it’s getting about time i should start playing around with some of these. they’re going to succeed at lowering the marginal cost of software development, but it’s hard to tell how that will actually play out. maybe something like jevons paradox where the overall demand for software developers increases, and/or we see widespread creation of limited-use, ephemeral software. applications that are created for a specific, immediate need and then quickly abandoned. i’m quite interested to see how this continues to mature.
The tech is exciting. The labor implications are not. Now how to get all of those disenfranchised tech workers away from Right Libertarian politics
ned ludd did nothing wrong. but that’s what i mean about jevons paradox, which is the observation that increasing the efficiency of using a resource can lead to increased aggregate demand. so if the marginal cost of software development goes down, there may paradoxically be an increase in aggregate market demand for software developers and tech workers.
I suspect jevons paradox is a very good analogy for this. My expectations is that most UIs will be focused around using chat prompts going forward. You just use natural language to tell the computer what you want and it’ll dynamically produce the UI you want on demand. Like you’d ask it to graph something, or look up some data, or play some music, or a combination of these things.
I can see software moving away from applications business logic with a UI towards simply making services you run on your machine, and then have a LLM on top of that orchestrating them and building visualizations for you using MCP on top of that.
Honestly, I’ve long wondered why this wasn’t the model already. Coupling the UI with the logic makes it very difficult to move data between apps or script things.
