The future of programming (1982)
dl.acm.org
October 17, 2026
1 min read
45/100
Summary
Communications of the ACM’s “The future of programming” page includes a “Cited By” section listing works that cite it. The visible list includes N. Sophatsathit’s 2026 publication, “Supporting Traditional Manual Software Process With AIware,” in the International Journal of Service Science, Management, Engineering, and Technology. The listing identifies volume 17, issue 1, pages 1–27, DOI 10.4018/IJSSMET.416696, and an online publication date of September 22, 2026. The cited-by list also includes A. Laaksonen’s 2026 “Future of Programming,” a chapter in Guide to Using Generative AI in Programming. Its listing gives DOI 10.1007/978-3-032-07453-9_12, pages 177–180, and an online publication date of February 10, 2026. Another visible entry is G. Lomow and B. Unger’s “Distributed Software Prototyping And Simulation In Jade,” listed in INFOR: Information Systems and Operational Research. That entry gives volume 23, issue 1, pages 69–89, DOI 10.1080/03155986.1985.11731945, and an online publication date of May 25, 2016. The page provides a “Show More Cited By” option for additional citations.
What the discussion said
Commenters treated the 1982 essay less as a relic than as an unexpectedly sharp sketch of today’s AI-assisted programming debate. Several readers felt its near-term prediction landed: programming productivity has indeed been pushed by better tools, and modern code models such as Claude are the most consequential extension of that trend. Others noted that the essay goes beyond ordinary IDEs, anticipating programmers stating desired behavior declaratively through rules or specifications, a recognizable ancestor of prompt-driven and spec-driven development. The surprise was not merely that AI arrived, but its form. Readers argued that older science fiction could imagine giant automated systems or dangerous centralized machines, yet did not seriously anticipate broadly useful systems that help solve novel engineering problems. At the same time, the thread resisted declaring the old vision fulfilled. One view cast current AI programming as a revival of earlier fourth-generation-language and fifth-generation-computing ambitions that repeatedly failed to transform mainstream development. That skepticism was reinforced by the observation that no-code visual programming keeps resurfacing without durable adoption, while specification-first workflows still leave developers struggling to diagnose what went wrong. The resulting mood is cautiously impressed: the essay identified the direction, but today’s tools have not erased the hard parts of turning intent into reliable software.
Where opinion split
The central dispute is whether LLM-assisted programming realizes the long-promised shift from writing procedures to describing outcomes. Optimists see code models as the first practical tools that make high-level intent materially useful; skeptics see another cycle of 4GL, no-code, and specification-language promises, still unable to deliver dependable development at scale.
Community Sentiment
Positives
- Modern coding models make the essay’s productivity forecast feel prescient: AI now helps turn ordinary developer intent into working code rather than merely polishing the editor.
- Its vision of describing desired behavior instead of hand-authoring every procedure reads like an early blueprint for specification- and prompt-centered programming.
- Readers see today’s AI as more startling than classic fictional computers because it can assist with unfamiliar engineering work, not just automate routine operations.
Concerns
- No-code and visual-programming schemes have repeatedly promised to replace coding, then stalled in practice; commenters see that history as a warning against inflated AI programming claims.
- Spec-driven development still resembles the old debugging problem: expressing intent at a high level does not make failures easy to locate or correct.
- Much of programming remained keyboard-and-text-editor work for decades, suggesting that better abstractions routinely fail to displace established development habits.
