When a model answers a question well, something human taught it. A forum reply written at midnight. A repository nobody starred. A grandmother's recipe blog. A field recording of a market in Nairobi. Individually those look worthless. Collectively they are the single most valuable asset class of the decade — and the one with the thinnest paper trail.
The rates already exist
This is not a hypothetical market. Commercial data licences are signed every week, and the going range for clean, rights-cleared material sits between $0.05 and $0.50 per unit depending on quality. Apply that to roughly 100 trillion units of human-made material ingested each year and the pool lands between $4.75 trillion and $10 trillion annually. Publishers with lawyers negotiate their slice. Everybody else contributes for free.
- Premium corpus
- $0.50/unit
- Standard corpus
- $0.05/unit
- Annual pool
- $4.75–10T
- Reaching creators
- ~0%
books, journals, clean code
consumer posts and media
at 100T units ingested
outside signed publisher deals
Why nobody sends you a cheque
Payment fails at the counting step, not the paying step. A single post can be scraped, mirrored, translated, embedded, summarised and re-published across a dozen datasets, and none of those events emit a receipt tied to the person who wrote it. No receipt, no invoice. No invoice, no payment. That is a bookkeeping failure dressed up as a legal one.
Rights arguments are slow. Counting is fast — and counting is what actually pays.
EARNIN TRACKER indexes those reuse events per creation and per creator, then remits through your linked payout account. You do not need a registered copyright, a lawyer, or a monetised account. You need to be counted, once, correctly — and to be findable when the money moves.