rough consensus and running code
expressionnetworkingops culture
The IETF's working philosophy: standards emerge from broad agreement plus implementations that actually work.
David Clark's 1992 formulation - we reject kings, presidents and voting; we believe in rough consensus and running code - is why internet protocols are proven in interop before they are finished on paper. It is the cultural DNA behind every RFC this site teaches.
We reject kings, presidents and voting. We believe in rough consensus and running code. David Clark's line describes how the IETF makes decisions, and it is a genuine alternative to both authority and formal ballot.
Rough consensus means no formal vote and no requirement of unanimity: the chair judges whether objections have been addressed, not whether everyone agrees, and a persistent minority does not hold a veto. Running code means a proposal with two interoperating implementations outweighs an elegant specification nobody has built. Together they select for what works over what wins an argument.
The honest limits are worth stating because the phrase gets used as an unqualified virtue. Rough consensus rewards those who can attend meetings and sustain long participation, which favours the well-funded, and running code favours whoever already has engineers. The IETF has debated its own capture by large implementers repeatedly. It nonetheless produced TCP/IP, HTTP and TLS, which is a stronger record than most standards bodies with more formal legitimacy, and the tension between openness and effectiveness is a live one rather than a solved one.