Two Takes on the Eternal September

11 minute read

September, and with it the start of classes at most educational institutions in the United States, is almost upon us. In honor of this, I’d like to look back this week on the September that started twenty-eight years ago in 1993 and is still continuing: Usenet’s Eternal September. Twenty-eight years in, I still don’t think we understand exactly what the implications of this change were, nor have we decided what direction Internet communities should turn in response. I don’t have all the answers either, but I’d like to suggest two useful ways of thinking about it.

The story

For those who are too young or weren’t using the Internet at the time – a category which includes me – here’s the Internet mythology to which I’m referring. Usenet was a distributed system for newsgroups, the precursor to Internet forums. (Is, I guess: it still exists, it just isn’t culturally important anymore.) In 1993, a large portion of Usenet newsgroup users, like Internet users in general, came from universities. Every September, a bunch of freshmen would show up and start exploring the network, having no idea what they were doing. They would violate all the social norms of Usenet. Many of them didn’t even want what the community offered (although they didn’t know it yet). With some effort on the part of existing members, after a few weeks, the people who didn’t fit in would leave and the people who did would figure out how to behave, and Usenet would go back to being its cozy self for another year. Everyone learned to brace themselves for September, and the world was great.

In September 1993, AOL started offering access to Usenet through their standard online services package. The potential audience of Usenet became massively larger overnight, now including new and inexperienced users from all walks of life across the country, and the seasonality ended, because even once the current AOL users were integrated – difficult enough by itself – people could join AOL at any time of year, and it was the nineties, so people were doing that in droves. Thus was born the Eternal September, or the September that never ended. September became the normal state of Usenet. The community was permanently changed, and veterans spent the next years talking knowingly and sometimes longingly about the prelapsarian Usenet. The previous social norms of the community were overwhelmed and largely disappeared, many of the existing core group were alienated, it become impossible to know a significant proportion of the people participating in many groups that had previously been human-scaled, the same questions came up over and over again, people were more likely to be rude or post without having anything meaningful to contribute, and so on.

It’s tempting to think of the Eternal September as the Fall of the Internet. As noted earlier, I think there are several takes on this view, which are partially contradictory but each have something to contribute to the discussion. Let’s look at two of them.

1: The beginning of wisdom

Suppose that, seeing what was happening to Usenet during the early days of the Eternal September, some charismatic administrator decided that everyone now had to get a Usenet License to use Usenet. Or maybe everyone who didn’t want to get AOL-ified splintered off into a different network where such licensing was required. Let’s suppose that to get a Usenet License, you have to have a bachelor’s degree, be an excellent writer in English – good users have to be able to communicate with all other users optimally! – and sign an agreement committing yourself to stringent standards of etiquette. This is a crude way to limit the audience of Licensed Usenet – we can certainly come up with alternatives that will yield a better community – but let’s run with this method for a moment, since this wasn’t all that far off from the original audience of Usenet. We’ll be leaving out plenty of people who would be worthy participants, but we’ll at least ensure we keep most of the AOL users who spoiled Usenet out.

What now do we have? Is this the Internet at all? I’m not sure it is.

The Internet was and is revolutionary in that it breaks down the barriers between people who would never have met each other, shared ideas, and collaborated before. The Internet is how I can video chat with anyone in the world by emailing them a link, how I can edit Wikipedia, how I can go read fascinating stories told by people who are in totally different circumstances than me on /r/AskReddit. The Internet is not how I can share long-form correspondence with other smart, Internet-savvy, college-educated-and-socialized people who are like me. We were already doing that before by writing essays and books and going to parties and sending letters around to people who seemed like they were working on similar ideas and might enjoy getting to know us. Usenet was just faster and more convenient. Wikipedia and its ilk are the point of the Internet; these things were truly impossible before.

Bringing in the messiness of humanity by casting the original Internet denizens out of the Garden is exactly what made the Internet meaningful and different, what left us free to make our own choices and improve the world, what leaves room for ultimate redemption. The Internet lost its innocence in September 1993, but it also grew up.

2: The first widely obvious example of unscalability

If you’re part of any online communities at all, or ever have been, from worldwide social networks to obscure Discord groups, take a moment to think about this: Which were or are your favorites? And what did you like about them?

Did you say you loved, say, Facebook? Sure, it’s useful, and it keeps you connected, and all that, but is it your favorite, as a community? Now what about that obscure Discord group?

True communities are formed from small groups of people who share common characteristics or are interested in similar topics. Trying to pull everyone in the world into one community is silly (Christopher Burg argues that social media is “impossible” for exactly this reason). It reduces the community to average, making most of the discourse uninteresting, the norms lowest-common-denominator, the governance impossible except by administrative fiat. Nobody wants to talk about only the things that every human has in common!

Further, expressing unpopular opinions (not just unpopular among the people who are interested in them, unpopular among the whole population, because everyone’s in the community and can participate) creates abuse and pile-ons, because people who don’t know you will show up and react and share, creating a feedback loop. People who have lived together with you and your posts for weeks or months don’t usually make snap judgments and rant about your posts all around the Internet; people who see your content without that social context do it all the time.

The Internet, and most software for the purpose of creating online communities, is not currently designed to foster these small groups. Because everyone can participate, we assume that everyone should participate, so we make it open. The AOL users can just flood right in as in the Eternal September, make the community too large, and destroy its norms. Importantly, the AOL users aren’t acting maliciously – communities designed on a human scale, which are the best ones, just don’t respond well to this kind of pressure. They’re designed to change slowly. Physical communities moderate themselves, maintaining a stable equilibrium with occasional raging conflicts that result in incremental changes. Internet communities that have grown beyond the normal human scale naturally consist of raging conflicts; they have to be kept in shape by force and artificial barriers to be useful. And we often fail to do even that well: just look at your social media feed in 2021.

What we should get out of the Internet is the ability to incrementally improve on physical communities by removing the need for physical co-location. Nowadays, useful and meaningful communities can form from people sharing incredibly rare interests. Even if only fifty people in the entire world are interested in something, they can quickly and easily talk to each other at no cost. The Internet serves to help them find each other, which would have been nigh on impossible before.

Indeed, the communities with rare interests are lucky. Except for the occasional troll who drifts in and finds the topic weird, they naturally moderate because there aren’t enough people interested to create a problem. The more mainstream systems should consider erecting artificial barriers to participation or scaling; see Clay Shirky for more on this (search for “Four Things to Design For,” though the whole thing is worth reading – it’s fascinating and was a major influence on this post).


I think it’s true, as argued in #1, that the introduction of AOL and similar services was a (halting) beginning of wisdom for the Internet, and that without it the Internet wouldn’t have become its fullest self. Yet at the same time, as argued in #2, twenty-eight years later, we are still designing systems that don’t scale and expecting them to work well with thousands or millions of people. It seems to me that our reaction is to insist that they are working well. For instance, I don’t use Twitter much myself, beyond reading threads that other people have pointed me to, but my understanding is that it’s entirely possible to get a useful and positive experience from your Twitter feed – but only if you work at it by proactively blocking people, paying attention to who you follow, and so on. If you aren’t careful, the default state of Twitter is a toxic mess. (I do use Facebook, and my feed is not too far off from this, to be honest!) Yet Twitter fans will say if you “just” do this and this and this, everything will be fine and you’ll get lots of value out of the platform.

That’s probably true. But why is that considered acceptable? Can we not design our communities and social media in a more effective way, returning some of the unintentional barriers that we had before the Eternal September to the platform itself, instead of placing the responsibility on individual users? I’m glad that some people have found ways to make these tools useful for themselves, but it seems to me that we’re still missing something. With the minimal amount of experimentation we’ve seen so far, it seems unlikely we’ve found anything close to the best way to balance openness and sense of community on the Internet. In fact, we haven’t even concluded such a balance is important yet, which is why I’m writing this post in the first place.

The benefits of online communities are intrinsic to the Internet: anyone who has an Internet connection, anywhere in the world, is a potential member of the community. The drawbacks, however, are not: it doesn’t have to be possible for hundreds of AOL users to flood in all at once and ruin the community. Software can be made to do almost anything. And contrary to my Usenet License thought experiment, the barriers don’t have to exclude certain types of people (those who aren’t highly educated); we can still get the democratization the Internet allows, but select for people who are most interested (or some other useful dimension) instead. For instance, we could make a community you could only join in September. Implementing this would be trivial, and it would keep out eleven out of twelve people who weren’t that interested. The group of people who were interested enough to put it on their calendar and come back a few months later would make up most of the community. Moreover, all the new members would show up at the same time, so it would be much easier to get them integrated into the community’s social fabric.

I hope we start to see more experiments along these lines, to suss out which drawbacks are inevitable and which we can correct, and to understand exactly which parts of takes 1 and 2 are correct.