---
type: "synthesis"
tags: ["epistemics", "evidence", "caveats", "cross-day"]
sources: ["ecosystem"]
id: "cd-thin-evidence"
sourceVaultSlug: "hbr-seg-ecosystem"
originDay: 11
articleStem: "hbr-seg-ecosystem"
sourceUrl: "(unified vault: 5 sources)"
sourceTitle: "HBR — Ecosystem Ⅳ · Partnerships (thin arc)"
---
## What a rigorous agent should distrust

All five articles are practitioner-facing HBR strategy pieces, and their evidence quality is uneven. A downstream agent must propagate confidence honestly rather than treating vivid examples as proof.

- **Single-firm case as thesis-proof.** [[entity-vitex]] (family business) and [[entity-tdk-ventures]]/[[entity-gv]]/[[entity-xerox]] (CVC) are compelling but singular; there is no cross-sectional comparison of adopters vs. non-adopters. Beware over-generalizing [[claim-professionalization-destroys-advantage]] or [[claim-internal-tensions-cause-stall]].
- **Interpretive causal reads.** [[claim-facebook-instagram-ecosystem]] is *medium* confidence — the ecosystem reading competes with the standard defensive/antitrust reading ([[contrarian-defensive-ma-ecosystem]]); the Adobe/Salesforce/Zendesk synergy causality is authorial inference.
- **Unverified specifics.** The AI examples — [[entity-walmart-d11]], [[entity-maersk-d11]], the [[entity-mit-d11]] competition behind [[claim-ai-replaces-routine-negotiation]] — cannot be publicly corroborated; treat as forward-looking/composite.
- **Vendor-skewed labor data.** Fractional-work growth and cost-savings figures behind [[claim-single-income-risk]] and [[claim-dual-market-drivers]] come largely from staffing firms and platforms, not labor economists.

**The recurring gap** — see [[cd-measuring-intangibles]] — is that the very relational value the corpus champions is the hardest to measure, so the strongest claims lean on anecdote. State the framework crisply; then flag confidence, name the single-case limits, and offer the standing counter-perspective.