LockerGnomeTech news and nostalgia for today's technology professionals
Subscribe

Search

Retrointernet historyRFCARPANET

RFC 1 opened the Internet’s document series by admitting almost nothing was firm

Steve Crocker’s early network notes welcomed unfinished ideas, even one-sentence contributions, before the process acquired its official weight.

Chris in a 1969 lab pinning a single unfinished note to a corkboard crowded with drafts beside a cabinet-sized computer
Illustration: LockerGnome

Steve Crocker had agreements to write down and no grand authority to announce. The opening of RFC 1 names the people who had been meeting and practically puts a pencil in the reader’s hand to mark up the unfinished questions, which would be an unusually honest onboarding document today.

The memo divides responsibility between the packet-switching machines, called Interface Message Processors, and the computers attached to them. BBN was specifying the IMP software. The host groups still had to work out what their machines would say to one another. Somebody had to start that conversation before a network could carry it.

Its technical detail makes the modesty funnier. Messages could contain up to 8,080 bits plus a header. The header had destination, link, trace, and spare fields. This was careful engineering alongside openly unsettled decisions, a combination that tends to disappear the moment a document gets a handsome cover.

Crocker’s retrospective in RFC 8700 supplies the human problem. The junior group expected someone more senior to arrive with official responsibility. He worried their notes might look like a claim to authority. Bill Duvall suggested the label Request for Comments, and Crocker used it to make the invitation unmistakable.

The name survived rather better than the anxiety. Today an RFC number can make a sentence sound as though somebody carved it into a router. The original authors were trying to prevent precisely that reflex.

RFC 3’s documentation conventions pushed the door wider. A contribution could be one sentence. Questions could appear without answers, and an implementation suggestion could arrive without a polished introduction. Crocker explicitly worried that putting an idea in writing would make people treat it as authoritative.

“Notes are encouraged to be timely rather than polished,” he wrote. Imagine a documentation process whose submission requirement makes room for the thing somebody actually needs to say. Somewhere, a template with twelve mandatory sections just felt a disturbance.

That openness also explains why reading the status block matters now. RFC 2026 describes standards-related publications and distinguishes standards-track material from categories such as Informational and Experimental. Even the fiftieth-anniversary retrospective labels itself Informational. A document can belong in the permanent record without ordering your next deployment around.

The useful habit is to follow both the document’s category and its update history when applying it. A familiar number in an old blog post is a starting point for reading, and the RFC Editor’s document information gives you a route to later changes. That gives an operator a way to distinguish the historical document from the version that actually applies to the system being configured.

The first one asked for comments.

More Retro

Every Retro story →