In brief
FidoNet is a worldwide network of electronic bulletin boards that never had a company, a bank account, or a single server. It grew out of a hobby program the programmer Tom Jennings wrote for his own BBS in San Francisco, and it turned into a network when he dialed up an acquaintance in Baltimore. The peak was 39,688 nodes at the end of 1995, and the decline began after that, though not everywhere and not at the same time: in the former USSR the network kept growing for another five years, and from 2002 to 2009 it held more than half of all the nodes in the world. FidoNet did not die: in 2026 there are 1,233 nodes worldwide, 414 of them in the former USSR.
How it started (the founders)
Tom Jennings was testing the link between his own BBS (Fido #1, San Francisco) and John Madill's BBS (Fido #2, Baltimore) in the spring of 1984, reaching the seventh version of the program by June. The network started working in 1984, and three independent primary sources confirm that; the exact year in which the single-computer Fido program was written, before the network existed, is not named separately by any source we found (see "Discrepancies"). Jennings described the project as a hobby from the start: "totally supported by it's users and sysops, and in many ways is similar to ham radio," a business in no sense at all.
Growth outran what one person could administer: from ~2 nodes in the spring to ~50 by September 1984 and 160+ by the beginning of 1985. Errors in the list of nodes ("every single week without fail there was at least one wrong number") were the norm until August 1984, a matter of months after the start, when Jennings handed the keeping of the nodelist to a team of volunteers in St. Louis (Tony Clark, Ben Baker, Ken Kaplan, John Wichman, Mike Mellinger), who brought the error rate down to nearly zero within a few months; another source independently confirmed this two decades later. On December 4, 1984 the first issue of FidoNews came out, a mailing list in which Jennings, declaring himself editor, wrote: "I do NOT wish to be editor."
The ideology was deliberate rather than accidental: "Tom Jennings intended FidoNet to be a cooperative anarchy to provide minimal-cost public access to electronic mail"; in 1993 he himself said that "anarchism is a personal philosophy and not a political issue at all," and the design of the network was shaped directly by his broader punk and anarchist identity, which from 1988 also took the form of the zine Homocore, published together with Deke Nihilson. The attempt to formalize the network into an organization (IFNA) failed in the 1989 referendum, 480 votes in favor against 1,417 opposed: the network collectively rejected its own bureaucratization less than five years after launch.
Year-by-year timeline
- 1984 (spring): a test link between two nodes, San Francisco and Baltimore fact
- 1984 (August): the keeping of the nodelist was handed to volunteers in St. Louis fact
- 1984-12-04: the first issue of FidoNews came out fact
- 1985 (Jan–Feb): 160+ nodes already fact
- 1986: 1,209 nodes worldwide (zones were not yet broken out in the nodelist); in February, at a sysops' party in Dallas, the idea of EchoMail was born and Jeff Rush took on the implementation; in October the Standards Committee approved the future four-level zone:net/node.point addressing fact
- 1987: zones and zone:net/node.point addressing (the decision had been taken a year earlier, see 1986) showed up in the nodelist for the first time; 2,465 nodes; on December 12 the EchoMail standard, FTS-0004, was formalized fact
- 1989-06-09: Policy4, the network's constitutional document, was adopted; 1989-12-01: the referendum on making IFNA mandatory failed, 480 against 1,417; 6,169 nodes worldwide
- 1990: the first nodes in the USSR appeared in Novosibirsk, region 2:50, 5 nodes at the end of the year; the exact date of September 21 turns up in two secondary retellings, but that is a blogger's transcription of one particular nodelist rather than our own check of a primary source: our series confirms only the year, and that at the end of 1990 the region was coordinated from Novosibirsk estimate; 10,168 worldwide by then
- 1991: region 50 (the former USSR), 103 nodes; the world, 14,598 fact
- 1994: Zone 1 (North America) at its own peak, 19,025 nodes; region 50, 1,408 nodes, 3.9% of the world network fact
- 1995: the world peak by year's end, 39,688 nodes fact
- 1996: the world shrank to 33,898, but this was the year Zone 2 (Europe) reached its own peak, 18,242 nodes fact
- 1998: Jennings left the network for good, having never accepted its bureaucratization fact
- 2000: the world was down to 15,692 nodes, while region 50 was right at its peak, 6,671 nodes (42.5% of the world network); Moscow (1,583) and St. Petersburg (1,045) were at their maximum too fact
- 2002–2009: the former USSR accounted for more than half of all the nodes in the world fact
- 2026: 1,233 nodes worldwide, 414 (33.6%) of them in region 50
Lesser-known but significant facts
- The former USSR outlived the network's world peak by five years and for a while made up more than half of the whole network. The world had been shrinking since 1996, while region 50 grew to 6,671 nodes in 2000 and from 2002 to 2009 held no less than half of all the nodes in the world.
- EchoMail, the network's main feature and its answer to forums, was thought up at a sysops' pizza party in Dallas, not in a lab. Chuck Lawson proposed the idea, Jeff Rush implemented it in a few weeks, and the formal FTSC standard was adopted only two years later.
- Most of the real users never made it into the nodelist at all. Points, a lightweight client layer on top of a node, connected only to their own host and carried none of a node's network obligations; the official node count systematically understates the number of living people in the network.
- The network held social order together without a single moderator, by escalating the "plus marks" beside an offender's name up to full read-only, and as far as a sanction on an entire node, sysop included.
- A Fido station, by the account of one of the organizers, helped pass news out of the White House in Moscow in August 1991, when other channels of communication were being cut off estimate.
Legend vs. the record
- Legend: the world peak of FidoNet was 1996, around 40,000 nodes. The retellings give different years for the same round figure: Jennings recalled "32,000 by the early 1990's," Habr writes around 40,000 nodes with no year at all, and Jason Scott gives ~40,000 by 1994. The document: the weekly nodelists from July 1995 to February 1996 give a peak week of 39,827 nodes in the nodelist of November 24, 1995; a plateau within 1% of the peak held from July 7, 1995 to February 2, 1996. North America was already losing nodes in the spring of 1995, while Europe and the former USSR grew until the summer of 1996: the peak fell at the moment when the growth of one zone stopped covering the decline of another. Verdict: the peak was at the end of 1995, and around 39,000–40,000 in 1996 is the descent from the plateau.
- Legend: FidoNet is dead. In everyday talk, including among the participants themselves, who joke that Fido has not died, it just smells that way, the network is counted as deceased. The document: an independent present-day retrospective site records 2,000–3,000 active nodes in 2024 on the Binkp protocol over TCP/IP; our own count from the nodelists gives 1,233 nodes as of 2026, 414 of them in region 50; FTN technology, according to the same source, is used today for secure data channels in banks and security agencies. Verdict: the network lost 97% of its peak but did not disappear; the legend confuses having become small with having died.
- Legend: the internet killed Fido everywhere at once. The document shows the opposite, a divergence of almost a decade. The world as a whole began shrinking right after 1995; region 50 was only picking up speed at that point and reached its own peak in 2000, after which it held the majority of the entire world network from 2002 to 2009. The reason for the lag, according to an independent source, was that the internet was a completely inaccessible technology for most people while telephone service, on the contrary, stayed cheap; ADSL providers appeared in Moscow and St. Petersburg only by 2000, and the replacement of analog telephone exchanges with digital ones stopped letting the modem signal through. Verdict: false as a single global event, with at least two different decay curves separated by a lag of 5–10 years.
- Legend: the first Soviet node was in Moscow. The document and independent accounts from participants converge on a different city. The first node in the USSR appeared in 1990 in Novosibirsk, not in Moscow, and this is confirmed by a first-hand participant in the events and by two independent journalistic sources. The exact date of September 21 is named in two secondary retellings, down to a specific nodelist file, but that is a blogger's transcription rather than our own check against a primary source: our measured series confirms only the year, and that at the end of 1990 the region was coordinated from Novosibirsk estimate. Moscow got its node about a week later, or after an intermediate station in the Urals. Verdict: Moscow was not the first; the city is confirmed three times independently, and the exact day (September 21) is plausible but not confirmed by a primary check of our own.
The first growth lever
The lever was operational rather than marketing: the network cut the barrier to entry to zero (free software, your own modem and phone line, and that was the whole of membership) and almost at once removed the one real bottleneck that could have stopped that free entry. Over the first few months the number of nodes grew from 2 to 50, and the list of node numbers began to break down: "every single week without fail there was at least one wrong number." Jennings did not try to scale that work himself: in August 1984, two months after the launch, he handed the keeping of the nodelist to a volunteer team in St. Louis, which brought the error rate down to nearly zero within a few months, and the network went on growing to 160+ nodes by the beginning of 1985 with no single point of failure in the person of its founder.
The second element was the economics of the connection itself. The network was arranged so that even intercity and international mail cost almost nothing: nodes inside a city called each other directly, while traffic between cities was gathered through "inbound hosts" so that nobody had to call each node separately; intercontinental mail went through "zonegates," and all of it most often at night, because "US telephone rates are much lower in the middle of the night." Growth needed neither capital nor a sales department, only a willing sysop and a phone line he already had.
Parallels today (projects from the catalog)
- Blume (
blume): the same Policy4 formula, "amateur system... unpaid volunteers... not a commercial venture," free and open source from day one, zero friction. The difference: at Blume that is the choice of a single solo maker rather than a structural property of a network of thousands of independent operators. view this project's dossier → - UtilVox (
util-vox): FidoNet grew on top of someone else's infrastructure (the public telephone network, the night rate) with no money of its own and no wires of its own; UtilVox likewise builds nothing from scratch and charges nothing for use. The difference: UtilVox's growth was bought with programmatic SEO and a mass launch of 170 tools at once rather than with a chain of personal calls between sysops. view this project's dossier →
What a builder can take from this in 2026
- Cut the barrier to entry to zero, but close the first bottleneck that creates right away. Free entry produced explosive growth within months; had Jennings not delegated the nodelist in those first two months, growth would have choked on errors before the network became interesting to anyone.
- A single aggregated chart can hide two opposite markets. The claim that FidoNet has been dying since 1996 is true for the world and false for an entire region for another ten years; look only at the global curve and the best (or worst) market stays invisible.
- Building on someone else's already-paid-for infrastructure is a durable growth economy rather than a one-off hack. The telephone network's night rate made international communication almost free without a single wire of your own.
- Formal structures of control can be replaced by a transparent escalation of the rules, plus marks → read-only → a sanction on the whole node, instead of hiring moderators.
- A network outlives its inventor and its original economics as long as someone still needs the substrate. The 1,233 nodes in 2026 are not nostalgia; they are people who still need the FTN protocol for something.
Discrepancies and what we could not verify
- The absolute node counts by year. Our series from the nodelists matches neither Bush's table of 1992/93 nor the rough estimates on technicshistory.com; for 1986, for example, 1,209 (ours) vs. 1,400 (Bush) vs. ~1,000 (technicshistory). The order of magnitude and the shape of the curve agree, the exact numbers do not; this dossier uses our measured series and nothing else.
- The exact year the single-computer Fido program was written (before the network). The popular figure of 1983 was traced to a specific source, Cory Doctorow's 2019 essay for the EFF, which Wikipedia cites in its article about Jennings; that is a retrospective essay, not a document of the era and not the founder's words, and Doctorow's own direct source for the date could not be checked. No document of the era and no text by Jennings himself, neither the note of Feb 1985, nor the second part of Aug 1985, nor the interviews from his archive, names 1983 in connection with writing the code; the only mention of 1983 by Jennings that was found relates to conversations about anarchist philosophy, not to the code. Every source that gives a year at all names 1984 as the moment of the test link between two nodes. There is still no ground for fixing 1983 as a fact; 1984 remains the established fact of the network's launch, and the origin of the 1983 figure has now been traced but not confirmed unverified.
- The date of the first issue of FidoNews. Some aggregated search results give December 1, 1984, while the mirror archive's own index gives December 4, 1984; the archive's date is the one used.
- The third city of the first Soviet nodes. One participant in the events names Sverdlovsk/Yekaterinburg, a magazine long read names Chelyabinsk; both agree on Novosibirsk as the first city and on September 1990.
- Who founded IFNA. Only one source names Ken Kaplan as the sole founder; an insider account describes the founders collectively, with no name.
- The "peak" of the Russian segment as the participants remember it. One source says the peak was in 1998 and held until 2002; an exact measurement from the nodelists puts the peak in 2000, at 6,671 nodes, and the measured figure is the one used.
- The White House episode of August 1991 and the 1997 protest against per-minute billing by MGTS: both facts rest on a single publication; a second independent source was found neither in the first pass nor in a targeted repeat search, and one possible reason is that the episode may have been confused with the better-documented role of the Relcom network in August 1991.
Sources (primary first)
- Tom Jennings — FidoNet History, 1985 (mirror)
- Sensitive Research — Tom Jennings's personal archive on FidoNet
- Jennings History mirror (apple2.org.za)
- FidoNews No. 1, 1984-12-04 (archive mirror)
- FidoNet Policy Document 4.07, 1989-06-09
- FTS-0004, EchoMail Specification, 1987-12-12
- Randy Bush — FidoNet: Technology, Tools, and History (INET'92)
- Wynn Wagner — a history of EchoMail (a document of the era)
- Hacker News — a participant's first-hand account of the first Soviet node, 2017
Secondary (context, cross-checking):
- LiveJournal — "Fidonet: 33 years of the network in the USSR and Russia"
- BBS Wiki — the 1990 timeline (based on Jason Scott, The TEXTFILES.COM BBS Timeline) — independent confirmation of the IFNA vote, 480/1417
- Habr / 1cloud — "Almost anarchy": a short history of Fidonet
- Moskvich Mag — "The best internet that did not survive"
- ASCII (Jason Scott) — a blog post on the ROFLCON panel, 2010
- technicshistory.com — The Era of Fragmentation, Part 4: The Anarchists
- fidonet.ximg.app — FidoNet: The Internet Before the Internet
- Wikipedia (Tom Jennings, Homocore) — a pointer only, not a source of fact