Internet architecture principles revealed by IETF conversation analysis using vCons
An exploration of Internet architecture principles as they surface in real IETF working group deliberations — extracted and analyzed using vCons.
The Internet wasn’t designed in a single blueprint. It emerged over decades through thousands of conversations — debates, compromises, and moments of clarity in working group sessions at the Internet Engineering Task Force (IETF). The design principles that hold the Internet together are not just written in RFCs. They are lived in these conversations, invoked when engineers face hard trade-offs, and sometimes bent when reality pushes back.
This site is an attempt to surface those principles by analyzing 2,249 IETF session recordings from meetings 110 through 123 (March 2021 – July 2025). We processed the transcripts of every recorded working group session, searching for moments where participants discussed, debated, or applied core network design principles — from the end-to-end principle to Postel’s Law, from fate sharing to protocol ossification.
The result is a design guide to the Internet, not as it was imagined on paper, but as it is practiced in the rooms where the standards are made.
Each principle has its own chapter with:
Whether you’re a protocol designer, an engineering manager, or an executive trying to understand why Internet architecture matters for your business, this guide is for you.
The Internet’s architecture rests on a set of design principles. Some are formal and documented in RFCs. Others are cultural norms that guide how the IETF community works. Here’s what we found across 1,375 analyzed sessions:
| Principle | Sessions | Description |
|---|---|---|
| Principle of Constant Change | 565 | The Internet must be designed to evolve; permanence is not the goal |
| Fate Sharing | 544 | State should be kept at endpoints that have a stake in it |
| Layering | 456 | Organize complexity through well-defined protocol layers |
| Simplicity | 418 | Prefer simple designs that can be understood, implemented, and debugged |
| Rough Consensus and Running Code | 331 | Progress through working implementations and approximate agreement |
| Trust | 252 | Trust is an architectural decision with cascading consequences |
| Connectivity | 165 | Universal connectivity is the Internet’s primary goal |
| Robustness / Postel’s Law | 80 | Be conservative in what you send, liberal in what you accept |
| Protocol Ossification | 71 | Protocols calcify when middleboxes depend on implementation details |
| End-to-End Principle | 56 | Intelligence belongs at the endpoints, not in the network |
| Tussle in Cyberspace | 50 | Design for ongoing conflict between stakeholders with different goals |
| Principle of Least Surprise | 4 | Protocols should behave as users and implementers expect |
This analysis was produced as part of the IETF 125 Hackathon (March 2026, Shenzhen). It demonstrates how vCons (virtualized conversations) can be used not just to store conversation records, but to analyze them at scale and produce derived insights.
Source Data — 2,249 vCons from the vcon-dev/ietf-meeting-vcons repository, each representing a recorded IETF working group session with metadata and auto-generated transcripts.
Keyword Detection — Each vCon transcript was scanned for mentions of 12 network design principles using pattern matching against principle-specific vocabularies.
Derived vCons — For each session where principles were detected, an appended vCon was created following the vCon core specification. These derived vCons reference the original via UUID and content hash, adding analysis metadata without modifying the source.
Group vCons — Per-principle group vCons aggregate all sessions discussing each principle, creating corpus-level views with collected evidence and statistics.
Narrative Reports — Claude was used to synthesize the evidence from each group vCon into the narrative reports you’re reading now.
Every step of this pipeline produces spec-compliant vCons, demonstrating the format’s utility for conversation analytics at scale.
A vCon is a standardized JSON container for conversation data — recordings, transcripts, metadata, and analysis. The IETF vCon working group is developing this as an Internet standard. This project showcases three types of derived vCons:
This site is generated from vCon analysis. Source code and data are available in the ietf-hackathon-125 repository.