"I think Prolog ironically falls in the category as Forth (and Joy is not even as practical as Forth). It's incredibly expressive for some things, and it has ge

"I think Prolog ironically falls in the category as Forth (and Joy is not even as practical as Forth). It's incredibly expressive for some things, and it has generated many epiphanies. But if you try to use it in production, you eventually "fall off a cliff" for parts of the program that don't fit the model." I believe that's correct. It's why AI research usually used LISP with Prolog handling specific things it's good at. Often as a DSL, too. A good example which still sells is Allegro CL: https://franz.com/products/allegro-common-lisp/ https://franz.com/products/allegro-common-lisp/ On Prolog itself, there are commercial users outside AI/rules/NLP like this finance company: https://dtai.cs.kuleuven.be/CHR/files/Elston_SecuritEase.pdf https://dtai.cs.kuleuven.be/CHR/files/Elston_SecuritEase.pdf Far as efficiencies, most Prologs appear to be based on abstract machines that weren't quite optimized for real hardware. Whereas, Aquarius Prolog that Peter Van Roy posted here once seemed to show there's a lot of efficiency to be gained with optimized-for-metal designs: https://www.researchgate.net/publication/2813183_The_Making_of_the_Aquarius_Prolog_System https://www.researchgate.net/publication/2813183_The_Making_... So, that's just a few things I have. I never learned Prolog since I think it's too limited. There's quite a few systems that make tasks easy to express, with Prolog being one. I think ideal system would combine high-efficiency in base language, metaprogramming features (or term-rewriting logics), contracts (esp for automated analysis and testing), stuff like Prolog in DSL form, functional programming like Haskell/Mercury, and so on. All designed to interoperate well with C ecosystem and/or using assembly for optimized modules. General idea is you can always use the right tool for the job with base language integrating them smoothly in a way that maximizes productivity during development and performance during deployment.