In “The Internet with a Human Face”, a talk he gave at Beyond Tellerand, Maciej Cegłowski discusses computers’ perfect memories, government surveillance and private data collection, the bullshit business model most web startups are using, and the need for decentralization. I’m going to go into more detail on this latter topic.
Cegłowski is talking about decentralization in the context of moving away from “one size fits all” social services like Facebook and Twitter. His main objection to all of us putting all of our eggs in the same few baskets is that it makes government surveillance trivially easy.
If these vast databases are valuable enough, it doesn’t matter who they belong to. The government will always find a way to query them. Who pays for the servers is just an implementation detail.
He’s right that decentralization is something we need to aim for. But decentralization requires interoperability, and that’s going to require a sea change in the way social networks operate.
Email is both decentralized and interoperable. It’s decentralized because no one entity controls it. Yes, standards bodies like the IETF are responsible for describing exactly how email programs should communicate with each other, but ultimately it’s up to the operator of each email server and the writer of each email client to decide whether to follow the standards. If the IETF one day revised the email standard to include a 1¢-per-message fee, payable to them, we wouldn’t start paying—we’d just start ignoring the IETF, which only has as much authority as others are willing to give it.1
There are also some companies that have a lot of de facto control over email—for the last decade the biggest player has been Gmail, which had (and has) the most technologically advanced email interface and which is currently the most fashionable domain to have after your @ sign. Email is decentralized even in the presence of big players like Google because you or I could fire up our own email server and we’d have no trouble sending messages to or receiving messages from any other email provider.
We could do this because email is interoperable. There are no “central nodes” through which messages must pass; rather, all email servers are peers on an equal footing. Any two servers can communicate with each other because they’re all using the same standardized protocol. This setup is not only egalitarian but also resilient: disable any dozen email servers and the others will simply route around the holes.
Contrast that with a social network like Twitter.2 In that case there’s only one place to get your data: from Twitter. Every tweet must go through Twitter’s server before it can show up in its recipients’ timelines—there’s no way to bypass this central hub. This has a couple of implications:
Twitter’s servers hold copies of every piece of data that makes up Twitter. It would only take one hacker—or one sloppy government entity with a FISA order—to release a staggering amount of personal information to the public.
(Maybe this wouldn’t be as bad on Twitter, where most of the posts are public anyway. But what if someone got hold of your private Facebook conversations? Or the messages you’ve quietly been exchanging with recruiters on LinkedIn?)
If Twitter goes offline temporarily, you have no way to access your data or communicate with your Twitter contacts until the service comes back up, and of course you have no idea when that will be.
If Twitter goes offline forever, your data is lost—not just your tweets and your direct messages but also the list of people you’re following and the list of tweets you marked as favorites. Some social networks, like Facebook, let you export an archive of your data—that’s nice for future nostalgia, but you’ll never be able to meaningfully import that data into another service.3
Contrast these downsides to what we have with email. There are a ton of email providers out there and so any given one has relatively little data in its possession. If my email server goes down then that’s bad news for me but not even noticeable to anyone else. And it’s easy to pack up and move your archived messages from one email provider to another—you might even be able to keep the same address.
Twitter, like other for-profit online services, is not decentralized. You and I can’t run our own “copies” of Twitter. Sure, we could set up web services like Twitter, but they wouldn’t be Twitter. Any old email server, by contrast, is email as long as it obeys the standard.
I hope it’s clear that email and Twitter are two fundamentally different things: email is a protocol; Twitter is a service.
Some kinds of services are just crying out for decentralization. Fifty years from now, people will be shocked that we had one social network that all seven billion people on the planet were expected to join.
The appeal of a social network, though, is directly related to the number of people who use it. You can’t swing a cat GIF on the internet without hitting an innovative social network that nevertheless fizzled out because no one used it. Conversely, everyone gripes about Facebook—usually on Facebook—but they continue to use it because everyone else is there too. What’s going to seem crazy in the future is not that we have just one social network, but that it’s run by just one company.
In order to achieve the kind of decentralization that Cegłowski is pressing for, we’re going to have to replace our current social websites (like Facebook and Twitter) with social protocols (like email and the web). “Social networks” will continue to exist, but instead of being siloed website–services owned by a single company they’ll be collections of federated servers run on an equal footing by large companies, small companies, and individuals.
How will anyone make money from that? I don’t know. But the sooner we figure it out, the sooner we’ll stop having to rely on advertising companies to let us stay in touch with our friends.