It is true that, today, we have a vast array of impressive tools at our disposal, from parser combinators and generators, code generators to entire language workbenches with projectional editing capabilities. However, if one were to design a language for any reason, having a deep understanding of the expressiveness of computational models such as the lambda calculus would certainly be an "advantage" (especially for those who end up having to use the language): A Lisp/Scheme is as close to interactive lambda calculus as it gets. From there on, one can learn about implementing different evaluation strategies, scoping rules and continuations (William Byrd's "The Most Beautiful Program Ever Written" [2] also to mind).
Now, I am not sure whether homoiconicity tends to lead people astray. It is true that is it not strictly necessary to make a language extensible (e.g., Smalltalk has no macros and is very extensible due to its powerful meta-object protocol and an elegant syntax for closures), but it's still worth studying the concept. For example, writing a metaintpreter in Prolog [3] is surprisingly easy because of its homoiconicity.
[1] https://softwarepreservation.computerhistory.org/LISP/book/L...
The real thing with CL is that the entire language is available during parsing and macro expansion and that users can hook into these steps to influence them. No artificial limitations like you get in C++'s consteval, you can do everything and anything without having to use a crappy DSL to do it.
Also, I've never found SBCL slow to compile. Have you?
i had a phase when I was using PyPi a lot for branchy "old AI" kinds of workloads and felt it was an easy win but since then it has been either numpy or PIL or pytorch doing the heavy lifting or scripty stuff like uploading files to S3 where performance doesn't matter a lot.
I will grant that Common Lisp can be compiled to run amazingly quickly!
This is how https://github.com/marcoheisig/Petalisp#why-is-petalisp-writ... and https://github.com/numcl/specialized-function can exist.
Working with Clojure is an absolute delight. It strips down all the dogma and let's you deal with the "business logic" as if you're cooking steak using no BS ingredients - meat is meat, herbs are real, stove is hot.
Why would I ever choose bash for writing anything slightly more complex than simple redirection, when I can do things in way better fashion with babashka. Why would I wrestle a YAML CI pipeline that only fails on push, when I can drive the whole thing from a babashka task file, run each step locally in the REPL, and actually debug it?
Why would I ever deal with Lua, if I can't even format it for "readability" - no matter how I do it, it just looks darn ugly, and luafmt often makes it worse. Why, if I can just slash down dozen lines of Lua boilerplate compressing it into a three-liner Fennel macro? With Fennel, I can interactively poke through elements of my WM through Hammerspoon on Mac, and that's just bananas.
Why would I ever deal with JSON, when EDN is almost twice as compact and far more readable - I can align things and treat data as a literal table. Besides, I can group, sort, filter, slice, dice, salt & pepper that data easily, without ever leaving my trusted editor.
Why would I choose to build a web-scraper in Python, when I can use nbb driving Playwright and go through selectors interactively, directly from my editor, as if it is a devtools console. And I don't even have to restart anything, deal with state changes, etc.
How can I abandon Emacs where I can just open a scratch buffer, type some Elisp and change the behavior of my editor, my WM, my OS and even things on remote computers. No other text editing environment works the way Emacs does - nothing even comes close. It feels like playing a video game, where my controller in my editor.
Why would I write Flutter UIs in Dart, fighting the widget-tree ceremony and endless build() boilerplate, when ClojureDart lets me express the same tree as plain data and hot-reload it interactively? The layout is just nested maps and vectors.
Why would I reach for C when I need to embed a small, fast scripting layer. Text parsing alone would be a regex nightmare elsewhere.
Why would I bolt a templating engine onto HTML strings, when Hiccup makes markup just vectors - so my views compose, filter, and generate like any other data, no special templating DSL to learn
And with all sorts of different runtimes and dissimilar Lisp dialects, it still feels as if you're working with the same language. The mental overhead when switching is so negligible. While switching between just JS and TS - which are supposed to be of the "same family" - feels quite annoying. Despite the fact that I've put years into those - far longer than any Lisp I've ever used.
Sure, nothing special about Lisp at all. Except that practicing Lisp can actually make you a polyglot. You'd realize that it isn't syntax that makes a programming language, but runtime and semantics do. After years of dealing with different PLs, I lost a preference for one specific language - I'd choose the runtime best suitable for the task, and then see if I can bolt Lisp on top of it. And these days, it feels like there isn't a platform left where you can't meaningfully do things via Lisp.