“Code was never the hard part” is a lie programmers tell themselves when AI starts doing their job — and Senko’s post calling it out hit 117 points on Hacker News with 88 comments.
The pushback isn’t coming from AI skeptics. It’s coming from programmers who’ve watched agents generate code and realized the typing was doing more work than they admitted:
- Writing code forced incremental review. You saw dependencies, edge cases, and small decisions line by line. Agents collapse that into one review pass where you meet everything at once.
- The tools exist because coding was hard. Debuggers, syntax highlighting, intellisense, high-level languages, libraries — decades of tooling built to make the act of coding less painful.
- The phrase is self-inflicted. It’s not managers or execs saying coding was easy. It’s programmers themselves, rationalizing why AI replacing their typing shouldn’t feel threatening.
The Sanity blog frames it precisely: AI didn’t remove the hard part of engineering. It removed some of the breaks. The rhythm of typing code gave you time to internalize the system as you built it. Skip that rhythm and you’re reviewing logic you never built mental models for — catching bugs in code you don’t fully understand.
Martin Fowler splits code into two purposes: instructions to a machine (being commoditized by LLMs) and a medium for human reasoning (becoming more central). The “code was never the hard part” crowd conflates the two — assuming that because generating instructions got cheaper, the reasoning work wrapped inside that act was never real.
NOTE
The HN thread’s top observation: if coding was never the hard part, the entire history of developer tooling — from assembly languages to IDEs — was solving a non-problem. Nobody builds debuggers for something that’s easy.
The real risk isn’t that AI writes bad code. It’s that developers lose the incremental understanding that came from writing it themselves — and don’t notice until the review pass where they’re supposed to catch what the agent got wrong.