The idea is a second elevator system, but instead of going up and down inside one building, it runs sideways between buildings, connecting every fourth floor to the fourth floor of the building next door, and so on across a whole block or eventually a whole city. You'd step off a normal vertical elevator onto a horizontal one and just glide over to the building across the street without ever touching the ground.

The obvious part is the rail. You'd need a bridge-like track bolted into two separate structures, which is already how a lot of skybridges work, so that's not the hard bit. The hard bit is that buildings don't sit still. They settle, they sway in wind, they expand and contract with temperature differently depending on their materials and how much sun they get. Two buildings ten meters apart can drift out of alignment by a few centimeters a year in opposite directions. A rigid rail spanning that gap would eventually tear itself apart or jam.

You could solve it with a telescoping joint, something that absorbs a few centimeters of slack, the way expansion joints work on bridges. But then the elevator car itself has to tolerate riding on a track that isn't perfectly level or straight at any given moment, which feels like a much harder control problem than a normal elevator, where the shaft is fixed and known.

There's also the question of who owns the airspace between two private buildings and who's liable if the joint fails mid-ride. I keep getting stuck there.

Has anyone worked with expansion joints on pedestrian bridges and can tell me whether a few centimeters of seasonal drift is actually manageable, or whether I'm underestimating this by an order of magnitude?