Hi new friends, I'm Zeke. I live in Chicago and I love thinking about the future of computing in my spare time. In my day job, I'm a staff engineer at Abridge. I'm happy to find this space because I've been doing a lot of reflection alone or among friends and it seems like you are doing it together, here. I've recently been reworking a blog post about the cost of computing over time. It's still a work in progress, and maybe it will be forever, and I'd love to hear what you all think about it. I'm also doing some personal research projects (like a typescript / prompt IDE in tldraw!). I'd be stoked to talk to folks about that too. But, as a parent of young children, my time is limited so my work moves slowly!

Welcome! Love the comparison pictures between the first municipal computer and Raspberry Pi Zero!

On the subject of order-of-magnitude decreases in costs, Alan Kay’s discussion of “better old thing” vs. “brand new thing” (~27:00): And I can’t find what talk I’m thinking of but I’ve definitely seen a talk before that framed order of magnitude improvements in performance as new features/capabilities. Might’ve been Gary Bernhardt.

Surely, with AI, it is in fact getting much cheaper to write programs?

Yes, to write a certain kind of program for sure

I am pretty sure there are some deep economic forces that are at least a large part of why we get stuck with suboptimal programming languages, databases, etc. This is mostly intuition at this point. I should see if someone has dug into it.

Sounds plausible

I’m pretty sure that a lot of those forces go away with AI-assisted programming. My theory is that the long tail that will keep programmers busy for some time time to come is de-shittifying our industry. Programming will become cheap and efficient enough that it will pay to make good services that work well with good user interfaces.

28 years later and we're still on the pink plane

Love your hypothesis, that the quality of abstractions and the quality of UX is what matters for a computing environment. Can you think of cases where there was a good abstraction but the UX was so bad it destroyed it? Or the other way around? Do you think there are properties of an abstraction that invite a better/worse UX?

No amount of lipstick on the pig that is SQL will make it any prettier. Folks have spent a great deal of effort making all kinds of SQL clients but folks don’t use them much because they’d rather do data manipulation in anything else if they have any choice. This is the biggest single problem with the computer industry. We ought to be leaping to do our data manipulation using the Relational Model — that’s what it’s for. It’s the ideal paradigm for doing it. But you have to use SQL, which is so bad, it’s punishment for shoplifting in some countries.