This is really interesting. I have felt for the past couple years that the moldability of Common Lisp lends agents the appropriate affordances to “do the right thing” via experimentations.

I see there is a section on RLMs; have you ran Autolith via agentic benchmarks? I would love to see comparisons with Prime Agent.

We have not done that yet, it's one of the major priorities. Autolith implements RLM a bit differently than Prime Agent (although it both comes from the same paper):

In Autolith, the top level agent is traditional, but has RLM tools which it can use for the things RLM is good at, namely exploratory work, processing a lot of files at once, backward context research and so on.

Forth is unbelievably underrated and I think its value is locked in poor DX.

Hey Alex, we actually had a very similar idea this morning! I could not find your contact anywhere, would love to chat. This is really cool.

Sure! www.calendly.com/growth-kinetics

Maybe you can use Mercurial for this? You don’t need the complexity of Git for memory.

Could you share the code for some of the custom threading macros you created? Looks very useful.

will do when I share with you the streaming string library I talked about on your discord. I use these arrows in every single namespace I write so this is already part of the plan.

Sometimes, the repetition of encounters is just as meaningful as the people we meet.

https://en.wikipedia.org/wiki/Small-world_network https://www.quantamagazine.org/new-proof-shows-that-expander... https://telecom-paris.hal.science/hal-03814119/document

Tacit programming + llm control library (like LangChain) has a lot of untapped potential. Tacit programming shines when the control structures are simple (and functions/variables can be anonymized), and these patterns occur frequently in programming on top of llms.

For those who have the book, what is contained in the "discussion of the interactive Lisp programming process" as mentioned by Paolo?

The discussion basically says you don't follow an edit the full program-compile-run cycle like in other languages with batch compilers. Instead, you interactively create a program by writing individual expressions (e.g. function definitions) in the editor, sending them to the REPL for evaluation and testing, further experimenting at the REPL, and repeating the cycle. If you're already familiar with Lisp environments there's nothing new in the book.

Forth really changed the way I think. I now take notes in a Forth-like way. Very underexplored.

Could you say more about what it means to "take notes in a Forth-like way"?

Not sure what it means

carrot soup cookies bread butter cart grocery todo

would be one way :-)

I'd love to see your notes as a sample, please :).

I think he might mean:

1. Push items onto a stack.

2. Create new stacks (words) for categories of items.

FWIW, I do the same, and use a tool (now using almost exclusively the GUI tool I include in the github repo) to do it.

This is the explanation of the tree-of-stacks method that I am using: https://github.com/lelanthran/frame/blob/master/docs/FrameIn...

(yes yes, it's a slideshow, but until I can make a video, this is a better medium for visualising how it works).

In my experience, perfectionism is crafted post-procrastination. It probably makes no sense, but the mind often does this convoluted thing in order to protect procrastination.

No, but there should be. If interested in collaborating (I made ChatGPT.el), shoot me an email at [email-redacted].

[url-redacted].

Interesting project. I can't help but ask if you've ever considered how Richard Stallman would feel about people configuring Emacs to upload code into OpenAI's cloud. It amazes me even more that you get people to pay to do it. I'd rather see Stanford helping us put developers back in control of their own AI experiences.

I think it's important to notice that free software is free to use and extend by people who don't necessarily share philosophical or political convictions of the software's author.

I was talking about showing empathy. I'm not sure where you got those ideas.

https://github.com/karthink/gptel might interest you as well

I was one of the reasons llama.cpp instructions was recently added.

The ordering is because of date added I'm pretty sure.

However I bet an issue about ordering based upon freedom respectfulness would be well received.