Two-way
Share Data, Get Data
Published counties are free to anyone. Credits are for the rest: ad-free downloads, unopened community uploads, and the fresh vintages that are not given away. You earn them by adding counties nobody else has.
What a contribution is worth
| Event | Credits | Why |
|---|---|---|
| Create an account | +5 | Enough to open something and see how it works. |
| Upload passes the automated checks | +10 | Containment, schema and vintage checks all clear. |
| Upload verified against a reference copy | +40 | Machine-comparable to a known-good version of the county. |
| Upload verified by review | +75 | No reference existed, so a human confirmed it. The most valuable case. |
| Upload fills a bountied county | ×2 – ×5 | Multiplies the award above. See the board. |
| Ad-free download of a free dataset | −3 | The data stays free; skipping the funnel is what costs. |
| Open an unverified community upload | −15 | Honest uncertainty, priced accordingly. |
| Redeem a premium county | −250 | A current vintage that is not part of the free release. |
| Your endorsement is later confirmed | ×2 the spend | Paid only when a reviewer independently agrees. |
| Your report of bad data is later confirmed | ×1.5 the spend | Same rule, other direction. |
Why endorsements pay only when confirmed
Because that is the only design where telling the truth wins. A reflexive thumbs-up pays nothing, so there is no reason to give one. A spiteful report pays nothing either, so there is no reason to file one. The payout lands only when a reviewer independently reaches the same verdict you did — which makes every downloader a second, genuinely independent verification signal rather than a rubber stamp.
Your accuracy record is public and it weights how much your verdict counts. Volume is not the leaderboard; being right is.
What happens to an upload
- 01
Intake
Zipped shapefile, GeoPackage, GeoJSON, FileGDB or Parquet. Archives are unpacked under size and path constraints before anything reads them.
- 02
Structure
Polygonal geometry, a declared CRS, and a feature count in a plausible band for the county.
- 03
Containment
Centroids are tested against the county's TIGER boundary. This is the check that catches one county's data labelled as another's.
- 04
Attributes
Every field is profiled — type, fill rate, distinct count, examples. Placeholder text is counted as empty, because a column full of n/a is an empty column.
- 05
Vintage
Date-like fields are read for their upper envelope: a dataset cannot be older than its newest recorded sale. A large gap from the claimed vintage is a flag.
- 06
Routing
Clean and comparable to a reference copy publishes immediately. Clean with no reference publishes as an unverified upload with the uncertainty stated. Contradictory goes to review.