The Internet’s Past Was Distributed. Will Its Future Be?

Given the recent chaos at Twitter following Elon Musk’s long and successful quest to buy it, I decided to set up an alternative micro-content account. With, I’ve taken my first steps into the Fediverse, a world of services that use the same protocols but have many nodes run by multiple independent operators. In a way, it’s a bet on the future of the Internet looking rather like the distant past of the Internet.

I’ve been on the Internet since the early 1990s, when I dialed into a pilot program run by St. John’s University in New York. At first, I mostly used e-mail and text-based web access. At that time, users and Internet service providers (ISPs) were expected to be good citizens. An ISP’s subscription fee would include an e-mail account, access to Usenet, a little personal web hosting space, and perhaps other services as well. (Free versions of all of these also existed, but especially before Google came around, people were expected to depend on their ISP; users of free services were treated with suspicion.) ISPs would also mirror popular free software as a public service, helping to share the load and expense of file downloads at a time when both storage and bandwidth were expensive commodities. An ISP was also expected to protect its reputation by responding to reports of user misbehavior. Someone who misused shared resources or who made life miserable for other Internet users might be reported to their ISP, and their ISP would be expected to kick them out. The ideas of a person being banned from an ISP and of an ISP being cut off from networked services were severe punishments. In rare cases, a user or provider could receive a “death penalty,” to be decided by consensus of server administrators, which would cut them off from a major distributed system such as Usenet. The threat of a death penalty usually caused an ISP to take action against its worst users.

Early protocols of the Internet were designed to be distributed, relying on ISPs as hubs to route messages. The Simple Mail Transport Protocol (SMTP), for example, allowed systems to route mail through multiple relays, and it also allowed for mail to be held or rejected in case the destination was temporarily or permanently unavailable. Internet Relay Chat (IRC) relied on network nodes to keep messages moving even if one node’s connection went down, although it was also prone to “netsplits,” in which many messages or users would suddenly drop off the screen for no apparent reason. Usenet, a distributed bulletin board system, relied on ISPs to provide access to newsgroups; messages were relayed around the world, and, aside from the “death penalty,” Usenet wasn’t censored in any large-scale way. All of these protocols relied on the goodwill of network node operators to keep data flowing. There was no money to be made from the protocols themselves; they were just considered basic amenities of an ISP’s subscription fee, necessary to retain loyal customers and to keep distributed services useful for everyone.

Commercialism was not kind to these open, decentralized protocols. Without any kind of access fee to send a message, e-mail, IRC, and Usenet became havens for unsolicited commercial messages. Early efforts to shame commercial message senders by reporting them to their ISPs ultimately proved unsuccessful. Businesses, including my employer, learned that they could send e-mail very cheaply, and customers appreciated the convenience of doing business online. ISPs made more money by selling services to commercial clients. “Bulletproof” service providers charged even more than good-citizen ISPs did, with the expectation that, for example, a bulletproof web host would host content that a traditional ISP would not. That led to the development of now-ubiquitous software to keep unsolicited commerce away from users. Using e-mail without a spam filter is incredibly frustrating. All major web browsers block pop-up ads, and other ads are easily blocked as well. IRC and Usenet didn’t have widely-accepted filtering systems to block commercial bot mail, putting the burden on individual users to maintain ignore lists and “killfiles” respectively. In the end, most distributed chat and discussion systems were overrun by bot accounts, ending their appeal as mass media. Both IRC and Usenet still exist, and are still used by niche communities, but the masses have chosen centralized, controlled alternatives such as Slack and Discord for chat and Reddit for discussions.

As a software developer for nearly 20 years, I embrace the notion that “all good developers are lazy.” If someone else has already done the work, and can take care of the operations, and charges you little or nothing, the lazy thing to do is to delegate that work. I could run my own e-mail server, but that would be a big pain for little benefit, so I subscribe to a commercial service. I could set up a co-located (or “colo”) server in a data center to host this web site, but colos are expensive and require me to do maintenance, so I pay a web service provider to host everything for me. I published a book earlier this year, but rather than printing it and shipping it myself, I hired professionals to take care of all the work I’m too lazy to do myself. Almost all the great developers I know entrust their personal e-mail, photo storage, web hosting, and social networking to third parties, many of which are free, even if they could build something better with enough time and energy.

That brings me to Wes, a friend of a friend who recently started Mastodon Transit Authority, a small social network hosted on his colo server. Wes runs Mastodon, an open source social networking system that can talk to other Mastodon servers, called peers, over the Internet. Mastodon Transit Authority is small and will probably remain small. In small communities, like the Slack and Discord instances where I’ve been spending a lot of time lately, the norms of good behavior are a lot easier to enforce than they are on platforms with hundreds of millions of users. Just because a Mastodon node can interact with its peers does not mean that it must. It’s up to Wes to decide which Mastodon nodes he will and will not accept as peers, and it’s also up to him to decide when to censor Mastodon Transit Authority’s content or ban a user for bad behavior. Wes also has to keep his server running reliably, keeping up with software patches and making sure that the Fediverse doesn’t overwhelm his single colo server.

Mastodon isn’t going to solve the “echo chamber effect” common on social media; Wes’s node attracts people who largely agree with each other, and he’s free not to peer with nodes that we don’t like. Some large Mastodon nodes, like Gab and Truth Social, are known as hotbeds of politically-charged hate speech and conspiracy theories, and so I’m not expecting Wes to peer with them. This is fine with me; I know where to find contrarian opinions by well-informed writers, and I would rather read contrarian material as articles or books anyway, where there’s enough room to make a strong argument.

My big curiosity with Mastodon, if it continues its rapid growth, is whether it will end up looking like 1990s e-mail, when every provider was expected to run its own relay, or whether it will look more like 2020s e-mail, when a small number of providers dominate the market. The network effect matters; I don’t foresee people setting up their own Mastodon node when their friends are already on a popular server with its own administrators.

Could social media become something worth paying for? Aside from LinkedIn Premium and dating sites, there’s little precedent. (There are also “web3” services that use cryptocurrency as a revenue stream, but I’m ignoring them, because I don’t think people will want to pay every time they post something.) WhatsApp used to charge $1 per year, with the promise that it would never be in the ad business, although the founders sold for $19 billion and reneged on that promise. I still think that a paid, ad-free social medium can survive. Admins can run a lightweight Mastodon instance very cheaply and either absorb all the cost or charge users a pittance. In any event, I’m glad to see communities of all sorts realizing that they don’t need a monolithic corporation, controlled by one person, to oversee all their communications.

I currently work for AWS, a division of Amazon. I previously worked for Salesforce, the corporate parent of Slack. I hold shares of Amazon and Salesforce stock. The opinions in this article are strictly my own and do not represent those of either company or any of its subsidiaries.