#496 1969 · Steve Crocker / ARPANET Network Working Group · Computing standards
A UCLA grad student stripped his own memo of authority so rivals would dare write one back
the problem
No one owned the authority to set network protocol rules the four sites had to share
background
By 1969 the four ARPANET sites — UCLA, Stanford Research Institute, UC Santa Barbara and the University of Utah — had a rudimentary network but no agreed rules for how their computers should actually talk to each other. The people meeting to figure it out were mostly graduate students and junior staff, chosen from competing university labs with no formal charter, no appointed chair, and no professional standards committee anywhere above them; everyone assumed a 'real' team of senior engineers would eventually take over.
Steve Crocker, a 25-year-old UCLA student, volunteered to write up the group's early thinking on network protocols. The obvious way to do that job was the way engineering write-ups normally look: a numbered specification stating what the system should do. But nobody in the group had been given the standing to issue a specification, and Crocker had no way of knowing who, if anyone, might later claim that authority and object to a grad student having pre-empted it.
what everyone would do
Write the network's first protocol notes the way engineering documentation is normally written: as a specification stating what should be done. That format is exactly what a group with no appointed authority cannot safely publish, because it reads as one lab deciding for the others — and Crocker feared it would draw exactly that reaction from whichever senior engineers might later be placed in charge.
what they saw
Crocker saw that the danger wasn't the content of the memo, it was its claimed status. A document that asserts nothing has nothing for a rival team to fight over; strip the authority out of the format itself and the content becomes safe to write and safe to answer, regardless of who writes it or where they rank.
the move
Crocker wrote the note anyway, but titled it a 'Request for Comments' rather than a specification, and wrote it in deliberately tentative, non-binding language, explicitly inviting disagreement rather than announcing a decision, so that nothing in it could be read as one team asserting authority over the others.
why it works
Naming the notes 'Request for Comments' and writing them as open questions rather than decisions removed the social risk of contributing: nobody had to out-rank anyone else to publish one, and nobody reading one had to defer to anyone else's authority to respond. That low a barrier to entry let dozens of engineers across competing labs contribute freely, and because contributions could freely accumulate and be tested, the group could converge on shared protocols by simple voluntary adoption — 'rough consensus and running code' — instead of waiting for a committee empowered to rule.
the payoff
The tone held: contributors kept writing in the same disclaimed, first-person, no-authority register indefinitely. By the note's 40th anniversary in 2009 the RFC series numbered over 5,000 documents, all published free and open to anyone, and had become the Internet's actual mechanism for setting technical standards — not through a committee vote but through what participants called 'rough consensus and running code.'
where it breaks
The move only works before real institutional stakes exist — money, patents, or a legally binding standards body — because disclaiming authority doesn't remove genuine conflict, it only works where no one actually has enforcement power yet and adoption can stay voluntary. Once a standard has to be mandated rather than merely adopted, or once competing parties have something to lose from someone else's draft winning, the humble framing stops defusing the fight and a real authority structure becomes unavoidable.
what came after
The RFC format never lost its humble name even after it became the formal, load-bearing standards process for the entire Internet; the series now runs past 9,000 documents and is still how IETF standards are published today. Crocker received the IEEE Internet Award in 2002 for the process this single naming choice started.
references
- [1]RFC 1000 — The Request For Comments Reference GuideRFC Editor / Internet Engineering Task Force, 1987rfc-editor.org
- [2]How the Internet Got Its RulesThe New York Times, 2009nytimes.com