The Internet has become a cluttered, proprietary mess

Font Size


By Elaine Ou

THE INTERNET was built on open standards and interoperability, but networks tend to balkanize.

If you’ve ever tried to receive an e-mail from me through Gmail, it probably went to spam. My company runs its own e-mail server, and Google feels that our mail authentication is inadequate. And remember those Flash-based websites from the aughts? They’re no longer supported by major browsers, and entirely blocked from Apple’s mobile devices.

Every now and then, we see attempts to re-open the internet. Last month, for example, Twitter announced Bluesky, a project to create an open protocol for social media. If adopted, Bluesky would allow anyone to spin up a Twitter clone and host a social network. Content could be shared across platforms — for example, a Twitter user could retweet a post from a Bluesky version of Facebook, allowing her followers to interact with the post without switching accounts.

There are plenty of other examples. One of the older ones is XMPP, an open protocol developed for social networking and chatting. The project began in 1999, and was adopted and later abandoned by AOL Instant Messenger, Facebook Chat, and Google Talk. (Once a platform gains significant market share, there are competitive advantages to using a proprietary protocol.) A more recent introduction comes from Mastodon, a decentralized social network as well as a software, in the form of ActivityPub.

You know where I’d really like to see open protocols? On the entire rest of the internet. Take Wikipedia, regarded as a definitive source of truth by fact-checkers on Facebook and YouTube. After criticizing Wikipedia’s oligarchic editing process, a former co-founder created the Knowledge Standards Foundation to promote an open protocol where anyone can host wiki-type encyclopedia pages. Custom reading interfaces can then fetch pages from sources preferred by the individual user.

Then there’s Google Search. Google’s prowess doesn’t come from hundreds of petabytes stored on index servers, although that does create a massive barrier to entry for competitors. The company’s dominance comes from its proprietary ranking algorithm, designed to deliver the worthiest links to the top of a user’s search results.

Google has been accused of de-ranking certain sites, promoting political biases, or even spreading disinformation. An open protocol for Google Search could allow anyone to set up a web server to receive queries and forward them to the index servers, then rank results using competing algorithms. The approach could be extended to any content platform.

The challenge, of course, is making a profit. It’s hard for platforms to monetize an open protocol — it’s like trying to make money off of e-mail. Jack Dorsey explains how an open protocol might be good for Twitter: “It will allow us to access and contribute to a much larger corpus of public conversation, focus our efforts on building open recommendation algorithms which promote healthy conversation, and will force us to be far more innovative than in the past.” No company truly wants to be forced to innovate — Jack was just going for a tricolon — but the first two points are valid. As a non-dominant social network, Twitter could benefit from accessing external content, just like Google gains access to new data each time a Gmail user corresponds with an external e-mail address.

Distributed platforms also give companies a way to avoid the messy problem of content moderation. Google contracts with over 10,000 search quality raters who spend their days evaluating controversial search results, while Facebook has hired 15,000 moderators to manage problematic content. With a federated internet, these companies would not have to employ so many moderators, or worry about creating a catch-all policy that attempts to please everybody — anyone who didn’t like one server’s content policies could simply start their own social network. No one complains to Mastodon about unsatisfactory moderation rules, because the host of a Mastodon server can implement whatever filters or amplification algorithms they want without losing access to external content.

This doesn’t necessarily mean that nasty content will take over the internet. When a controversial social network called Gab created a Mastodon instance, many other servers simply preemptively blocked it.

Rather than adhering to content rules imposed in a top-down manner, individual communities make their own rules and create an actual marketplace of ideas.

Requiring open protocols could even be part of an antitrust approach to increasing competition among big tech companies. Back in March, Facebook CEO Mark Zuckerberg suggested that regulators should guarantee the principle of data portability between services. If regulators go a step further and promote interoperability as well, it could be the modern-day equivalent of forcing Bell Labs to liberate their patents. Compulsory sharing of technology may not have been advantageous to the Bell System, but it certainly increased US innovation in the long run.