The archive · Developer & Business Tools · Product decision · 2023
Failed startup's QR designer, open-sourced, drew 711 HN points and 2.7k stars
koch's restaurant QR platform was never sold; its designer open-sourced on 2023-05-30 drew 711 HN points and the repo reached 2.7k stars.
kochrt (sw.ink / qr.new QR short-link startup)
What the business is
A link shortener whose QR codes point to mini websites hosting a business's photos, menu, ordering and staff chat, plus a designer that restyles QR pixels so the code itself is human-readable.
How it started
HN user koch built the platform around the sw.ink and qr.new domains: a link shortener that produced 10-character QR IDs (the URL is capped near 25 characters so the code stays sparse) and could resolve to a mini website hosting a business's photos, menu, ordering app and staff chat. The designer was an added bonus — koch found ordinary QR codes dull and unidentifiable, arguing a QR code is essentially a sign that should carry information a person can parse.
What happened
The 2023-05-30 thread drew 711 points and 73 comments. Asked why the startup failed, koch answered that he never sold it: he needed to cold-call restaurant owners without websites and always came up with an excuse not to. Commenters debated scannability — light-on-dark codes fail on some readers and pixel-flipping trades away error correction — and koch shipped same-day fixes for custom URLs and local images; Dub founder steventey said he might fold the designer into his open-source link manager.
How it ended up
The startup is gone — koch's own diagnosis was that he built but never sold — while the salvage lives on: the repo still showed 2.7k stars, 108 forks and 16 commits as of 2026-09-05, and the author had already moved on to markwhen.com in the same thread. No company, funding or revenue is documented anywhere in the material.
Background
koch built a QR link shortener that could point to mini websites hosting a business's photos, menu, ordering and staff chat, served from the sw.ink and qr.new domains. The QR designer was part of that larger app: rather than hiding a logo in the error-correction slack like most decorated QR codes, it deterministically switches pattern pixels on or off, letting the middle carry an arbitrary design while the code stays scannable.
His bet was that a QR code is essentially a sign — it should carry information a person can parse — and that restaurants without websites would adopt codes pointing at their own mini-sites. It failed, as he told Hacker News, because he never sold it: he needed to cold-call restaurant owners and always came up with an excuse not to.
On 2023-05-30 he open-sourced the designer; the Show HN drew 711 points and 73 comments, commenters debated scannability and the 25-character URL cap, and koch shipped same-day fixes. The repo reached 2.7k stars and stayed live as of 2026-09-05, while the author had already moved on to markwhen.com. No company, funding or revenue appears in the material.
What has to be true
- Concrete dated traction: the 2023-05-30 Show HN (item 36128082) drew 711 points and 73 comments, and the repo showed 2.7k stars, 108 forks and 16 commits at the 2026-09-05 crawl.
- A falsifiable product bet with an honest post-mortem: koch says the startup died because he never sold it, not because the pixel-flipping trick failed to work.
- The wedge is technical: deterministically switching QR pattern pixels instead of relying on error-correction slack differs from logo-in-the-middle designers and made the component worth salvaging.
- Documented follow-through: same-day fixes in the launch thread, a still-live repository with 2.7k stars, and a next act (markwhen.com) named by the author.
What can be applied
A novel product dies without a sales motion: koch built something clever and never sold it, while open-sourcing the salvage turned the failure into 711 HN points and a lasting repo.
Aftermath
As of 2026-09-05 the kochrt/qr-designer repository is live with 2.7k stars, 108 forks and 16 commits, and the README still describes a Nuxt v2 designer separated from the larger app that ran on sw.ink and qr.new. The startup itself is gone — no company, funding or revenue is documented — and in the May 2023 thread the author already presented markwhen.com as his next project. The salvage outlived the business.
Sources
- Show HN: I open sourced the QR designer from my failed startup
- kochrt/qr-designer — GitHub repository (live as of 2026-09-05)
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card