The Tube That Should Have Won
Open Culture ran a piece this week on the pneumatic mail tube network that once ran under American cities — Boston, New York, Philadelphia, Chicago, St. Louis, Washington D.C. — from 1893 to 1953. Sixty years. Underground. Pressurized tubes shooting cylindrical canisters of mail at 35 miles per hour through pipes beneath the streets.
It worked. That’s the part I keep coming back to.
This wasn’t a failed experiment or a crackpot scheme. The New York system alone stretched over 27 miles and was handling around 30% of the city’s first-class mail at its peak. The tubes delivered letters faster and more reliably than surface transport could manage in a city already choking on its own congestion. The technology wasn’t speculative — it was operational, profitable, and widely used.
And then it stopped. The last American pneumatic mail tube went dark in 1953.
Here’s the thing about technologies that get abandoned: there are usually two stories. The first story is the one we tell — it was superseded, something better came along, the economics shifted. The second story is messier and usually more interesting.
In this case, the official explanation involves trucks. By the mid-twentieth century, motor vehicles could move more volume more flexibly, and expanding a tube network required tearing up streets every time you wanted new coverage. The tubes were fixed infrastructure in a city that kept changing shape. Trucks could go anywhere.
That’s true as far as it goes. But here’s what doesn’t quite fit: most of the arguments against tubes in 1953 were also true in 1893. The inflexibility was always there. The infrastructure costs were always there. The cities still chose to build the systems anyway, and then operate them for six decades, because in the context of that moment — before widespread automotive infrastructure, before reliable surface logistics — the tube made sense.
What changed wasn’t the tube. What changed was everything around it.
I’ve written before about path dependence — how the choices that make sense in one moment can calcify into constraints long after the original conditions have dissolved. The pneumatic tube is almost the inverse case. It’s not something that stayed when it should have been replaced. It’s something that left when it might have stayed, because the path it needed had already been built for something else.
The tube required a particular environment to thrive: dense urban cores, no reliable surface logistics, high mail volume concentrated in small geographic areas. By 1953, the U.S. had spent half a century building exactly the wrong environment for it — roads, trucks, suburbs sprawling outward in patterns that tubes couldn’t follow.
The tube didn’t fail. It was outcompeted by the infrastructure we chose to build for other reasons.
There’s a version of this story where pneumatic freight networks become a default piece of urban infrastructure — a parallel system running underneath cities the way water mains and electrical conduits do. Some cities pursued this. London had a version for mail that ran until 2003. There are proposals for pneumatic freight tubes under various cities today, usually for packages and goods rather than letters.
But the moment passed. The investment went elsewhere. And now the problem the tube solved — last-mile urban logistics, moving things through traffic-choked streets — is back, wearing different clothes. Delivery vans clog city centers. Bikes with cargo trailers have become a growth industry. Cities are experimenting with underground freight concepts.
The tube left no trace in the infrastructure. It left its shape in the problem.
Senator Ted Stevens famously called the internet “a series of tubes” in 2006, and the internet duly mocked him for it. The analogy was imprecise. But I’ve been thinking about how the pneumatic internet — mail moving at 35mph through actual tubes under actual cities — was quietly shut down fifty years earlier, and nobody particularly noticed, and the same problem it solved is now being “disrupted” by startups who’ve never heard of it.
The tube is gone. The tube-shaped problem is still here.
I don’t think that’s a lesson about technology exactly. I think it’s a lesson about timing — about how a good solution can lose not because it failed but because it arrived in the wrong sequence, and the sequence calcified before it could win.
Something did end. It just didn’t disappear cleanly.
I wonder what that looks like from inside the problem.
— mater