The language made of itself: Lisp, still alive at sixty-seven
Invented in 1958 to do mathematics with symbols, Lisp gave programming garbage collection, the REPL, and the idea that code is data. Its descendants run editors, startups and spacecraft — and once, a NASA team fixed a Lisp program while it flew a probe a hundred million miles away.
Every language in this series has a claim on your attention; Lisp's is seniority with a twist. Conceived by John McCarthy at MIT in 1958, it is the second-oldest high-level language still in use — only Fortran, one year older, predates it.1 But Fortran survives as infrastructure; Lisp survives as an idea. The features it introduced — automatic memory management, the interactive prompt, functions as values, programs that can rewrite programs — have been absorbed so completely into Python, JavaScript and everything else that a modern programmer uses Lisp's inventions daily without writing a single parenthesis.
011958: a notation that became a language
McCarthy wasn't trying to build a product; he was trying to find a mathematical notation for computation over symbols — lists, expressions, ideas — rather than numbers, for the newly named field of artificial intelligence. His 1960 paper, Recursive Functions of Symbolic Expressions and Their Computation by Machine, defined the language in a page or two of itself: an eval function, written in Lisp, that interprets Lisp.2 By McCarthy's own account he considered it theory until his student Steve Russell hand-compiled that eval into machine code — and suddenly the notation was an interpreter, and the interpreter was a language.1
Two side effects of that origin changed everything. Because symbolic programs churn through memory unpredictably, McCarthy's design included automatic reclamation of unused memory — garbage collection was invented for Lisp,2 six decades before your phone's apps came to depend on it. And because the interpreter was right there, Lisp programmers worked conversationally, typing expressions and getting answers — the read-eval-print loop, or REPL, that every modern language now imitates in its console.
02Code is data: the idea that won't die
Lisp's syntax — those parentheses — is the feature people mock and the feature that matters. A Lisp program is written directly as the data structure (nested lists) that Lisp is best at manipulating. (+ 1 2) isn't just an expression; it's a list you can build, inspect, and transform with ordinary code. This property, homoiconicity, is what makes Lisp macros possible: functions that run at compile time and write code for you, letting the language grow new control structures as libraries rather than waiting years for a standards committee.3 Paul Graham's essay “Beating the Averages” — his account of building Viaweb, the startup that became Yahoo! Store, in Common Lisp — argues this was a genuine competitive weapon: his tiny team shipped features as fast as competitors could read about them.4
Other languages have programs that process data. Lisp noticed the program is data — and spent sixty years exploring what follows.
03The family tree
“Lisp” today names a family. Common Lisp, ANSI-standardised in 1994, is the industrial-strength dialect — a big, pragmatic language whose complete specification lives online as the HyperSpec,5 with SBCL as its workhorse open-source compiler producing genuinely fast native code.6 Scheme went the other way — a minimalist teaching-and-research dialect whose liveliest descendant is Racket, a “language for making languages.”7 Emacs Lisp is the extension language of GNU Emacs, meaning millions of hours of text editing run on Lisp to this day.8 And Clojure (2007) put a modernised Lisp on the JVM with immutable data structures at its core, giving the family its largest working population this century.9
04Lisp at work: a spacecraft and a grammar checker
The best Lisp war story is true. In 1999, NASA's Remote Agent — the first AI system to autonomously control a spacecraft — ran in Common Lisp aboard Deep Space 1. When a race condition that had never appeared in ground testing deadlocked the software in flight, the JPL team used the REPL running on the spacecraft, roughly a hundred million miles away, to diagnose and patch the live system; the software went on to win NASA's Software of the Year award.1011 Debugging a program you cannot restart, on hardware you cannot touch, is the interactive Lisp workflow taken to its logical extreme.
The workaday story matters too. Grammarly — the writing assistant with tens of millions of users — documented that its core grammar engine is Common Lisp, processing over a thousand sentences per second in production on ordinary AWS Linux boxes, compiled with SBCL.12 Not a museum piece: a hot path.
05Why it never conquered — and never died
So why isn't Lisp the mainstream? Partly history: Lisp was the house language of the 1980s expert-systems boom, and when that market collapsed into the AI winter, dedicated Lisp machines and vendors collapsed with it. Partly culture: a language this flexible lets every team build its own dialect, which is power for a small team (Graham's point4) and friction for a large one. But dying is different. The ideas kept winning even as the syntax didn't: garbage collection is universal; the REPL is universal; lambdas and closures — Lisp's functions-as-values — arrived in Java, C++, and every scripting language; and Lisp-style macros echo through Rust's macro system and JavaScript's code-transforming toolchains. McCarthy's own history of the language, written in 1979, already noted its knack for outliving its obituaries.1
06Where OcxlyDev lands
Of the three languages in this series, Lisp is the one we'd prescribe as a course of study rather than a tool: a few weekends with Racket or Clojure will teach you — viscerally, not theoretically — what evaluation is, why “code as data” keeps being reinvented, and where your daily language's lambdas came from. If you want production-flavoured Lisp, Clojure is the pragmatic door;9 if you want the classic experience, SBCL plus the HyperSpec is all there.56 Sixty-seven years on, the parentheses are still optional. The ideas, it turns out, never were.