Wish I had the time to take part in this. 11yrs later and still nobody has actually made the tools demoed in Inventing on Principle

That is because it is easy to build demos. Working tools that do the right thing in the general case are really hard.

This ^! Demo helps show the vision, but making this happen likely means re-thinking every part of the stack to support this principle of “show your guts!”

Only thing I'll add is: it's really easy to turn a demo into something that works for you! Unfortunately nobody believes me because they can't experience my things working for them 🥲

My hunch is that it's not that hard to make most of the demos "real" in the sense of being real implementations which share the limited scope from IoP

I meant it as, "something useful to me in my life."

It's true that not everything is easy. I'm still far away from reproducing inkandswitch.com/potluck for myself: merveilles.town/@akkartik/109293317013542368

But I'll get there. More generally, I think "working product for myself" is an excellent intermediate step between "demo" and "multitenant product". And going by 💬 #of-end-user-programming@2023-03-28T18:13:25.608Z, it's possible that's all we need.

This is a rather startling post, from end-to-end.

I liked this part:

Darklang's major advantage remains the same: the holistic or integrated nature of the platform, language, editor, and infrastructure. Darklang's major features: Deployless, Invisible Infrastructure, and Trace-Driven Development, were all difficult before Dark, because they occurred at the intersection points of the editor, language and infra. By removing that intersection in Darklang, these features were very obvious and fell out fairly naturally.

What are the intersection points in AI-generated code? Well, look at what Copilot is building: Copilot Chat in the IDE, Copilot for Pull Requests, Copilot for Docs, and Copilot-cli. Super exciting, but those seem like 4 different products, meaning a lot of intersection points with existing workflows, code bases and production deployments which we have the potential to remove.

What might an AI-first environment look like? DarkLang mentions "the holistic or integrated nature of the platform, language, editor, and infrastructure." This feels right, because AI that can instrument all of these pieces will be more effective and helpful.

The article also suggests that a strong, static type system might keep LLM hallucinations in check because they highlight logical inconsistencies.