The sharpest line in “Code has always been the easy part” isn’t really about AI at all—it’s about the perennial mistake of treating code as the primary artifact rather than the system it enables. The author opens with a concrete memory from Etsy: a team deep into a rewrite (in fact, “two distinct incompatible elegant architectures”) that had gone two years without shipping a customer-facing feature, until a pivot toward standardizing on PHP helped unblock progress.
Code fetish vs. system value
A throughline here is the industry’s long-running tension between obsessing over how code looks and acknowledging what actually creates value: a human-technology hybrid system that can deliver a product, adapt over time, and reflect the “spoken and unspoken needs” of a domain. That mismatch shows up in places like hiring, too—where the post calls out how technical evaluation can drift into ritual (including the now-mocked trope of asking candidates to reverse a linked list on a whiteboard).
The framing is blunt: code is the easy part, and has been for decades.
Why this AI moment still feels different
The post doesn’t dismiss the current wave of AI-assisted coding. It argues the last few months of frontier models—explicitly calling out Claude Code and similar tools—represent something genuinely new, primarily because the cost of producing code is collapsing toward near-zero.
But it also points out that “cheap code” has never been the core difficulty. What’s changing—again—are the social contracts of software teams, with one example being how review can feel fatiguing in a way creation does not. And while the tools are new, the pattern isn’t: the web, CI/CD, mobile, SPAs, and machine learning all forced teams to reinvent how work gets done.
The thread worth pulling
There’s a lot more texture in the original—skepticism about AI hype, a sense of temporal whiplash, and an interest in tiny-team “software factory” stories—but it all funnels back to the same claim: even if code gets dramatically cheaper, the hard part remains building organizations and processes that can ship, learn, and evolve.