History of the Internet

Message Had a Standard. Mail Still Rode on FTP: RFC 733

The Message Had a Standard. Mail Still Rode on FTP: RFC 733

In 1977, RFC 733 gave ARPANET text messages a common syntax while mail delivery still used FTP’s MAIL and MLFL commands. The standard drew a careful boundary: make messages legible between hosts without prescribing one mail program, user interface or fully adopted transfer protocol.

RFC 4714 Put Copyediting Under Change Control

An RFC could have passed technical approval and still face another consequential review. A copyeditor might be correcting grammar, but after approval even a small rewrite crossed a boundary: the published sentence had to remain answerable to the process that had agreed it.

When RFC 6410 Removed Draft Standard, It Left Old Labels in Place

RFC 6410 removed the Draft Standard rung from the IETF’s maturity ladder in 2011. Its transition clause did not move the documents already there: their old classification could remain until someone took explicit action.

RFC 2860 Draws a Boundary Around IANA’s Technical Work

The sentence that gives RFC 2860 its sharpest edge is also the one most easily overread: policy issues in domain-name and IP-block assignments sit outside this memorandum, while named technical assignments remain inside it. The distinction matters because the document is a working agreement about protocol parameters, not a charter for every role that carries the IANA name.

Great Renaming Gave Usenet Operators a Better Feed Filter

The Great Renaming Gave Usenet Operators a Better Feed Filter

In 1986–87, Usenet did more than replace old group names with a tidier vocabulary. Topic hierarchies gave site administrators a clearer way to choose which discussions their machines would carry. The surviving debate also draws a line between people consulted about a change and the operators who could actually configure its distribution.

NORSAR Kept 2.4 kbit/s for Seismic Data as ARPANET Arrived

When NORSAR’s voice-grade line was upgraded for ARPANET, the seismic stream did not disappear into a larger network. The year-end 1973 operating report says 2.4 kbit/s remained assigned to seismic data; NORSAR and the University of London Institute of Computer Science shared the other 7.2 kbit/s. The connection was an allocation problem before it was an origin story.

Who Could Apply for a Top-Level Domain? RFC 920’s First Eligibility List

In October 1984, the first top-level domain list was also a set of admission rules: some names were temporary, others stood for broad categories, and none could enter the tree without an administrator and a route to authorization.

Three Networks, One Transmission: The 1977 Internetting Demonstration

The famous van was only one endpoint. The real test was whether a message could cross packet radio, the ARPANET and a satellite packet network without turning them into one enlarged network. The surviving record supports a successful route—and a much narrower claim than “the Internet was born.”

Glossary Left Interoperability Undefined: RFC 1208’s Moving Target

The Glossary Left Interoperability Undefined: RFC 1208’s Moving Target

Adapted from a pocket guide for Interop ’90, RFC 1208 helped readers navigate a fast-moving field without pretending that one short definition could prove two vendors’ systems actually worked together.

Cache Changed the Destination, Not the Root: AlterNIC’s 1997 InterNIC Redirect

The Cache Changed the Destination, Not the Root: AlterNIC’s 1997 InterNIC Redirect

A resolver in one operator’s network returned an AlterNIC address for `www.internic.net`. Its cache record exposed the route the redirect took: additional DNS data had become a local answer. That was evidence of a changed destination at one resolver, not of a new authority taking over the public root.

A Customer Link Became a Security Boundary: RFC 2267 and BCP 38

In 1998, an informational memo moved one check from the machine under attack to the provider edge nearest the traffic’s declared source. RFC 2267 asked whether a packet arriving from a customer network used a source prefix that network could legitimately send. The rule narrowed where an attack could appear to come from; it did not authenticate a host, stop misuse of an allowed prefix or prove that a filter was running.

Pigeon Returned an IP Packet. A Human Still Had to Finish Delivery: RFC 1149 in Bergen

The Pigeon Returned an IP Packet. A Human Still Had to Finish Delivery: RFC 1149 in Bergen

In Bergen in 2001, a Linux user group tested a ten-year-old specification for carrying IP datagrams by pigeon. Four replies reached the log, but only after a person removed the paper, scanned it and corrected the OCR. The surviving record shows what this one route tested—and what it did not.

One Connection Was One Point: How RFC 1025 Tested TCP/IP Compatibility

Before compatibility could be assumed, it had to be demonstrated between actual implementations. RFC 1025 turns that early Internet practice into a scorecard: start with one connection, then add peers, gateways, damaged packets and cases designed to expose what a successful handshake alone could not prove.

Frame Held 512 Bytes. The Datagram Could Span 120 of Them: RFC 1201

The Frame Held 512 Bytes. The Datagram Could Span 120 of Them: RFC 1201

The 1991 ARCNET specification made a small fixed frame carry a much larger IP datagram by moving fragmentation down to the data link. Its 60,480-octet ceiling was only a format limit: every network still had to choose and share a practical maximum.

Before Joining the Internet, You Needed Its Oral Tradition: RFC 1118

In September 1989, RFC 1118 set out to help a small, isolated IP network cross into the Internet. Its most useful contribution was a map of people, operating centers and unwritten practice; its own disclaimer showed how quickly that map could age.

One Empty Query Could Show Who Was Online: Name/Finger’s Presence Boundary

A 1977 empty CRLF asked one host to enumerate every user on that machine. The request could travel; the answer stayed a site-produced report whose fields and disclosure boundary were local.

Three-Year Plan for Internet White Pages: RFC 1107

The Three-Year Plan for Internet White Pages: RFC 1107

In July 1989, a plan for finding people across research networks arrived with an unusually clear qualification: a service might be possible in three years, but the memo did not represent a committed Internet-community research activity. That gap between timetable and commitment is where RFC 1107’s history begins.

A Name Server Could Not Deliver a Datagram to the Group: RFC 1001 and RFC 1002

A name directory can tell a sender where a name is registered. In the March 1987 NetBIOS-over-TCP proposal, that answer still did not distribute a group datagram. RFC 1001 and RFC 1002 gave that second job to a distinct logical service, then left its forwarding result largely outside the sender’s view.

Ethernet Handoff Was Common. An Open IGP Was Not: RFC 1009

The Ethernet Handoff Was Common. An Open IGP Was Not: RFC 1009

In 1987, the National Science Foundation required gateways from different vendors to meet at a familiar boundary: Ethernet, with serial links as a minimum companion. The same specification admitted that the vendors still had no open interior gateway protocol in common. One layer of interconnection had a shared floor; another did not.

Archie’s Search Was Only as Fresh as Its Last FTP Listing

Archie made a scattered set of anonymous FTP archives searchable by collecting their directory listings. That small distinction shaped the service: its answer could point to a filename and a place, but the file still lived on someone else’s machine—and the index could only describe what it had last seen.