1978 talk by Professor Perlis at APL’78 frames APL as intellectually closer to French than to FORTRAN’s universal lingua franca.
Key Takeaways
FORTRAN is cast as the “air we breathe” of 1978 computing: on every machine, from every manufacturer, documented in every language.
APL is positioned as deliberately non-universal, more French in character: precise, expressive, and alien to those raised on imperative idioms.
The talk attacks the idea of adding while-loops, if-then-else, or for-statements to APL as a category error, not a missing feature.
The “spherical language” framing distinguishes languages that try to cover everything from those that commit to a single expressive axis.
Array languages ask you to think in shapes and rank, not in control flow; Perlis treats that as a feature, not a limitation.
Hacker News Comment Review
The strongest pushback targets Perlis’s dismissal of control flow as “ridiculous”; commenters with array-language experience see structured iteration as a real gap, not a category error.
Adampunk draws a productive line from APL to R: once you drop spherical-language ambitions, APL-likes develop their own coherent beauty, where complexity grows from the programmer, not the language itself.
The 1978 timestamp is load-bearing context; some predictions about FORTRAN’s permanence read as obviously dated, but the philosophical split between array-native and imperative thinking remains an active debate in J, BQN, and Uiua communities.
Notable Comments
@adampunk: “fractals of badness from us, the humans, as kernels” – sharpest summary of APL/R’s inversion of where language complexity lives.
