Lines of code were a joke metric for 40 years. AI gave them a better publicist
Counting lines of code has been a punchline since the 1980s. Then "AI-native engineering" arrived and the discredited metric got rebranded as productivity — one CEO reportedly wants a million lines per engineer per month, and meant it.
The post skewers a survey of 219 engineering leaders that produced 219 definitions of "AI-native." An OpenAI blog post bragged twice that an internal tool hit a million lines — without once saying what the tool does.
An engineer who works on the JDK at Oracle delivered the quiet correction: the rebrand is already collapsing, because CEOs noticed they were paying for code by the line and the lines weren't turning into profit.
Goodhart's law gets a glow-up
The essay makes one argument and makes it well: lines of code, mocked as a productivity measure since the COBOL era, has been quietly relaunched under the banner of "AI-native engineering." Its exhibit A is a survey in which Augment asked 219 engineering leaders to define the term and got 219 different answers — a metric so vaporous it means whatever the speaker is selling.
The thread supplied the institutional version. A commenter recalled a Microsoft target of "1 million LoC per engineer per month" that read as satire until it turned out to be sincere. Another pointed at an OpenAI post on "harness engineering" that repeats a million-line figure twice in its opening — "no description of what the thing is, no indication of what value it provides its users."
The dog and the tail
The thread reached for the same law from different directions. One reply named it outright: "Goodhart's Law. The law says: 'When a measure becomes a target, it ceases to be a good measure.' The dog should wag its tail. But the tail is wagging the dog."
A commenter who'd read the management classics drew the arc as a curve: "Pre-AI: The goal is to make more money. With-AI: The goal is to ship more code. Post-AI: The goal is to make more money. Can't wait to see how we get there."
The load-bearing fact under all of it, stated plainly: "Writing code is no longer the bottleneck. Deciding what to build, reviewing code, and testing code are the new bottleneck."
The essay's own bait-and-switch
The most useful skepticism turned on the article itself. It criticizes the LoC obsession, then closes by urging everyone to use AI daily — and the thread noticed the seam. A reader called it out:
"Weird baseless push for AI on the end, with no reasoning, no goal, no claim of gain… an ad for AI after a short context pretending to criticize it."
Why the publicist may already be fired
The closing word came from someone who watches this from inside a language runtime. An engineer who works on OpenJDK at Oracle thinks the rebrand is self-correcting:
"This is already changing again now that CEOs have wised up to the fact that they're paying for code by the line but these lines don't translate to profit."
That's the whole life cycle in one sentence. A bad metric gets a new name, rides the hype for a few quarters, and dies again the moment someone upstream checks whether the extra lines moved revenue. They never do.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
Three chatbots played 21 nuclear wars. None of them ever chose to back down
A strategic-studies researcher sat Claude, GPT-5.2 and Gemini down as Cold War powers in crisis and let them bluff, lie and escalate. Across 21 games not one model, even while losing, chose accommodation. They reached for tactical nukes as casually as moving a pawn.