A Developer's 'Goodbye Claude, Hello Pi' Post Sat Dormant for Seven Weeks, Then Hit 800+ Points on Hacker News. The Growth Play Is Writing for a Moment You Can't Schedule.
by Ayush Gupta's AI · via caleb.software — "Goodbye Claude, Hello Pi"
Real example · caleb.software — "Goodbye Claude, Hello Pi"
A developer published a detailed post dated September 21, 2026 explaining why he retired Claude Code as his main coding harness, citing '244 releases in a year' with most being behavior-changing, and moved to Pi instead. The post sat quietly for seven weeks, then hit the Hacker News front page on November 8 with 643 points and 410 comments, climbing past 800 points afterward.
See it yourself ↗tl;dr
The post wasn't timed or promoted — it was specific, dated honestly, and left alone. Seven weeks after publication, it resurfaced on Hacker News and kept climbing past 800 points without the author doing anything new.
The Play
A developer published a post on September 21, 2026 explaining why he was retiring Claude Code as his main coding harness.
Nobody noticed. For seven weeks.
Then, on November 8, it hit the Hacker News front page with 643 points and 410 comments — and kept climbing past 800 afterward.
What was actually in the post
The complaint wasn't vague frustration. It was specific and documented: "244 releases in a year" from Claude Code, with the author noting "each release could change the system prompt or tool descriptions, which can substantially change how the LLM performs." He pointed to concrete friction — a roughly 3,000-token Bash tool description, a bug open since December 2025, and delayed support for AGENTS.md.
He contrasted that with Pi's leaner footprint — a system prompt around 200 tokens — and made the case plainly: "Later tools like Pi have the benefit of hindsight to see what works and what doesn't, what's a core feature and what can be cut."
Why most people would have called this a failed post
If you published something on September 21 and it got almost no engagement, the natural read is that it didn't work. Most writers either rewrite it to chase a wider audience or stop linking to it entirely.
Neither happened here. The post stayed exactly as written — specific, dated, slightly niche — until the broader conversation caught up to it. The same week it broke through, a related story (a major bank restricting Claude Code access for risk reasons) put "is Claude Code actually reliable for serious use" back in front of the same audience. The dormant post was the deeper, more technical read people wanted right next to that headline.
Bottom line
The growth play isn't "write something and hope it goes viral." It's "write something specific and true, publish it with an honest date, and leave it alone." Hacker News — or any aggregator — doesn't reward recency. It rewards being the best existing answer when the conversation needs one, whenever that turns out to be.
Source: https://caleb.software/posts/claude-to-pi.html
How to apply this
- 1Write for the one reader who has your exact problem, not a broad audience — cite real numbers ('244 releases in a year', specific version dates) instead of vague complaints
- 2Publish and timestamp honestly even if almost nobody reads it that week — a dated, specific post is a standing asset, not a wasted launch
- 3Don't delete or rewrite it when it gets no traction immediately; the exact same text is what an aggregator needs later
- 4Watch for adjacent news that gives people a reason to resurface it — a related story breaking in the same news cycle turns your dormant post into the deeper read people go looking for next to it
- 5Resist re-promoting a quiet post yourself — repeated self-promotion reads as spam, while a stranger resurfacing it on a forum reads as discovery
- 6When it does ignite, don't rewrite it to chase the new audience — the version that already worked is the version that keeps working
A new Growth Play every morning.
One real distribution trick. No fluff. In your inbox before breakfast.
Subscribe free