I’ve seen them around in London - they are incredibly slow. There’s no way they’ll succeed here - speed is important for getting around.

This is what happens when people who live in a bubble automatically assume what works there must scale and make economic sense to the rest of the world.

That and cyclists/pedestrians can be kind of anywhere at any time. Again not a quite the same in America.

We'll see I guess.

They're just really cautious and do need an extra 10s to think sometimes, and there's a lot of driving scenarios that don't allow that much time to decide to go or not. The biggest culprit seems to be unprotected intersections with the potential for human/bike crossing, which includes driveways which they insist on using for dropoffs (even when they're obviously not public driveways), which they then take a very long time to leave.

At the same time technology doesn't stand still and the algorithms now have an "aggressiveness" slider that goes up if the car can't make any progress.

I wonder how reliant self-driving cars are on unchanging road infrastructure.

    > half the interchanges weren't on Google Maps because they were built so recently

Here's the interchange near my hotel that's not on satellite but is on Google Maps. Unfortunately satellite coverage on Google Maps is older than Google Maps is, but you can see on streetview in April 2025 it mostly wasn't there and in May 2026 it mostly was. I think I visited a month or so after which is why it was so recent.

Sounds like someone who's never in their life driven in Austin.

It wouldn't surprise me to find it's nearly a half dozen per year, on a slow year. I only drive down there a few times a year and yet almost every time I go on my random route at least two stretches of road are just radically different from the last time I drove on it.

Sometimes it really feels like Texas' biggest industry is highway bridge/overpass construction.

If you've never experienced the rapid redevelopment of Texas exurbs and what not it's something of a marvel to behold, like watching a train wreck take place over months.

Like did a bunch of teenagers sneak out at night and build a highway?

How hard can it be to have a process where mapping companies can register with a state’s DoT and get information when they build stuff?

With how some of this urban design happens it almost feels to be the case at times.

A bit more real of an answer is: federalism. You're assuming the state DoT is entirely in charge of these projects. Lots of different government organizations are in charge of these things, and they don't always communicate well. Cities, counties, toll operators, and others can decide when these ramps are opened or closed. It's not just the state DoT that does these things. There's no single org that can authoritatively say "these are all the roads which are open, right now."

Now, I do agree with you, making such a data set complete and public should be some level of priority. There should be reporting standards involved in defined, standardized formats. These do not really exist, and making such a thing happen is a political question not a technical one.