All that is solid melts into code

It’s been wild to see the arc of AI bend so powerfully towards software development this year. I’m not sure that, in the summer of 2024, anybody was sitting around saying, wow, one of the chief problems facing the world today is the difficulty of producing lines of code. (It was not.) Yet language models (1) are naturally very good at operating inside this magic circle, and, more importantly, (2) can very effectively be trained to become even better.

The second point in particular makes it feel, in retrospect, inevitable: code matches formal(ish) verification, “answers you can check”, to broad application, “answers you care about”, in a way that might be unique. Let ’er rip!

Look at this fellow:

That’s an over-the-row olive harvester. Most olive oil production at medium-or-greater scale depends on machines of this kind; they trundle over trees planted in long rows, almost like continuous hedges, and collect the fruit with vibrating fingers. Machine-harvested olives (1) are cheaper to collect, and (2) arrive at the mill in better shape than olives harvested by hand.

One catch: most olives can’t be cultivated in this