Methodology
Auditable definitions that keep the platform and slides in the same statistical universe.
1. Units of analysis
The platform separates document names from proposal lineages. A draft name is an identifiable administrative record. A lineage is a component connected by unique Datatracker replaces relations and may combine an individual draft, its WG/RG adoption, and successors. Volume is measured with names; pathways and origin are measured with lineages.
2. Populations and denominators
The full catalog (35,072 unique names) is used only for scale. The 31,919 documents with a dated first submission support observed outcomes. Conversion and association analyses use the mature 2005–2018 submission cohort (22,098 documents), allowing comparable follow-up time. Pathways use 25,410 dated lineages formed from 6,775 unique replaces relations.
3. Outcomes
The hierarchy has five states: published RFC, replaced, active, expired, and discontinued. Replacement is a transition between names, not failure. Expiration is an administrative state, not rejection. Active documents and recent cohorts are right-censored and must not be treated as definitive failure.
4. Time to RFC
The canonical duration is the RFC Index publication date minus the first submission of the RFC-linked draft. The comparable summary uses 2009–2024 publications; 2005–2008 are omitted to reduce left truncation. In that window, n=4,140, median=2.08 years, and mean=2.57 years.
5. Stream, area, and origin
Stream is the document-name classification, with a prefix fallback when needed. It is not the proposal origin. For stream or area comparisons, a lineage enters in the year of its first occurrence in that stream; legacy app/rai labels are harmonized as Applications & Real-Time and Transport as Web & Transport.
6. Geography
The map uses the identifiable country of the first author, normalized to ISO2 from the country field and, as support, the email ccTLD. Coverage is a floor: missing countries remain missing and are not imputed. Generic-use ccTLDs are not treated as residence. Results are aggregate and do not describe individual merit.
7. Affiliations and organizations
Organizational diversity counts distinct author affiliations. In the mature cohort, 1,485 of 22,098 documents lack a known count (1,478 recorded as zero and 7 missing); they are excluded from comparisons by number of organizations rather than classified as one organization. Observed differences are descriptive associations.
8. Reproducibility and causal caution
Raw IETF Datatracker API collection is separated from processing. The pipeline produces columnar tables, a per-draft base, and an audited premises summary published with each refresh. Revisions, presentation, leadership, and collaboration are descriptive associations; they do not establish causal effects.
Source: IETF Datatracker and RFC Index. Analytical snapshot: 20 July 2026; collection date is reported separately.