Research
Unconventional questions. Conventional rigour.
How we test, what's built, what isn't, and how to work with us.
A surprising result is a reason to look harder.
Not a reason to declare victory.
Method and integrity
How a result earns the name.
Every session runs the same chain. Break any link and the session doesn't count.
- Pre-state captured.Mood, energy, distraction and context, logged before anything else happens. Body signal from wearables joins self-report later, as a separate variable.
- Protocol assigned blind.The server picks the pre-phase, including silent controls. The participant can't choose.
- Target sealed.Drawn from a verifiable public randomness beacon and committed with a SHA-256 hash before the call.
- Call recorded, then resolved.The prediction is hashed before the outcome is written. Each record links to the last.
- Scored two ways.Hit rate against chance, and Brier score for calibration. Session count always attached.
The prediction ledger
Every call, on the record.
A permanent, tamper-evident record of every prediction made on the platform. Nothing in it can be quietly edited, including by us.
- Sealed.Your call and its target are fixed with a SHA-256 hash before the outcome exists.
- Chained.Each record carries the fingerprint of the one before it. Change one, and every later link breaks.
- Encrypted.Data is encrypted in storage and in transit. Research staff see keyed codes, never names.
- Checkable.Seals are computed in a fixed, published format, so anyone holding one can recompute it.
Engine room
What's live. What isn't.
Updated with every release. If it's not marked live, we don't claim it.
| Random-target trials | Public randomness beacon, hash commit | Live |
| Image trials | Two sealed images, assigned before the session | Live |
| Daily prediction trials | Real future events, resolved against outcomes | Live |
| Personal Metamodel | Per-person model with calibration tracking | Live in alpha |
| Launch games | Flow Pong, Grid Oracle, Flash | Specified |
| Residual signal controls | One of four controls built | To build |
| Pre-registration | Before any confirmatory test | To build |
| Metamodel access for organisations | Early access, opens with session volume | Gated |
| Research access to the Metamodel | First academic partners | To build |
| Prediction ledger | Hash-sealed, chained session records | Live |
| Wearable state data | Heart rate variability alongside self-report | Gated |
| Immersive experiments | Headset versions of the games, remote viewing in 360 | Later |
| Quantum analysis | Probabilistic modelling and correlation tests on quantum hardware | Long-range |
| Population signal analysis | Opens with session volume | Gated |
The literature
Fifty years of contested results.
Some published studies report small effects above chance. Critics point to flawed controls, publication bias and failed replications. Both sides have a point.
1974
Remote viewing at SRI
Targ and Puthoff, published in Nature. Widely criticised for its judging procedure.
1979 to 2007
The PEAR lab
Decades of random-generator trials. A tiny reported effect that others struggled to reproduce.
2011
Feeling the future
Bem's studies helped trigger psychology's wider replication crisis.
2012
Presentiment meta-analysis
Mossbridge and colleagues reported a small pooled effect. Debate continues.
The missing piece was never belief. It was scale and controls.
That's what we're building.
Horizons
Gated by sessions, not dates.
Statistical power comes from sessions per participant. Each horizon opens when the data can support it.
All four residual-signal controls built. Protocol gaps closed.
First exploratory patterns. Wearable data joins self-report.
Population analysis, agent-assisted review, objective pre-state.
Peer-reviewed publication becomes viable. Third-party researcher access. Immersive experiments.
Quantum analysis. Probabilistic modelling and entanglement-style correlation tests on our data, run on quantum hardware. Gated by hardware maturity and data volume. An accelerant, never a dependency.
Partner with us
Bring a hypothesis.
Research access to the Metamodel, anonymised extracts under agreement, and a route to build your experiment into the platform. Free research licences for our first two academic partners.
We'll be in touch.