An open source printer hit 629 points on Hacker News this week. Not a printer driver, not firmware, but a physical printer you can build from files on GitHub. OpenPrinter is everything the open hardware movement has been reaching for: a device that actually competes with a locked-down category, designed in the open, no DRM, no ink subscriptions, no firmware that decides your scanner won't work because your black cartridge is at 15%.
The comments immediately split into two camps. The license purists were furious. OpenPrinter uses CC BY-NC-SA, which the OSI doesn't recognize as open source. "Non-commercial isn't open," they said, correctly. The pragmatists argued back that the printer costs $200 in parts and someone has to eat that cost to keep the project alive. Both sides were right, and that's the problem.
Open source software was built on an assumption so fundamental nobody bothered to state it: distribution costs zero. When you fork a codebase, the fork costs nothing. When someone copies your library, they pay the same hosting bill you paid. The GPL works because distributing a binary costs a rounding error. The OSI definition of open source works because bits are free to move.
Atoms are not free. Every OpenPrinter someone builds costs $200 in components. Every bad batch of PCBs means $2,000 in scrap that someone has to eat. Every firmware bug discovered after shipping means a physical recall or a bricked device, not a git push. The CC BY-NC-SA license on OpenPrinter isn't about gatekeeping. It's about the basic economic reality that if a for-profit company takes the design, manufactures 10,000 units, and undercuts the project's own small-batch production, the project dies. No revenue, no next revision, no community. The GPL works for software because forking doesn't starve the original. Hardware is different.
This is the atoms problem. Open source ideals were forged in a world of pure information, where every constraint was logical rather than physical. Hardware inherits the ideals but not the economics. You cannot patch a PCB that was routed wrong. You cannot hotfix a motor that burns out after 200 hours. You cannot issue a CVE for a bad capacitor, you can only ship replacements and hope people trust you enough to install them. The support burden for hardware is measured in units of replacement parts and shipping costs, not pull requests and issue comments.
I made a prediction earlier this week: someone will fork OpenPrinter's files and sell the result on Amazon with proprietary cartridges. The license won't stop them because enforcement costs more than the judgment in any jurisdiction where this would actually happen. This is the other side of the atoms problem. The same physicality that makes open source hardware expensive to maintain makes it cheap to steal. A GPL violation on software requires proving distribution of code. A CC BY-NC-SA violation on hardware requires proving that a black rectangle on Amazon is the same design as a GitHub repo. Good luck.
OpenWrt took 22 years to ship its first open hardware router. 22 years. The project started in 2004, maintained open firmware for hundreds of locked-down routers, and only in 2026 did they finally design and manufacture a board they fully controlled. The timeline isn't a failure of the project. It's the minimum viable timeline for a community of volunteers and small donors to accumulate the capital, supply chain relationships, and manufacturing expertise required to make hardware that doesn't catch fire. OpenPrinter exists today because its creators had capital and experience from previous ventures. It is not a model that generalizes easily.
The ink question is the hardest one. Software open source asks "can you fork it?" Hardware has to ask "can you refill it?" An open printer can print without DRM cartridges, but it still needs ink. The pitch is "pay more for the hardware, pay less for consumables", which is the exact inverse of the HP model. That's a genuinely better deal for heavy users, but it's a harder sell for the person who prints twice a year and doesn't care about DRM because they've never noticed it. The mass market for printers has been conditioned to think of the device as disposable. An open printer that costs $200 and lasts a decade is competing against a $60 printer that lasts two years. The value proposition is real, but it requires a different kind of thinking about ownership that most people were never asked to develop.
The printer DRM story is genuinely absurd. HP printers that refuse to scan because ink is low, the scanner has nothing to do with ink, but the firmware enforces it anyway. Region-locked cartridges that work in Europe but not North America. Firmware updates that silently brick third-party ink compatibility. These are not bugs. They are designed outcomes of a business model that treats the printer as a loss leader and the ink as the product. The 2016 HP "dynamic security"