# Standards Win When Re-Instrumentation Is the Expensive Part ## Claim A neutral standard succeeds in domains where switching vendors would otherwise require rewriting integration code, because the standard's value is the migration it prevents rather than the capability it adds. <!-- atomicity: 1 idea ✓ --> ## Scope & conditions - Holds where integration is deeply embedded and switching is otherwise costly — instrumentation, authentication, data formats. - Breaks where switching is already cheap, in which case a standard adds ceremony without removing pain, and breaks where one vendor's capability advantage exceeds the switching cost entirely. - The account is retrospective and self-reported by the standard's own governance committee — treat it as a hypothesis about mechanism, not a proven causal account, until tested against a standard that failed. ## Evidence OpenTelemetry's own Governance Committee attributes its graduation (CNCF's highest maturity tier, reportedly its second-largest project after Kubernetes) to three factors: users refusing rip-and-replace when switching observability vendors, unified signals rather than siloed logs/metrics/traces, and a welcoming contributor community that compounds. The talk now extends the same argument to agentic AI, pitching GenAI semantic conventions as the standard to adopt before proprietary tooling fragments the space. ## Links - [[AI Agent Safety MOC]] — the condition under which a neutral standard actually displaces vendor lock-in - [[3. Resources/Permanent Notes/Cheap Adaptation Beats Good Prediction When Standards Multiply]] — because contrasts; that note argues the winning move under standard proliferation is adaptability itself rather than betting on one standard, this note argues a specific standard wins for a specific structural reason — worth reconciling which applies when - [[3. Resources/Permanent Notes/Lock-In Removes the Discipline That Would Punish Decay]] — because contrasts; a neutral standard is explicitly the mechanism that keeps switching cost low and vendors honest, the inverse of the lock-in this note describes - [[3. Resources/MOCs/Open Source Governance MOC]] — because applies-in; standard-adoption mechanics are core open source governance territory ## Tests / falsifiers - Test the three-factor account against a standard that failed to gain adoption despite addressing genuine re-instrumentation cost. If failed standards show the same three factors present, the account is incomplete. - If agentic AI observability tooling fragments into proprietary formats despite GenAI semantic conventions existing, that specifically refutes the extension of this claim to the agentic domain, independent of whether the original OpenTelemetry case holds. ## Maintenance - Last tended: 2026-08-16 - Next review: 2026-11-16 ## Source - [[3. Resources/Readwise/Full Document Contents/Articles/OpenTelemetry Celebrates Graduation and the Next Era of Agentic - Alolita Sharma & Ted Young]] — KubeCon Japan 2026 keynote, Governance Committee members Alolita Sharma and Ted Young.