Ken Iverson’s 1990 paper questioning APL’s direction, which served as the first public release of the J programming language.
Key Takeaways
The paper is the founding document of J, an APL-lineage array language that strips APL’s special character set while preserving its array-first computation model.
J’s design choices in the 1990 paper remain directly comparable to the current J vocabulary, making it a live reference for the language’s evolution.
The question mark in the title signals a design critique: Iverson was asking what APL should become, not just documenting what it was.
Canonical HTML version is hosted at jsoftware.com/papers/J1990.htm; the PDF source requires workarounds for Cloudflare access.
Hacker News Comment Review
Commenters note that APL’s failure to resurge alongside data science is surprising given its array-oriented model maps naturally to tensor/dataframe workloads that dominate modern ML pipelines.
One active builder is combining APL, Prolog, and Lustre lineage into a new language, using this paper as a recovered primary source, suggesting the 1990 ideas still seed original language design work.
The 1975 APL demo videos are cited as a canonical example of computing ideas that were decades early, reinforcing APL/J as underappreciated prior art for array and functional paradigms.
Notable Comments
@tern: actively designing an APL+Prolog+Lustre language; found this paper mid-build, treating it as a recovered design reference.
@quad62246: confirms this paper is J’s first public release; suggests comparing the 1990 language summary against the current NuVoc at code.jsoftware.com/wiki/NuVoc to see what held.
