9/1 Video call with my son. I blundered through the usual, boring questions – how was school, who is your favorite teacher. The conversation was dragging until we got to the Roman telescope, at which point it turned into an excited, bubbling exchange of facts after facts.

It’s gonna take months until it gets to the Lagrange point!

It has a low-powered engine that gradually raises its orbit!

How many spacecraft are currently parked at the Lagrange point? (Turns out four.)

What does that even look like? How far apart are they from each other? Meters? Must be kilometers at least! (Turns out they are hundreds of thousands of kilometers from each other. They are circling the Lagrange point in what is called Halo or Lissajous orbits.)

The lives of the engineers tending to these spacecraft seem to be an entirely different world.


I had to build a small but very complex piece of software for a customer, in a programming language I don’t actually know. I asked the AI whether what I wanted to do was possible, but I was careful to point out, in the very first sentence, that I didn’t want it to actually build it just yet. All I wanted at this point was to discuss possible approaches to the project. Reluctantly, it seemed, the AI agreed to just talk, for now. I then answered some of its questions about details of the functionality, and this time I failed to mention that I didn’t want it to go ahead and just build it. And so it did. Five minutes later I had the finished result.

André Spiegel @drmirror