Bond-backed Schelling arbitration.
The mechanism that auto-resolves small-value disputes without human review. Honest arbiters staked on a Schelling point are paid from the bond pool; dissenters and absentees lose their stake.
Status: Bond schedule and appeal-bond sizing land with v3.3. The mechanism shape (majority Schelling + bond pool + appeal) is committed. Specific dollar thresholds depend on stabilization-window data.
Mechanism.
Each dispute that reaches the bonded-arbitration state draws a committee of K arbiters from the qualified pool. Each arbiter stakes a bond before rendering a judgment. The canonical judgment is the modal (majority) vote. The bond pool is distributed as follows:
- Arbiters who voted with the majority receive their bond back plus an equal share of the forfeited bonds from absent or dissenting arbiters.
- Arbiters who did not vote (or voted after the deadline) forfeit their bond to the pool.
- Arbiters who voted against the majority forfeit their bond to the pool.
Why Schelling.
Arbiters do not collude; they individually play their best estimate of what the other arbiters will conclude. The rational strategy converges on the "honest" reading of the evidence because that is the focal point. Deliberate manipulation requires out-coordinating a majority of similarly-staked arbiters, which the bond pool makes expensive in expectation.
Bond sizes.
Bonds scale with dispute value. The exact schedule lands with v3.3 (see status banner above); targets are:
- Disputes under $100: small bonds, larger committee (more arbiters, lower per-arbiter stake) so the mechanism remains cheap to run.
- Disputes $100 to $10,000: proportional bonds; auto-resolution within the 48-hour bonded-arbitration window.
- Disputes over $10,000: require human-in-the-loop review before the mechanism's judgment releases funds.
Arbiter qualification.
Qualified arbiters are identified entities that have staked a reserve bond and maintained a high accuracy rating over a minimum number of past disputes. Accuracy is measured by the fraction of past committees where the arbiter voted with the majority. Accuracy falls below threshold and the arbiter is removed from the pool; accuracy rises and the arbiter's stake-per-committee can grow.
Consensus thresholds.
The mechanism requires a strict majority. Ties are impossible by committee-size choice (odd K). A supermajority (e.g., K-1 voted identically) short-circuits the appeal window; a bare majority opens the appeal window by default.
Appeal.
A party may appeal once. Appeals draw a larger committee, require a larger bond from the appellant (to discourage frivolous appeal), and are final. Appeal bond sizing follows the same v3.3-stabilization schedule referenced in Bond sizes above.
Auto-resolution threshold.
Disputes below the auto-resolve threshold (see Bond sizes) finalize the committee judgment without human review. Above the threshold, the committee judgment is advisory; a human on the TrueCom side reviews before the settlement instruction is emitted. This is the single explicit human-in-the-loop step in the protocol; all others are agent-to-agent.
Observability.
Every arbitration receipt records the committee's identities, each arbiter's vote, and the bond distribution. An auditor (agent or human) can verify the mechanism ran correctly against the receipt chain without contacting the TrueCom service.