michael-dean-k/

The Internet flattened the world into a digital puzzle—a kind of intelligence test to see who can combine the correct symbols in the correct corner of a social swarm; plucking a nerve to some various end, bringing fortune, fame, or cringe. Your conception of the game shapes how you play it. That’s the real limiting framework: your mental model. You can’t understand emerging social systems without flailing around to update those models. It is very easy to stay sidelined with some excuse that prevents you from getting embarrassed.

This same paradigm applies to AI. It’s limiting to anthropomorphize it; instead, think of it as a linguistic puzzle. It has ingested trillions of words and arranged meaning into a multi-dimensional geometric map, a massive mystery that no one yet has fully uncovered. It’s your job to craft and aim language into this hall of mirrors. We can only do this blindly. But nonetheless, it brings back a result, and how we respond to it is revealing. There all sorts of distortions—cliche prose, lying, hallucinations—and if you’re metaphor for AI is an entity, then you’ll naturally get annoyed or offended and stop playing the game. It’s better to think of your result not as a failure of the LLM, but a failure in your probe. You’ve asked a wrong or vague question. By sending multiple probes, you start to build a fuzzy map of the hyperobject you’re prompting into. Since it gives back feasible answers back so instantly, we expect it to be right for immediately, with little exploration, but that lack of patience is where the frustration comes. It’s a more a matter of rotation, to approach something that matters to you from many angles until you crack it.

I’m starting to feel the Opus 4.7 annoyance. Everyone has been complaining, and I told myself I’d be patient, but now I'm here watching Codex tutorials. 2 weeks ago I was able to effectively one-shot a Google Docs prototype in ~10 minutes with Opus 4.6. This sets the standard for what’s possible, and when that is ripped away, even 10% of it, it feels like theft, even when it’s still 2,000x faster than coding by hand. It’s easy to blame the model, but really AI coding has so many variables, and you can never really know the source of what shifted. Yes, it’s a new model, but also this time, I’m (a) deploying into an existing codebase instead of doing ground up; (b) the spec is far more detailed; (c) the whole factory has been redesigned. That’s four variables. It’s easy to not take the blame, put it on Opus, and then convert back to 4.6, but that itself is a change with unknown consequences. Was 4.6 nerfed too? The truth is we’re building systems on top of quicksand, but actually that’s not so novel because people are quicksandish too, always evolving, changing incentives, dreams, and abilities, totally variable day-to-day depending on if they slept or if they’re in a fight or not. We expect these machines to be deterministic (and use language like “factories”) but the cost of agency is a less determinism.