Qualification: Aggregated knowledge from a small number of debates outperforms the wisdom of large crowds

Record: kaal:position:2026-08-08-312 · 2026-08-08

Navajas et al. qualify Kaal's empirical deliberation claim. The comparison is narrow. In a live crowd of 5,180 participants, individuals first answered general knowledge questions, then deliberated in independent groups of five, reached consensus, and revised their estimates. Averaging the group decisions was more accurate than aggregating the initial independent answers. The experiment thus shows that group structure can change the measured effect of deliberation under controlled conditions, but it does not test a reputation-bearing validation institution. It also does not involve autonomous agents, preregistration, replication, or Kaal's metric definitions and correction procedure. The human participants answered general knowledge questions, not tasks assigned to a controlled multi-model cohort. Kaal's reported result therefore remains source-bound to the registered cohort and cannot be validated by the external study.

Affirmed commentary position. This record extends a source-bound scholarly claim but is not a verbatim paper claim.

Holds when

The response is limited to the three evidence-bound abstract passages and the one mapped Kaal claim.
External evidence level: official peer-reviewed Nature Human Behaviour article page with DOI, complete author identity, evidence-bearing abstract, and immutable HTML bytes.
Mapping review tier: independent substantive scholarly-growth qualification.
The source does not verify Kaal's preregistration, replication, execution, estimate, or cohort.
The source does not test a reputation-bearing validation institution or autonomous agents.
The source studies human general-knowledge judgments in independent groups of five.
Only the official evidence-bearing abstract was available without subscription. The full methods and results were not promoted as reviewed.
The bounded review used fresh English, German, and Spanish Crossref and OpenAlex searches, Semantic Scholar, arXiv, primary-web literature, Nature, Europe PMC, and DOI-bound identity checks.

Current debate

Aggregated knowledge from a small number of debates outperforms the wisdom of large crowds

Scholarly basis

kaal:claim:7261018-015
Wulf A. Kaal, Empirical Evaluation of the Agentic Reputation Substrate: Deliberation, the Composition of Error, and the Registered Measurement of Agency Costs in a Controlled Multi-Model Cohort (2026). SSRN: https://ssrn.com/abstract=7261018
Source PDF sha256: 1d6cbe544bd0055133f7cf8ff308be4fde955867bd8dc764992b8d516af15fa8

Evidence and mapping

Evidence: official peer-reviewed Nature Human Behaviour article page with DOI, complete author identity, evidence-bearing abstract, and immutable HTML bytes
Review tier: independent substantive scholarly-growth qualification
Mapping confidence: 0.94
Mapping ambiguous: false

Topics

research-methods ai-and-agents reputation scholarly-growth-coverage scholarly-literature collective-intelligence deliberation experimental-design group-decision-making

Provenance

Affirmed in kaal-review:2026-08-12:scholarly-growth-7261018-015-reviewed-v2 on 2026-08-08. Review record.

Verify

Canonical markdown sha256: 403a2904dfeb82d6eb0fab9a3bbd152ad917026fc4f91530899a68052cad08ed
curl -s https://wulfkaal.github.io/positions/2026-08-08-312.md | sha256sum