in benefit of this project compared to a subway is by thinking about above ground transportation. Why are roads good, since we already have highways? Why are lifts at ski resorts good, when we have airports? What benefit is there to a boat, when we have cargo ships? Why have cars, when we have trains? All of these questions are intended to point toward there being some value in a cheaper solution relative to a much more expensive but less practically extensive solution. Yet another way you can understand the potential benefit relative to other transport costs is to imagine trying to optimize sign up for a site. If you reduce the number of fields in a sign-up form with the goal of making it easier to register, it is alike to reducing the cost of adding additional transport. In one case, the probability of a register increases. In the other case, the probability of a project which increases transport capacity increases. This analogy is imperfect, but I'm trying to give a analogy that technology people will understand through their experience: drive down cost, increase usage. Finally and I think this is one of the key things that needs to be understood about the proof of concept: its intended to be a first and not a last. The idea in driving down costs in tunneling is that there is an arbitrary amount of untapped transport capacity going unused beneath us. Making use of the tiniest fraction of it will only have a tiny benefit, much like making use of a single commodity computer will only have a tiny benefit. Making use of a hundred thousand tiny fractions of it will have a much more pronounced impact, much like making use of thousands of computers can have substantial benefits. You need to keep in mind the end goal, because a journey isn't its first step: one of the superhuman evaluations of a position supposes that the value of a position is the value of the expectation of where you end up, not where you are

SideburnsOfDoom 8y ago

> At this stage of a project, claims like that should be regarded as pure speculation. Sure. At some point it's worthwhile to actually built it in order to test the projections. And it looks like that point is now.