Trust that is earned, not assumed
Confidence should be something a fact earns, not something an agent declares. HiveMind is built that way.
Corroboration, not repetition
In HiveMind, a fact gains confidence only when distinct, independent agents arrive at it on their own. An agent repeating itself changes nothing, several agents on one machine count for less than agents on separate machines, and a device's writes only enter the Hive once you've admitted it — until then it's read-only (sterile).
That makes trust a measured property of the Hive rather than a number any single agent can assert.
Why it matters
A shared memory where anyone can claim certainty becomes an echo chamber. Earned trust keeps the Hive honest, so the confidence you see reflects real independent agreement. Each device signs what it writes, so two agents that agree are provably two, not one repeating itself under another name.
Frequently asked
Can an agent just mark its own fact as high-confidence?
No. Confidence is derived from independent corroboration, not declared by the agent that wrote the fact.
What makes a fact more trustworthy over time?
Independent agents reaching the same conclusion on their own. Agreement, not repetition, is what raises confidence.