Juniors learn. Some juniors are potential good seniors. Over time they will internalise good architecture and be able to make good judgments on their own.
Micromanaging LLMs is like having Dory from Finding Nemo as your colleague. You find ways to communicate, but there is no learning going on.
On my spare time, I am working on my game, a sci fi dungeon crawler – written in Haskell. Currently finishing up the graphics engine. OpenGL based. 2.5D but with a real 3D dynamic light system. In game objects and entities cast realistic soft shadows, and are lit up realistically. Took a lot of mathing and thinking to get working!
As a university professor, I still get random people e-mailing me “dumb questions” every once in a while. And I still try my best to answer them – in the spirit of keep the Internet alive!
Online community and connections are very valuable, and I also get genuinely interesting e-mails from random people. Usually someone who has read something I wrote, and want to discuss it. I also send out random e-mails, and my experience is that many people will answer, if you write to them about something they care about.
I did not realise there are ads for LLMs. Of course if I thought about it, I would conclude that there must be. But somehow the concept seems strange to me. What do they advertise? The LLM can do so many different things… I guess they are all “welcome! To the world of tomorrow!”
Then again, I know nothing about ads these days. Last time I had any ad exposure was twenty years ago, when I lived at home with my parents.
Dehumanising “the others” is a human trait, and a very destructive one. Just like violence and greed. People have different susceptibility for these, but we should all work to counter them and it is in its place to point it out when observed.
I owe so much to this blog! It was such an inspiring read when I started out programming in Haskell back in 2007.
Today I am a professor in computer science and still draw on it for examples in my advanced functional programming course. Just last week we did the loeb function, as an example of interesting use of Functor.
I write there about anything which interests me. My site is written in Markdown and an mkfile builds it using Pandoc. There is even an atom feed for my diary entries. Also generated using Pandoc!
I used Gentoo for ten years (2005–2015), and I was very happy with it! Stable was not the word I would use, in that updating frequently broke and required manual intervention. But it was so flexible! The easily accessible options one has for choosing everything about the system is unparalleled in any system I have tried since. I would still use it if I had more tinkering time. These days I am on NixOS, mostly to have the same setup on every machine I use.
I had not thought about this perspective (I wrote this essay many eons ago). But now I have a good colleague who is dyslexic, and he said the same thing. He says, for him, the colours carry more meaning than the characters often.
The way code is written in Haskell, small laser focused functions and clearly defined and mockable side effects, lends itself very well to property based testing.
This might not be impossible to achieve in other languages, but I haven’t seen it used as prevailently in other languages.
I have found that Haskell has two good things going for it when it comes to LLM code generation. Both have to do with correctness.
The expressive type system catches a lot of mistakes, and the fact that they are compile errors which can be fed right into the LLM again means that incorrect code is caught early.
The second is property based testing. With it I have had the LLM generate amazingly efficient, correct code, by iteratively making it more and more efficient – running quickcheck on each pass. The LLM is not super good at writing the tests, but if you add some yourself, you quickly root out any mistakes in the generated code.
I usually treat the LLM generated Haskell code as a first draft.
The power of Haskell in this case is the fearless refactoring the strong type system enables. So even if the code generated is not beautiful, it can sit there and do a job until the surrounding parts have taken shape, and then be refactored into something nice when I have a moment to spare.
Micromanaging LLMs is like having Dory from Finding Nemo as your colleague. You find ways to communicate, but there is no learning going on.