Disagreement Register¶

Standing register — instantiates Consensus Convergence

Keeps a live, classified ledger of every open disagreement — factual, interpretive, preference, risk, authority — so the group always knows exactly what still divides it.

A Disagreement Register is the artifact that makes divergence explicit and keeps it that way. It is a running ledger in which every unresolved disagreement is logged as its own item, tagged by type — is this a dispute about facts, about how to interpret them, about preferences, about risk tolerance, or about who has the authority to decide? — and tracked from open to resolved. Its distinctive move is classification: by separating a factual dispute (resolvable with evidence) from a preference clash (resolvable only by a decision rule) from an authority question (resolvable only by someone with standing), it tells the group what kind of convergence work each open item needs, instead of arguing about all of them the same fruitless way. And because it is a live list, a still-populated register is itself proof that apparent agreement is not yet real.

Example¶

A city transit agency and three neighborhood coalitions are converging on a bus-rapid-transit corridor, and the meetings keep looping because everything blurs together — ridership models, curb-parking losses, and who gets the final say all get argued at once. A planner starts a disagreement register.

Each open item gets a row and a type. "Projected 2035 ridership" is tagged factual — resolvable by rerunning the model with agreed inputs. "Whether losing 40 parking spaces is acceptable" is tagged preference/risk — no amount of data settles it; it needs a decision rule. "Whether the coalitions can veto station placement" is tagged authority — it isn't about the corridor at all, but about who decides, and it goes to the agency's board. Sorting the argument this way, the group discovers that two of its five "disputes" were the same factual question in different clothing, and that the parking fight had been stalling because people kept throwing ridership data at what was really a values disagreement. The register didn't resolve anything by itself; it told everyone which tool each fight actually needed.

How it works¶

The register earns its keep through structure, not content:

One row per disagreement, with a type tag. Each item is named and classified — factual, interpretive, preference, risk, or authority — because the type determines the remedy, and mixing types is what makes deliberation loop.
Status is tracked to closure. Every item carries a state (open, in progress, resolved, parked) and, ideally, what would resolve it. Agreement is only real when the material rows are closed.
It is maintained live, not written once. The register is updated every session, so the group is always looking at the current divergence rather than a stale snapshot or a hopeful memory.

Tuning parameters¶

Classification scheme — how many disagreement types and how fine the tags. More types route remedies more precisely but add overhead and edge-case fights about which bucket an item belongs in.
Materiality threshold — which disagreements are worth logging versus which are noise. Too low and the register drowns in trivia; too high and a real fault line goes unrecorded.
Resolution-criterion field — whether each row must state what would settle it. Requiring it turns the register from a complaint list into a work list, at the cost of more effort per entry.
Ownership — whether items have named owners responsible for driving them to closure. Owners create accountability but can turn shared problems into someone else's problem.

When it helps, and when it misleads¶

The register's strength is that it refuses to let disagreement go invisible: it converts a vague sense that "we're not really aligned" into a specific, typed, trackable list, and it doubles as a false-convergence check — you cannot claim consensus while material rows remain open.^[1] Its failure mode is that a register records disagreement without resolving it; a fat, stale list can substitute the feeling of rigor for actual convergence work, and precise-looking rows can lend false weight to disputes that don't matter. It can also be weaponized — kept open on a trivial item to stall a decision someone dislikes. The discipline is to tie each row to a resolution path and a decision rule, review the register for staleness, and let the materiality of what remains, not its length, decide whether the group can proceed.

How it implements the components¶

The register is the archetype's divergence-tracking artifact:

disagreement_map — it is the disagreement map made concrete: the live, typed inventory of what still divides the group and why, which no other mechanism in the set produces.
false_convergence_check — because it holds the open items, it is the standing test of genuine agreement: an unresolved material row is direct evidence that a claimed consensus is premature.

It does not decide when the remaining differences are small enough to proceed (closure_rule — Consensus Vote and Consent Decision Rule), combine the resolved positions into a statement (synthesis_frame — Consensus Workshop), or supply the evidence that settles the factual rows (evidence_standard — Evidence Briefing Packet).

Instantiates: Consensus Convergence — the register is where the archetype's divergence is held and worked.
Consumes: the factual rows are settled by the evidence in an Evidence Briefing Packet; the preference rows route to a Consensus Vote or Consent Decision Rule.
Sibling mechanisms: Minority Statement Protocol · Consensus Vote · Consensus Workshop · Consent Decision Rule · Delphi Process · Evidence Briefing Packet · Facilitated Decision Meeting · Scientific Consensus Process · Stakeholder Alignment Session · Standards Committee Process

References¶

[1] The false-consensus effect (Ross, Greene, and House): people overestimate how widely their own views are shared. A register that forces disagreements to be named and typed is a direct counter — it makes the group prove agreement item by item rather than assume it. ↩