Mostly dead influential programming languages (2020)

Original Article ↗ Hacker News Discussion ↗

What does “mostly dead” mean?

  • Strong disagreement on the article’s use of “dead” / “mostly dead.”
  • Some argue “dead” should mean no practical use, only historical interest (e.g. ALGOL, Simula); by that standard COBOL, Pascal, PL/I, Smalltalk, BASIC, ML, etc. are clearly not dead.
  • Others define “dead” as “no meaningful number of new greenfield projects,” in which case COBOL and friends might qualify despite heavy legacy use.
  • Several comments call the article misleading or “unscientific” for equating “no longer mainstream” with “no longer in use” and say that damages its credibility.

COBOL: dead, alive, or undead?

  • COBOL is described as still handling a huge share of global financial transactions, with codebases that continue to grow.
  • Dispute over “no one’s starting new projects”:
    • One side says new COBOL projects are vanishingly rare compared to other languages.
    • Others with banking experience say new development around existing COBOL cores is constant, and COBOL remains “relevant whether we like it or not.”
  • Some marketing about “COBOL modernization in 2025” is called out as obvious AI-generated hype with fabricated claims.
  • Noted historical contributions:
    • Early, explicit emphasis on data portability across machines.
    • High‑precision fixed‑point decimal arithmetic, crucial for finance and supported in mainframe hardware (often via BCD encodings).

Other languages: how dead, how influential?

  • ML: commenters insist it didn’t “mostly die”; OCaml, SML, and F# are active, especially in theorem provers. Rust is seen as ML/Haskell‑inspired but constrained by its borrow checker and lack of higher‑kinded types.
  • Pascal/Delphi: still commercially used; Free Pascal/Lazarus highlighted as a strong cross‑platform option. Decline blamed largely on Borland’s strategy and pricing.
  • Smalltalk: still in use; discussion of its conceptual descendants (Ruby, actor-model systems) and contrast with Python’s less “pure” OO.
  • Lisp: far from dead—Common Lisp, Clojure, Scheme/Racket, Guile, etc.—with macros, REPL, and homoiconicity cited as still-unique strengths.
  • Algol: practically dead in new work, yet its ideas (block structure, types, structs/unions) were widely absorbed; some niche use remains (e.g. Unisys systems, JOVIAL).
  • Forth, Prolog, awk, Modula‑3, Eiffel, Ada, Perl, APL all appear in side discussions as candidates for “influential but niche/aging,” with mixed views on how alive each really is.

Broader themes

  • Many comments trace influence trees: APL → MATLAB → S/R → NumPy; COBOL‑like Englishy syntax → SQL; ML → OCaml/F#/Rust; Perl → PCRE and regex literals.
  • Several express nostalgia for learning and working in these languages, and curiosity about revisiting them with modern tools.