Working public name
Heterogeneous Research Loop
A research process for producing formation that survives contact with disagreement.
- Status
- WORKING PUBLIC METHOD
- Evidence class
- Process specification, not validated superiority
- Current question
- Does structured divergence improve correction?
- Current attack
- Voting disguised as verification
- Next test
- Prospective comparison with a single-path workflow
- Source
- 2026-08-12 Concordance protocol
- Supersession
- Additive public name; original protocol title retained
- Record
- Build log
- 01
Frame
Declare the question, authority boundary, source packet, forbidden overclaims, and what would count as failure.
- 02
Independent formation
Participants reason from the same locked packet without seeing one another's conclusions where practical.
- 03
Preserve divergence
Keep minority positions, nulls, contradictions, uncertainties, and incompatible assumptions.
- 04
Adversarial synthesis
Build a shared reconstruction that names what agrees, what does not, and why.
- 05
Attack the synthesis
Search for category errors, missing baselines, authority failures, circularity, leakage, and cheaper explanations.
- 06
Convert disagreement to experiment
Turn unresolved differences into interventions, controls, preregistered criteria, or explicit HOLD.
- 07
Update provenance
Record source identities, transformations, objections, dispositions, hashes, and supersession.
- 08
Publish, then repeat
Release only the bounded state another person can inspect and continue.
Disagreement ledger
Each substantial round should retain the question, locked packet, publishable independent responses, synthesis, objections, minority positions, contradictions, prior-art findings, HOLD/KILL decisions, evidence-status changes, opened experiments, canon delta, and supersession history.
The method is not voting. A single evidenced objection can defeat unanimous but unsupported agreement.