Post

We should stop thinking about software as anything other than getting the computer to do what we want it to do for us

44 reposts367 likes

weirdly though i think the malleable software people were onto something and that that means thinking a lot more about software, really carefully, and figuring out what will actually serve us towards that end. the llm's definitely do a lot but whats underneath imho matters.

Also getting the computer to not do what we don't want it to do for us, although maybe that's necessarily consequential

Do the dishes and the laundry, clean the apartment, make dinner? And "say it" can be/is the hard part. I've used Claude to build a few very, very small utilities but have zero idea how I would automate, say, the workflow that ends with one of those utilities.

The trouble is the correctness and determinism in "what we want the computer to do" is far more important if what we want it to do is, say, keep an airliner in the sky, or operate a medical imaging device within safe limits, or keep track of people's finances, compared to e.g. make the next Wordle.

well, video games are an art form. in like an abstract sense you're trying to get the computer to do what you want, but in practice it's much less "shuffle these bits from A to B" and more "this code needs to induce emotion in another person" which is a very different way of making software

I think you have to separate the code from the art? The code in a game that's moving pixels around is doing some complex geometry, or is executing a response to something that happened or that you did. The art is in what some writer wrote, or some artists drew.

Like, at least 90% of what you download isn't code, but is literally art. The code is what displays or enacts that art, and is extremely generic; it doesn't resemble that art at all.

I mean I guess. This feels sort of like saying that both Van Gogh and people who make road signage put red paint onto a canvas, and in a concrete sense that's true but those two things are so different I don't think collapsing them is conceptually helpful.

i think this is overly reductive, like the quine relay is technically getting the computer to do what you want but that isn’t /really/ what it’s about.

I was like the person you first responded to: good thought by Paul, but is there really any other way to think about software? But I really like your answer.

People make software. If we devalue how software works it devalues the people making the software, transitively. When you say the first one, leadership hears the second one. These two statements are tightly related. Your flippancy may secure your bag. But it's corrosive to products and people.

if you think I'm about getting my bag rather than empowering people to use computers better then you really dont know my deal

And then there's how future bots feel about this attitude. Might not be good! Can imagine them making the second statement about us. Maybe eventually.

relevant

maxwell ⎔@mfzx.net

to me the dream of computing was always to have a computer that could react to and fulfill the endlessly arbitrary and constantly-changing needs/desires of the user, and it seems like we're actually getting there