June 22, 2026
Code war from the punch-card crypt
My 1992 view of the problems of computer programming in 1992
An old coding rant sparks a very modern fight over whether programmers have improved at all
TLDR: A rediscovered 1992 essay argues that better programming tools never solved the real problem: humans still struggle to clearly tell computers what to do. Commenters loved the writing but fought over whether student-made tools are truly “passable,” with others joking that today’s giant projects are still painfully slow to build.
A dusty 1992 essay just crashed into the present and, honestly, the comments are doing most of the yelling. The writer’s big point was simple enough for non-coders: making the tool that translates code into computer instructions got easier over time, but actually telling computers what humans want is still a mess. In other words, the machines improved, but the human chaos stayed right on schedule. That gloomy little mic drop got a warm reception from some readers, with one calling it "Beautifully written" and wondering if it was just a personal note to self because it felt that sharp and honest.
But this wouldn’t be internet discourse without a fight. The spiciest pushback came from people side-eyeing the essay’s claim that students could build a “passable” compiler in a semester. One commenter basically said: hold on, what counts as passable here? Yes, students can make school projects, but useful in the real world? That’s where the eye-roll starts. Another commenter dropped a deliciously nerdy history flex, noting that an old IBM compiler was once so fast it outran the speed of feeding in punch cards, while modern giant codebases can still take six or seven hours to compile. Ouch.
So the mood is a mix of admiration, nitpicking, and existential dread: people agree the essay still hits, but they’re also arguing over whether programming is still a black art or just a very expensive, very slow group project. Even the shortest comment — "what do you think of it?" — feels like someone tossing a match into dry grass.
Key Points
- •The article says IBM used FORTRAN extensively for internal programming in the mid-1970s and already had a capable compiler, FORTRAN G.
- •IBM is described as having spent millions of dollars and hundreds of programmer-years to create the FORTRAN H compiler.
- •The author states that compiler construction became much easier over the following fifteen years, to the point that undergraduates could build passable compilers in a semester.
- •The article argues that by 1992 compiler quality was no longer the main bottleneck in software development.
- •The author says the more fundamental problems were programming methods, language design, expressing intent, and managing software complexity.