Obstacles to a federated Internet

Seth Godin has a blog post today that draws a line between the slow modems and underpowered computers of the early days of networked communications and many of today’s Internet ills. In essence, his argument is that slow connections and de facto thin clients meant centralized servers and the desire for those running them to turn a profit. In contrast, he’s encouraging the concept of a federated Internet, where we take advantage of the incredible bandwidth and processing power we have at the edges of the network.

He doesn’t bring in this example, but I couldn’t help thinking about how Usenet—and even email early on—was a federated system where each host forwarded on to the next. I could be reached at ace@tidbits.uucp for years early on. That was a case of cooperation winning out over centralization.

Another trouble the federated Internet faced was peer-to-peer file sharing, which was demonized as a huge threat to content companies (which it was, to be sure, but that didn’t mean the underlying concept of peer-to-peer communications was itself problematic).

Anyway, just an interesting post that triggered a bunch of neurons for me as Twitter spirals into insanity and the federated Mastodon gains momentum.


Just thinking about the security implications of this kind of “federated peer-to-peer” architecture gives me shivers down the spine…

This is an issue near and dear to my heart, but I’m afraid that like capitalism itself, the barriers we have erected to this sort of emancipation will take quite a lot to dismantle. The biggest ones I see at present to getting more hosts on the open Internet are the IPv4 address scarcity problem (which can of course only be solved by end-to-end deployment of IPv6) and the strong collusion of the big tech, big telco and big government in maintaining a fundamentally producer-consumer extractive relationship. I really do hope that the spread of FTTP begins to shatter this, as it becomes clear that everybody can benefit from being a producer as well as a consumer, but that won’t happen until the monopolies that maintain the asymmetry of bandwidths are broken. We can hope.

See also:

I love nostalgia as much as the next guy, but you have to remember that prior to the time you could get a finally get a .uucp address that worked automatically you needed to know the “bang path”, i.e. you had to know how to manually route your message/files. That wasn’t that difficult because you only really needed to know the big site that your site forwarded to and the big site that your recipient connected to; they usually connected to each other. UUCP was never a “decentralized” network. Technically it could be, but in practice it was carried by big sites (largely AT&T as I recall, since they had “free” bandwidth). Usenet was never the main reason for UUCP.

I just checked and MacPorts has a uucp package, so in principle you can install uucp on your Mac and have some fun. I’d rather just use what we have now. Usenet was fun, but it’s long over.

It seems like the Internet was more federated when the connections were slower. You had USENET groups and little online forums and niche sites everywhere that have been eaten up by the big, centralized social networks.

The problem with any federated system (especially Mastodon) is complexity. Mastodon has all these different instances that link up in different ways, and some don’t link up at all.

Could you elaborate on what you mean by “different ways” and “don’t link up at all?”

My understanding is that Mastodon and the larger Fediverse link up through ActivityPub. By definition, ActivityPub is the standard communication for any of the various Fediverse instances. As a user of Micro.blog I can follow and comment with Mastodon users as well as Pixelfed users and so on. And a user of Mastodon or Pixelfed can follow my micro.blog account in return.

Although I basically agree with most of what you say, I don’t think that’s the “biggest barrier to everybody being a producer as well as a consumer”. I think the biggest barrier to that is that most people have neither the skills nor the desire to produce anything – they want to just consume. This is evidenced, for example, by the majority now using smartphones (a mostly consumption device) instead of real computers. Back in the day, in the early web or even on Usenet, we had a biased sample: the users were fewer, more technically oriented, and generally smarter than the typical person online today.

1 Like