A Reviewers Point of View Is Context Plus a Stance

# A Reviewer's Point of View Is Context Plus a Stance ## Claim A reviewer — human or AI — gives specific, actionable feedback only once it has both the domain context to recognize a stack's real failure modes and a stated stance committing it to name them, and either piece missing collapses feedback back into generic hedging. ## Evidence **Source**: [[3. Resources/Literature Notes/Using Code Review Personas in Claude Code]] ## Scope & conditions - Holds when: the reviewer (model or person) is asked to evaluate work in a specific domain and nothing has committed it to that domain's known failure modes ahead of time — a written persona, a stated role, or lived domain experience all count as supplying the "stance" half. - Breaks when: context alone is supplied without a stance (an encyclopedic but noncommittal reviewer still hedges), or a stance is asserted without real context (confident feedback grounded in nothing produces plausible-sounding but wrong specifics). ## Links - [[AI & the Future of Knowledge Work MOC]] — because applies-in: reviewer-persona design is part of this MOC's judgment/review-as-bottleneck thread — a persona is one way of supplying the "stance" half this note names. - [[3. Resources/Permanent Notes/Generated Accessibility Targets Do Not Transfer Verification Responsibility]] — because contrasts: that note is about a generated artifact's output quality; this one is about what makes a reviewer's *feedback* specific rather than generic, a different point in the same review pipeline. - [[3. Resources/Permanent Notes/Automated Review Is Only As Accountable As The Standards Behind It]] — because extends: standards give an automated reviewer its context; this note names the second, separate ingredient — a stance that commits the reviewer to actually applying them.