Hacker News Reader

Comment by calvinmorrison | Hacker News Reader

why write programming languages if LLMs are fast enough to talk plaintext everything to eachother with perfect memory

Why go see Rome when you can google up some photos? Why climb Mt Everest? Why have your own sex when there's so much porn? The answer is going to be the same it's always been: because we fucking can.

This is totally how I’m feeling right now.

This stuff has empowered me to finish the BS I don’t want to do rapidly then do other cool stuff I want to do.

This is true too! I have ADHD-like executive problems, and sometimes doing the boring but necessary stuff is so painful I abandon the project entirely. Now, though? I entrust it to the agent, and even if it's slop it's still done, and this actually gives me the impetus to start moving again.

I am not an ADHDer (though I often wonder if I'm on some spectrum or another), but this is one of the things I have found great success in. If I can yeet through most of the busy work associated with a project, and focus on the stuff that really matters?

TBH, if you're wondering if you're on a spectrum but never bothered to check, my advice is to check. I was always distrusting of labels, then got diagnosed with autism spectrum disorder at the ripe old age of 31. And holy mother of Czestochowa, the story of my life suddenly made so much sense.

I take your point, but unlike everything you mentioned, programming languages have usually been written for others to use, with a handful of Intercal-like exceptions.

Any fully new languages we get will be either doomed from the start, or the language equivalent of TempleOS. Or both.

I'm writing a new language and I know it's doomed. I expect exactly zero users beyond me in the age of AI. But I love languages, I love language design. So, I'm doing it because it's fun and rewarding. The joy I get from seeing my language written out and thinking "I came up with that, that's mine" can't be beaten.

Anyway I have one blog post about it. I'm working on the implementation now, just takes a while. I'm close to having a parser (not semantic analysis) and I'll probably do a dev log.

Well, that makes sense too. I don't believe evolution of programming languages is just going to halt here, but that's just like my opinion man.

At some point (soon), the massive inefficiency of this system will stop being subsidized. Also, it’s nondeterministic and unreliable.

Perfect* memory

*semi-perfect

Even if that were true: to save on inference costs.

Reply on news.ycombinator.com