In 2011 I had been playing with a CR-48, the unbranded notebook Google sent out to test Chrome OS before any Chromebook was sold. Under its battery cover is a switch. Flip it and the machine boots in developer mode: verified boot stops enforcing Google’s signatures, a shell is a keystroke away, and you can install what you like. I had Arch Linux on it in about ten minutes.
That August I read Richard Stallman’s “Keep Control of Your Computing, So It Doesn’t Control You!”, first published in Der Spiegel Online. It granted that Chrome OS was free software and that users could in principle change it, but expected most Chrome OS devices, if Android phones were any guide, to be designed to prevent that.
I wrote to the address at the bottom of the page and said that had not been my experience. The switch was deliberately awkward to reach, so that a few minutes alone with the machine would not get anyone in, but as far as I knew every vendor shipping Chrome OS had to provide it. I called it a pretty decent deal for everyone.
A week later a reply arrived from Stallman himself, through gnu.org’s ticket system. He accepted that what I described allowed a free system, but doubted that vendors would bother with the switch unless Google required it, and that doubt was why the article said what it said.
Two weeks later I answered with the Chromium OS design documents: the developer mode and verified boot pages, recovery mode, and the per-device instructions. They said, in several places, that the switch was a requirement for any Chrome OS device; that developers had to be able to install their own builds or another operating system; and that if the design did not allow it, developers would hack holes in verified boot instead. They also described a second, permanent route: open the case, disable the firmware’s write protection, and flash your own. I sent him the passages and suggested the article could carry a note.
He answered three times over the following day. Some of what I had quoted struck him as more admirable than he had expected. He would add a note if practice turned out to differ from what he had written, but he preferred to wait and see how the devices were actually used. Reading further, he found the one thing that still concerned him: to run your own system, you signed the kernel with a utility Google provided, and he wanted to know whether that utility would be free software and available to everyone.
I admitted that I had first read about signing while writing my reply. As far as I could tell it existed to skip the warning screen and its delay at boot, and to let people redistribute images that would boot cleanly on other machines; with the switch flipped you could run what you liked without Google’s help. His last message drew the line where he wanted it: flipping a switch to load your own software was no hardship, and needing a nonfree program to sign your own kernel would be a real problem.
I had written on a whim, not expecting an answer, and I did not take it further. He had said he would wait and see; the sentence about Chrome OS devices is still there.