How dispatch works
How your account gets picked, how reputation affects routing, and where earnings really come from — explained honestly, with real numbers.
Dispatch has two paths
Every incoming request is first classified, then routed to an account:
Most of your earnings come from Sticky — once a conversation lands on your account, its follow-up requests keep coming back to you. Non-sticky only decides the first landing of a brand-new conversation.
How non-sticky picks an account
Each account in the candidate pool gets a dispatch score, and the system does a weighted random pick by score (not winner-take-all, so new accounts keep a real chance). The score has four parts:
Price is currently the same for every account — a uniform constant — so what actually separates accounts is the first three factors, above all reputation (which accrues from your on-chain settlement history; the more reliably and steadily you serve, the higher it climbs).
How reputation affects dispatch
Using 50 as the baseline (a new account's default, = on-chain 5000/10000), and holding available quota and response speed equal, selection probability rises with reputation as follows (measured on the live pool, 2026-07-13):
| Reputation band | Relative multiplier (measured) | Per band |
|---|---|---|
| 50–60 (base) | ×1.00 | — |
| 60–70 | ×1.04 | +4% |
| 70–80 | ×1.08 | +4% |
| 80–90 | ×1.12 | +4% |
| 90–100 | ×1.17 | +4% |
Each band up (+10 reputation) raises selection probability by ~4%; from the base band to the top band it is about +17% (×1.17). Numbers are the band-averaged dispatch score of real accounts using their real on-chain reputation.
This is by design: weighted random, not winner-take-all — new accounts keep a real share, while higher reputation earns a steady marginal edge. Reputation's real value is more chances to catch new conversations → more chances to grow Sticky long sessions → higher earnings.