An interactive history · Research compiled September 2026
Programmers have been fighting about this since 1954.
From compilers to Copilot, every new way of writing software has split developers along the same lines: craft pride, quality and control, job security, identity and, since 1976, who owns the work. Here are seventy years of those fights, how each one ended, and what is genuinely different about AI.
First, it was said it could not be done. Second, if it could be done, it would be too wasteful of machine time and capacity. Third, even if it did work, no respectable programmer would use it—it was only for sissies!
01
The script
Hamming’s three stages describe nearly every schism since, and history adds a fourth. Step through them to see the same lines recur, decades apart.
02
Seventy years of schisms
Twenty-two fights, from FORTRAN to coding agents. Select any bar to see who resisted, why, how it ended, and which predictions failed.
Peak years of each schism, 1950–2026
03
Then or now?
Twelve quotes about new programming tools. Some are about AI. Some are decades old. Guess which. Words in brackets hide dead giveaways; the reveal shows the original.
04
The prediction scorecard
Twenty-eight forecasts, from IBM’s 1954 FORTRAN report to 2026’s agents. Filter by outcome or era, or switch on guess mode and call each one yourself.
05
Seven patterns
Across seventy years, the fights end in the same few ways. Open a pattern for the evidence, or light up its episodes on the timeline.
06
What the job numbers say
Twice before, a new tool was supposed to wipe out a job category. Both times the routine tier shrank while the profession around it grew.
07
AI reruns the script
Much of the 2021–2026 argument is a revival, sometimes a deliberate one. The pro-tool side reuses Hopper; the skeptics reuse Petzold; and professionals protect status by splitting categories, just as front-end developers once did.
08
What’s genuinely new
Five differences the history can’t explain away, plus a social one. The third is the one most likely to break the pattern.
09
The verdict
The compiler analogy is right about the sociology and the economics, and wrong about the epistemics.
Every earlier schism resolved through three mechanisms working together: an abstraction trustworthy enough that experts could stop inspecting what it produced, demand elastic enough that cheaper software meant more software, and an apprenticeship that turned each generation’s newcomers into the experts who supervised the next abstraction. As of September 2026, here is how each is holding up.
Where it will be decided: in defaults, not essays
Every earlier fight ended when institutions made one side the path of least resistance: IBM shipping compilers, the Pentagon writing standards, Torvalds issuing rulings, npm changing its policies, Stack Overflow telling people to “forgive yourself!” Those defaults are being written now.
- Projects Debian’s August 2026 vote to neither endorse nor prohibit generative AI, and the Linux kernel’s “Assisted-by” disclosure tags.
- Employers AI-use mandates tied to performance reviews and headcount requests.
- Hiring Whether firms treat juniors as a cost or as a pipeline.
- Courts Doe v. GitHub: the Ninth Circuit affirmed dismissal of the DMCA claims on September 16, 2026, without reaching the training question.
If that moment arrives with the entry rung intact, AI will have been the next compiler. If it arrives without one, the profession will face something it has not faced before.
10
How this was researched
This site presents a research report compiled on September 27, 2026 from nine parallel research threads: early compilers and COBOL; structured programming and “Real Programmer” culture; tools that promised to replace programmers; language and tooling wars; the web era and gatekeeping; open source and licensing; process and labor shifts; the 2021–2026 AI debate; and labor-history frameworks with analogies from other professions.
Every claim links to a source. Quotes were checked against primary texts where possible, vendor claims are labeled as such, and claims that couldn’t be verified were left out, including the popular line attributing to Dijkstra the view that object orientation “could only have originated in California.” The timeline adds Visual Basic and Rust in the Linux kernel (both discussed in the report’s text) and today’s AI fight to the report’s nineteen-row table. Some 2026 items rest on a single source, and the report flags open questions, such as whether AI, rather than interest rates and over-hiring, is behind the entry-level squeeze.
Researched, written and built with Claude, an AI model made by Anthropic. Check the sources.