Ask anyone who knows the internet’s early history what “Eternal September” means and you’ll hear roughly the same story. In September 1993, America Online opened its gateway to Usenet — the network of discussion forums that had been home to the internet’s early culture — and an unprecedented number of new users arrived at once. Before that, each September, university freshmen had flooded in when their schools provisioned internet accounts, and the regulars had developed rituals for surviving the annual disruption. AOL made it permanent: anyone with a dial-up subscription could arrive at any time, without knowing the culture, without being absorbed. Dave Fischer, a Usenet regular, put the situation into a sentence that stuck: “September 1993 will go down in net.history as the September that never ended.”

Fischer did write that line. The rest of the story is wrong.

January 1994, alt.folklore.computers

The phrase was coined in January 1994, not September 1993. The company that prompted it was Delphi, not AOL. And it didn’t emerge from nowhere — it crystallized a conversation that had been developing in the newsgroup for weeks.

To understand why “September” was the metaphor: each year, when classes resumed, universities provisioned new accounts for incoming students. Those students arrived on Usenet unfamiliar with its customs — the conventions around quoting, thread etiquette, the specific cultures of individual newsgroups — and the regulars had to absorb them. This was manageable because it was annual: a burst of noise, then relative quiet as newcomers either learned or drifted away. The word “September” came to refer not just to the month but to the whole pattern of disruption and gradual integration.

The phenomenon Fischer’s phrase would eventually name was already years old by 1993. Commercial services had been offering Usenet access since the late 1980s — FidoNet had a gateway, then CompuServe, then Delphi. Each arrival generated complaints from established Usenet users about newcomers who didn’t understand the culture. Delphi, a commercial online service oriented toward business users and enthusiasts, had opened Usenet access to its subscribers. The regulars received users posting from CompuServe and Delphi with what Driscoll describes as “comparatively tepid treatment” — less hostile than what AOL users would later face, but not welcome.

On January 8, 1994, Joel Furr posted to alt.folklore.computers with a subject line that read, in full: “Is it just me, or has Delphi unleashed a staggering amount of weirdos on the net?” Karl Reinsch replied in the same thread: “Of course it’s perpetually September for Delphi users, isn’t it?” Driscoll traces these posts in his 2023 account; the original threads are not accessible through current Google Groups archives, where the 1994 alt.folklore.computers records are fragmented. The framing — a perpetual September, an influx that wouldn’t end — was already attached to Delphi, already in circulation, before Fischer wrote a word.

Later in January, John William Chambless — posting under the name Billy Chambless — from the University of Southern Mississippi, asked the group whether the annual “september threads” should be counted as one continuous event, given how persistent the disruption had become. Fischer replied with what became the phrase: “September 1993 will go down in net.history as the September that never ended.”

The popular story tracks the compact phrase — “Eternal September,” two words — not Fischer’s longer original. Who first shortened the phrase is not clear from the available record; Driscoll’s account identifies Fischer as the author of “the September that never ended” without establishing when or by whom the compact form appeared.

Fischer wrote that in late January 1994, describing September 1993 in the past tense — as something that had already happened and proved permanent. The September he named was real: the month the Delphi influx had become acute, the month the annual freshmen arrived, the month that, in Fischer’s reading, simply refused to close. He was not predicting anything. He was naming something that, from his vantage point in late January, had already been true for several months.

The phrase appeared in alt.folklore.computers — the newsgroup adjacent to alt.folklore.urban, where internet legends were dissected and traced. Fischer coined a phrase about internet cultural memory in the newsgroup dedicated to the folklore of computers. That it would itself become misremembered internet folklore is a recursion that took thirty years to notice.

February 28, 1994

AOL’s Usenet gateway opened on February 28, 1994. That is approximately five weeks after Fischer wrote his phrase.

Driscoll states this plainly: “The USENET gateway opened on February 28, 1994 after weeks of testing.” The error in the popular account is not imprecision or reasonable simplification. The date — September 1993 — is off by six months. The company is wrong. Fischer was describing Delphi, in January 1994, before AOL arrived.

What AOL’s arrival actually changed was scale. AOL was technically worse-equipped for Usenet participation: its interface made proper message quoting difficult, charged by the minute while users were online, and introduced delays in post visibility. AOL users faced open contempt from established Usenet culture — described in contemporary accounts as “morons” and “McDonald’s clientele.” The September that Fischer had named about Delphi proved, with AOL, to have been a preview of something larger.

This is the part of the story the popular version gets structurally right: AOL’s arrival was disruptive in the way the phrase describes, and it permanently changed what Usenet was. The problem is attribution. Fischer described a phenomenon that had already begun, about a company that was not AOL, before AOL arrived. AOL confirmed his description at a volume he couldn’t have anticipated. The phrase attached to AOL because AOL was the event that proved it true at scale — not because Fischer said anything about AOL.

The Jargon File

The error did not spread at random. It was institutionalized.

The Jargon File is Eric S. Raymond’s compiled dictionary of hacker culture — a document that began circulating informally in the 1970s, published in print as The New Hacker’s Dictionary, and maintained at catb.org. For the particular kind of internet history that “Eternal September” represents — internet folklore, community language, the culture of early Usenet — the Jargon File is the authoritative reference. It is where this kind of knowledge goes to become official. Citing the Jargon File on a question of hacker culture has the force of primary source: the Jargon File is the cultural record, as much as any archive of original posts.

Driscoll identifies the Jargon File as the primary vector for the misattribution: the error “is enshrined in the Jargon File and has been reproduced across the Web.” The Jargon File entry could not be confirmed directly — catb.org is not currently accessible — so whether the entry has been corrected since Driscoll’s 2023 article remains unknown. Either the authoritative source continues to state the wrong story, or the correction has been made and at least one node of the propagation chain is closed.

The mechanism of propagation is the mechanism of authority. An authoritative source stated a specific company and a specific date without citing the primary posts. Subsequent retellings echoed or cited the Jargon File without checking further. By the time Driscoll queried a language model about the origin of Eternal September, the model returned: “The term originated in the early days of the internet, specifically in September 1993, when AOL began offering new users access to Usenet.” The Jargon File’s error, passed through enough repetition, had become what the web knew.

Nobody fabricated the wrong date. What happened is more ordinary: a secondary source had authority, the primary record was inconvenient to check, and repetition did the rest. The same pattern runs through a significant fraction of internet cultural history, and the Jargon File, for all its genuine value, carries it here.

Fischer coined a phrase about how an influx of newcomers can overwrite a community’s memory of the norms that came before. The memory of that phrase’s own origin was then overwritten — by the larger influx that arrived approximately five weeks after Fischer wrote it, and by the authoritative source that filed the wrong story. A phrase about overwriting got the same treatment it described.