I like the sensation of feeling at home whenever I use a system like Emacs or Guix, that is using Lisp. It is a sort of XML or YAML format, but with the additional power of macro, and a greater domain of applicability.
So I would create a minimalistic "shared-lisp" that one can use for configuring and customizing things.
Then every serious Lisp system like Common Lisp, Clojure, Racket, Guile, etc.. can easily convert this "shared-lisp" code into data and/or native Lisp code. And the rest of system will be implemented using some serious but opinionated Lisp environment.
So we should sell a simplified shared-lisp as a configuration language. An YAML on steroids. Then with time, Lisp can become common also on the "backend".
Additionaly: IMHO, the worst disservice done to Lisp, was the path towards not-so-much interactive Lisp environments. Common Lisp is a joy to use. Clojure and Guile acceptable, but not optimal. Racket not so much fun. We should empathize more the interactive aspect of programming.
@screwtape
I skimmed your links. To be fair, it seems one of the many/few academic projects, about AI and collaborating agents, and so on.
They are hard to evaluate, because many of them sound logical and well designed, but IMHO only usage in real-world scenario can tell the truth about their effectiveness. For example, there are many types of logical theories that can be used.
I'm usually interested to these topics. I have also some project to finish, on these topics. Maybe some day...
@mzan
I know it's a bit to ask, but have you had a look at Sandewall's draft book on knowledge representation viz a viz everything you just wrote. In particular Sandewall's Leordo/Leonardo framework https://www.ida.liu.se/ext/leordo/ . I think you and I (well, you and Sandewall) will be quite compatible. https://www.ida.liu.se/ext/aica/
Okay I'm going to stop spamming every reply on this thread x_x I want it printed as a book though.
@svetlyak40wt