You?

In situation (B), there's another problem I'd love to work on: software components. A very frequent situation in my work (as a computational scientist) is that I have all the building blocks for e.g. a data analysis but I can't just assemble them together because of technical details: dependency hell, incompatible languages, etc. A big issue that I see is the impossibility of building hierarchically modular component assemblies (see https://www.jstor.org/stable/985254). It's impossible because software toolchains heavily rely on globally unique resources: namespaces, HTTP port numbers, etc. It becomes impossible to combine two components that depend in incompatible versions of these global resources. That sounds like a solvable problem, but I don't see anyone working on it.

Assuming (B), I would explore/build two more of my language designs: Val and Nat.

Both Val and Nat are situated in an under-explored area of language design that is to the right of traditional PLT, assuming a spectrum (C→HL) from compilers/computers (C) to the left, and human language (HL) to the right.

While I’m not going to argue the case here, Val and Nat are based on a hunch of mine that HL’s are much more regular and mathematically beautiful than they are typically given credit for**; that under the surface there are lots of regularities/principles/features potentially useful in language design; the point being that those features have the empowering property of built-in support in the human language capacity (HLC). That’s a pretty intriguing property. Val is situated a bit to the right of traditional programming languages, in the middle of the C→HL scale and Nat further to the right. Both build on the ideas of deep syntax (a system of term structures with regularities that maps to a perceiver model). • Val exposes deep syntax as its syntactic top-level, much as Prolog uses it’s terms and LISP its s-expressions; • Nat maps deep syntax into tokenized structures reminiscent of word classes in human languages. Besides being located in an immensely fun and non-boring area of exploration; • the Val language might enable a beginning exploration of the impacts of deep syntax on dx of programming languages. • Nat might offer the language designer a hands-on way to peek under the hood of human languages, maybe finding features that can be borrowed etc. • Doing design work on Val and Nat might give opportunity to try out and extract principles for mapping out features of the HLC. --- ** Actually, under their irregular surface, I propose that HL’s are much more regular and mathematically clean than most current programming languages, if coverage is taken into account. Furthermore, I propose that language designers needs/perspective is well suited for appreciating/uncovering/exploiting such regularities.