[dpark]

This is neat but also seems like an insane solution to the problem of “I don’t like seeing service trucks”. How many such tunnels and elevators would it take to supply the buildings in a typical city’s downtown area?

And what else could we do with that investment?

[m4rtink]

I would argue the ammount of space in cities wasted on cars and their infrastructure is totally insane.

[dpark]

And again, the conversation here is about delivery trucks, not cars.

Eliminating cars doesn’t eliminate the need for infrastructure for moving goods.

[hogrug]

> Eliminating cars doesn’t eliminate the need for infrastructure for moving goods.

Burying other last mile utilities that waste less land was not insane when real estate was a fraction as valuable as now and engineering technology was worse.

[dpark]

I wonder how many miles of fiber would have been laid if fiber required a tunnel big enough to drive a delivery truck through.

[hogrug]

You bring up an interesting point for the US to have a first world level of fiber to the home it needs to require diesel.

[dpark]

When I see bad faith arguments like this I earnestly worry that maybe sometimes I do the same and just don’t recognize it.

Did you read my comment about the cost of laying fiber being far different from the cost of digging truck-sized tunnels and make a conscious choice to pretend I was making a nonsense argument about diesel-powered fiber, or did you construct this strawman without realizing it?

[randunel]

I wonder how many people such an automated freight system would kill per year, compared to cars in the same cities. Once we have some numbers, you would probably reconsider the use of "insane" there.

[dpark]

The context here is not cars but delivery trucks.

The tunnels in question also did not transport pedestrians and were essentially focused on coal delivery and ash removal.

[randunel]

Sure, let's restrict statistics to any automobile used to transport merchandise. What do you think, ready to imagine such a comparison?

[dpark]

Sure. In 2022, 672 pedestrians were killed by large trucks. How many of those do you believe happened in say New York City? And how many of those do you think you could eliminate with this hypothetical tunnel system? And how much will this hypothetical tunnel system cost?

New York has about 300 pedestrian deaths total from all vehicles every year. So my guesstimate is that if you eliminated all of the trucks from New York City, you might save 50 lives a year, max. I would also guesstimate that it would probably cost well north of $50 billion to create this tunnel system to connect all of the major buildings in New York City. So we’re looking at about $1 billion per life saved. I bet you could save more than one life per billion if you put that money somewhere more useful.

> What do you think, ready to imagine such a comparison?

What are you intending to accomplish with your snarky and condescending tone?

[metalman]

it is not possible to eliminate the risk from the absolute requirement to move heavy bulk stuff, through and in citys. roads need work, big things break and fail ,wherever they are, new stuff gets built, again, precisly, wherever, it gets built. civil engineering is completly mature,and wildly boreing,and will dry your eyes out. much of the cannon is millenia old, with fuck all room to "inovate" and what works in one place, is a total fail somewhere else. what is more is that the ancient world is littered with the ruins of civilisations and citys, that did fail, and in every case part or all of that failure was to overextend, undermaintain, there infrastructure, or worse, jump to some new flashy thing that then fails, spectacularly. having walked through those ruins, and marvled at there engineering and planning of infrastructure, and also become a keen reader of all things civil or infrastructure engineerin, and also aircraft engineering, where the most important concept is up front, "failure mode", I have no respect for sudden ideas, and approve of what China has done to prohibit un educated comentary on infrastructure development and implimentation. "influencer engineering" by way of "saftey"

[MagnumOpus]

Nitpick: it is one life per year for the billion invested. Which after the typical metro infrastructure lifespan of a century is actually viable because the cost for a working-age US life in medical and other contexts is probably around the $10mn ballpark.

(Assuming financing happens cheaply by the federal state rather than via PPP grift; and assuming that $50bn is the number, which in NYC is an underestimate by a factor of at least five…)

[dpark]

That’s a really good point. This is all very back-of-the-envelope, but if the total cost per life saved were 10 million it gets into the ballpark of sane.

But as you noted this 50 billion is likely a major underestimate (for comparison the recently built SR-99 tunnel in Seattle cost 1 billion per linear mile and connects to approximately zero buildings via elevators). NYC covers 300 square miles and estimates are that there are upwards of a million buildings across 120k city blocks.

[xbmcuser]

they are no longer needed though with last mile delivery robots being introduced that can use the same elevator and stairs as humans

[dpark]

Have we already forgotten that a self-driving uber ran over a pedestrian a few years ago or that Tesla’s autopilot has caused multiple crashes?

I’m not optimistic that a bunch of robots sharing stairs with pedestrians is going to work out great.

[oblio]

Which robots?

[trimethylpurine]

Check out the recent moves on Atlas.

https://youtu.be/I44_zbEwz_w?si=sFS5XUhNtwEz_ebH

[oblio]

How much will those robots cost? How many will we be able to make within the next 30 years?

How will they be autonomous considering bipedal operation in random environments is MUCH, MUCH harder that full self driving for cars on public roads? And that's just moving around, we're talking about actual judgement to do a human job that requires reasoning and practical skills.

Jetson type robots are a pipe dream at this point. I don't expect to have a robot maid within my lifetime.

Let's be realistic and not plan society today around scifi fantasies, please.

We're probably lacking 80% of the basic science needed for autonomous robot maids.

[trimethylpurine]

Right. My comment wasn't about maids from the Jetsons. General purpose robots are not soon. But for more specific tasks we've come very far in the last decade.

Warehouse automation is a reality today. Package delivery is also, just not broadly across the US. But it is very much happening right now.

Specifically my comment was about package delivery, which appears to be around the corner for most major cities, and already in place in several major cities around the world.

For indoor delivery, you don't really need Atlas. A 4 wheeled "full self drive" can fairly easily navigate cubicles and press elevator buttons. It's really not that crazy, and doesn't require any reasoning whatsoever. Basic preprogrammed pathfinding borrowed from any modern video game works fine for this. I don't think you need any advanced AI, let alone AGI.

[oblio]

Going up random stairs is not the same thing, though.