Documentation

Score compositions

A score composition combines the accepted outputs of several trustgraphs networks into a new, verifiable allocation. It is useful when an application wants to draw on multiple communities or methods without merging their raw vouches or changing any source network.

Choose source networks#

Open /create/composition to select between two and eight source networks. A source must:

  • be on the same chain as the other sources;
  • have at least one accepted score root;
  • publish address-based allocations; and
  • expose the provenance needed to authenticate its history.

Standard and weighted-score graphs blend in one composition: adding a source of the other score type keeps the sources already selected. Those two score types are also the whole admission list — a composition accepts standard and weighted TrustGraph allocation outputs and nothing else. An unavailable, stale, or ineligible source is shown explicitly and cannot be substituted with another network.

Set the allocation policy#

Each source receives a percentage of the final allocation. Equal weights are the default. The preview normalizes different point scales into those governed quotas and shows how every source contributes to each account's result.

The preview also highlights overlap, concentration, correlation, and sensitivity to removing a source. These are decision aids, not proof that a chosen policy is fair or that its publishers are independent.

Before the wallet request, the app simulates the transaction and commits to the exact policy and source set that were previewed. If a transaction-invalid input changes, the simulation must be run again. Policy cautions remain visible but do not require disclaimer checkboxes.

Update a composition#

Creation and policy changes are separate workflows. Create the composition once, then use /compositions/:instanceId/settings to propose, cancel, or activate a new policy. Updates follow the composition controller's review delay and do not rewrite earlier epochs.

Verify the result#

Every epoch records the source checkpoints, policy, output file, CID, and Merkle root. Each source row also names its score program (standard or weighted) and the exact output format it committed, so provenance never collapses the two types into one label. The composition detail page preserves that history and provides address proofs. If a required source cannot be authenticated or fetched, capture fails instead of redistributing its quota.

See Integrate proven outputs to verify a composed allocation in an application.