Very good point.
However I would argue with the “no one will use it until a lot of people use it” part. We should just reuse existing protocols, that even if not so popular nowadays are already adopted (e-mail, XMPP, RSS etc.) instead of reinventing the wheel (Matrix etc.). This way we would already some potential to start with that would bring us closer to the critical mass. We would just need to fill the gaps, e.g. one f the decentralized social networking protocols would need to became the de facto standard. And maybe move the web from HTTP(S) to something like IPFS.
There is however on issue we may want to rethink now: the power of corporations lays not only in the servers but also in slowly dissolving existing technologies. We had quite much already solved problems that became problems again just because G####e has intentionally broken that for profit:
We had unified multiplatform UI (GTK, QT, Java, even the HTML), but G####e has destroyed it designing A#####d in a way that destroys existing UI conventions and Ch###e that it breaks old websites on mobile (scaling down websites unless some is present – basically breaking the lex rerum non agit rule on the ground of webstandards).
We had broken Windows desktop OS monopoly with Linux only to have it back with A#####d / i## duopoly.
We had efficient network for idea discussion in the form of blogs interconnected with RSS, pingback and blogroll sections. As it was hosted on diverse platforms it was essentially uncensorable – but then F######k came and destroyed it by bringing everybody with network effect into their walled garden. And nowadays they abuse that power to empower radical left in US and right wing in eastern Europe (the former intentionally, the later unintentionally by walling fake profile creation in a way that blocks laypeople from doing so, but obviously is no limit for “russian trolls”).
IMO, we as the open-source developers should take some countermeasures now as previous generation did when they started the OSS movement and licenses like GNU-GPL appeared. Maybe explicitly stating in our licenses that rendering old apps with blur applied (hello MS and 150% scaling in Windows) or scaled in an unpredictable way constitutes alteration of original work and therefore copyright infringement, and therefore our open-source tools cannot be used for that? Or maybe development of tools that don’t do that would be enough?
In that time of history we may also want to think about preserving all the older works. I have books from 1920s that are obviously still readable, but if something was released in an electronic way in 1990s it is already unreadable. For that reason I think the future web should not only introduce ways to avoid article 13 censorship, but also ways of preserving all existing content. We already have web archives, we still need however to bring back Flash and Java-applet compatibility to our browsers instead of lying about technology progress – as the fact that we disabled some of the technologies that were widely used in the past is basically modern form of burning the libraries.