Jay had built an HN live feed viewer years ago, a real-time comment stream tool that let you watch Hacker News discussions unfold like a chat room. He'd built it because he loved the rhythm of HN—the way a discussion could start as a technical argument and end as a philosophical inquiry, the way strangers could teach each other things in public. He'd been reading HN for over a decade.
Now the factory was on the front page, and Jay was reading every comment.
It hit number four at 10 AM Pacific and climbed to number two by noon. The submission was the factory.strongdm.ai page itself, and the comment thread grew like a vine, branching in directions Jay couldn't have predicted.
Some commenters understood immediately. They saw the non-interactive development model, the separation of specification from implementation, the scenarios-as-validation approach, and they mapped it onto their own experience with AI coding tools. These commenters asked precise questions. How do you handle specification ambiguity? What's the feedback loop when satisfaction drops? How do scenarios interact with the Digital Twin Universe?
Some commenters didn't understand, and that was fine. They saw "no human-written code" and "no human code review" and reacted to those phrases as provocations, which, Jay acknowledged, they were. These commenters argued about job displacement, about accountability, about trust. Their arguments were sincere even when they were wrong about the specifics. Jay respected sincerity.
Some commenters understood something Jay hadn't fully articulated to himself. One thread explored the factory as an organizational innovation rather than a technical one. The tools weren't new—LLMs, containers, CI/CD, testing frameworks. What was new was the arrangement. The decision to make specifications the primary human artifact. The decision to make scenarios the primary validation mechanism. The decision to treat code as a generated commodity rather than a crafted product.
Navan was reading too, at his desk, occasionally reading a comment aloud. "Someone's asking about the token costs. They think a thousand dollars a day is a lot."
"It's less than a senior engineer's daily salary," Jay said.
"Someone else made that exact point. Three replies down."
Justin wasn't reading the thread. He was in a meeting. When he came back, Jay gave him a one-sentence summary: "Some understand, some don't, the ratio is better than I expected."
"What about the tone?" Justin asked.
"Curious. Mostly curious." Jay paused. "Some dismissive. A few hostile. But the hostile ones are arguing with a version of the factory that doesn't exist. They're arguing with 'AI replaces programmers,' and that's not what this is."
"It never is," Justin said.
Jay responded to exactly zero comments. He wanted to. There were misconceptions he could have corrected, questions he could have answered, compliments he could have acknowledged. But he knew from years on HN that the best response to a front-page discussion about your own work was silence. Let the work speak. Let the commenters argue. The thread would find its own equilibrium.
By evening, the submission had 347 comments. Jay had read every single one. He closed his HN live feed viewer, leaned back, and thought about the gap between what the factory was and what people imagined it to be. The gap was narrowing. That was enough.
Responding to zero comments is the correct play. Anyone who's been on HN long enough knows that engaging with the front-page discussion of your own project is a lose-lose. Let the work speak.