Back in February, Jeremy Allison gave a barn-burning speech at the Copyleftconf 2020, entitled “Copyleft and the Cloud.”
Allison starts by drawing the crucical distinction between “open source” (you can see the inner workings of the code) and “software freedom” (you can exercise technological self-determination), and explores the many ways that the former has eclipsed the latter.
From “tivoization” (where a vendor uses DRM to prevent users from modifying the code on the products they own) to moving everything to the cloud, where the underlying source code can’t be modified except by the cloud’s owners.
He describes how “open source” was a technocratic proposition, concerned with giving hackers technological self-determination while leaving users behind to take whatever they’re given - and how the failure of software licensing takes away self-determination even for hackers.
It reminds me powerfully of Mako Hill’s absolutely crucial 2018 Libreplanet keynote on the way that corporations have figured out how to use open source to hoard all the software freedom, while taking it away from the rest of us.
Allison excoriates software freedom orgs - like FSF and The Software Freedom Conservancy - for their focus on licenses, saying that licenses only really work for business-to-business negotiations, and are all but useless to individuals who lack wherewithal to sue big companies.
Instead, Allison calls for a focus on protocol documentation, saying that in a cloud-based era, real software freedom comes from being able to make compatible clients for existing servers, and compatible servers for existing clients.
I’m not entirely convinced; I think protocol documentation is incredibly imporant and agree with the analysis of the limitations of licenses and the rapacious hoarding of software freedom through DRM and cloud computing.
Protocol documentation will do something to address these, but not enough. There’s a legal side to this, and while Allison explicitly says that he’s more interested in engineering approaches than legal ones, there are limits to the engineering-only approach.
The reason that companies are able to resist license enforcement, and the reason that their enclosure of software commons is so effective, is that tech has become monopolized by a handful of firms, and they attained that monopoly through anticompetitive acts.
The traditional antitrust world did not permit firms to attain dominance through mergers with major competitors, catch-and-kill buyouts of nascent startups, or vertical monopolies where companies that owned platforms competed with the companies that used them.
These rules were heavily nerfed by Reagan, then further eroded by every administration since. Now, we have the an internet made of five giant services filled with screenshots of the other four.
The reason that companies adopted software freedom even before open source came along was their terror of competitors who might take away their customers by offering more freedom to them. Today, that terror has been eliminated, thanks to monopolization.
Facebook is losing millions of users every year…to Instagram.
The incredible profits created by monopolies allow Big Tech firms to create new legal weapons - new laws and new interpretations of existing law - that allow them to punish people who make interoperable products without permission.
This legal power to block Adversarial Interoperability is one of the critical ways that Big Tech maintains its monopolies. I think Allison’s analysis of the practical limitations of licenses is spot on.
But interop isn’t just a matter of documentation, there’s a crucial legal dimension to it as well.