When Code Is Nearly Free, Taste Becomes the Only Scarce Skill

The essay argues that code-generating AI has erased the old barrier of software: the hours of typing, manual-reading, and API-wrangling that used to stand between an idea and a working program. That distance has collapsed to almost nothing. But the author’s point is that the value you earned by clearing that barrier doesn’t disappear—it relocates. What used to matter was the ability to make something exist at all; what matters now is judgment about which of several plausible outputs is actually right.

Effort, the author contends, was never just an obstacle—it was a filter. Because building things was expensive, output was rationed and a quality floor was enforced; nobody shipped a thousand mediocre variants because each one cost real time. With generation now cheap, ‘good enough’ becomes a solvent that dissolves the incentive to do better, and the only remaining arbiter of what’s worth keeping is human taste. Borrowing from Pirsig’s notion of Quality, the piece frames taste as a fast, wordless verdict—recognizing the wrong version before you can explain why.

The uncomfortable mechanism is that taste is built through friction: you ship something bad, live with the failure, and file it away, over and over. Removing that friction lets newcomers become fluent without the apprenticeship that once produced judgment—productive by every measured metric, yet unable to tell whether what they can make is worth making. Worse, taste is economically invisible. It ships at the same speed as its absence, its prevented disasters leave no trace on any dashboard, and ‘slop’—output that is indifferent rather than wrong, passable and now effectively free—becomes the field’s most abundant substance.