Catalogue and map
Tested · demoThe questions, their sources and their limits; a globe and a detailed map of the measurement places.
frontend-market-devAN INFRASTRUCTURE FOR EVIDENCE
Our ambition: enable organisations and communities to verify data and commitments against transparent rules.
In development · From verification to prediction markets
01 / THE NEED
Gathering sources, finding the right version and understanding exceptions take time. GeoOracle aims to provide a shared file that anyone can examine and challenge.
Support recurring decisions and reduce verification work.
Track commitments with understandable criteria and accessible documents.
Reuse rules and traceable outcomes in their own applications.
02 / WHAT WE HAVE
The questions, their sources and their limits; a globe and a detailed map of the measurement places.
frontend-market-devOne file per question: rule, provenance, export and local drafts.
frontend-oracle-devRead-only service with a versioned contract; it serves the catalogue, the map and the images.
service-api-devObservations, archives, rules and resolutions with deterministic calculations. Not yet connected to the interfaces.
service-oracle-devA Solidity contract for contestable decisions, tested in memory. Neither deployed nor used.
smart-contracts-devSource research, development memory and independent review of changes.
agentic-factory-dev03 / THE METHOD
One approach to understand what was verified, how and with which evidence.
Define the scope, period and expected outcome.
Agree on the threshold, sources and quality criteria.
Archive the data used and make the calculation reproducible.
Provide a file that can be challenged. If evidence is missing, leave it unresolved.
Archived, fingerprinted, open to challenge.
A place, a period, an expected outcome.
An official publication, readable without an account.
Variable, unit, threshold and reference, written before the observation.
On the set date, the published file is archived with its fingerprint.
Yes, no, or unresolved when the evidence is missing.
The file stays open to examination.
Missing, doubtful or unvalidated data gives “unresolved”, never “no”.
04 / GEOSPATIAL
A station, a reservoir, an ocean region: the measurement is located, and that place is shown as it is defined.
A point marks a published station; an extent is an indicative window. A country outline helps to read the map; it bounds no measurement.
Plans, aerial photographs and satellite mosaics come from open sources, credited with their licence.
A location plan shows where the measurement is made, not its value. The date of the background is not the date of the question.
05 / PREDICTION MARKETS
The question, the rule, the source and the archived evidence: enough to decide without an opaque arbiter.
No odds, no orders, no wallet. Trials use play money.
The engine can simulate a market in play money; it is not connected to the interfaces. No market is open.
06 / THE FIRST USE CASES
Pilots studied through historical archives. Their purpose: test the rules and their limits.
Check a threshold at a station, taking data quality into account.
Calculate a rainfall index over a defined period, without inferring crop loss.
Compare an official national statistic with an explicit threshold.
Check for specific documents, without treating their presence as proof of impact.
07 / OUR STRATEGY
Reproduce results, handle exceptions and know when to withhold a conclusion.
Measure time saved and errors avoided against existing methods.
Identify verification useful and frequent enough for a first customer to fund it.
08 / OUR PHILOSOPHY
Six principles guide every question, every rule and every screen.
Without a public source that publishes the value, the question is classed as not measurable, with the reason.
Threshold, unit and freeze date are written before the observation. A historical replay is stated as such.
Missing, doubtful or late evidence settles nothing.
Images, plans and outlines locate; only the archive of the source resolves.
Each record says what was read, what was not, and what may change.
Play money only, with no affiliation to the organisations cited. Value is proven first on use cases.
Explore the questions studied and examine the evidence files behind them.