Seal a hunch before the event.
Log an impression before it happens, seal what would confirm it, record how likely it was anyway, name the channel it arrived through — then resolve later and let the math, not your memory, tell you which of your channels actually land.
Maybe it's noise. Maybe it isn't. This is how you'd actually find out.
What it measures
Rangefinder is a precognitive impression journal. It doesn't know whether precognition is real — neither do we, and that's exactly why it's worth measuring, honestly, for you. The instrument only checks one thing: does what you wrote before the event correspond to what happened after — scored against how likely the impression was anyway, so a lucky call on an obvious thing earns nothing and an honest long shot that lands means something.
How an entry works
- The impressionWrite it exactly as it arrived — a dream fragment, a flash, a feeling. Don't clean it up.
- Make it checkableSeal the confirmation — the thing that would count as a yes — and a check-by deadline.
- Honest priorHow likely was this anyway? An honest 10% makes a later yes interesting; a 90% hit proves nothing.
- The channelSingle-select, frozen at capture: dream · edge-of-sleep · waking flash · meditation · synchronicity.
Later, resolve each entry: hitpartialmissvoid
Why both halves are sealed
Precognition's one adversary is memory rewriting itself. So both the impression and its confirmation are sealed, immutable, in advance — that's the entire validity claim.
If you could edit either after the outcome, a hit would prove nothing; you'd just narrow the criterion to fit. The channel is a frozen fact of arrival — you can't half-dream something, or later re-attribute a lucky hit to a different channel.
How the score is made
Your own hit / partial / miss call is the only input to every number. From your record, Rangefinder builds:
- Lift — how much your outcomes beat your own prior, in points. The number no belief-app will show you.
- Brier score — how well your confidence tracked reality, where 0.25 is coin-flipping.
- Calibration verdict — a plain-language read that names where your confidence and your results drift apart.
- The Channels scoreboard — each channel's hit rate against every other, shown as a ratio, only once both sides have enough data. Labeled correlation, not proof. If it were real you'd expect an edge in some channels; if it's noise you'd expect flat rates. The scoreboard just shows you which.
The AI — a second read, not a verdict
After you resolve, an optional on-device model reads what you sealed against what actually happened and gives its own take on the overlap — capped by your own hit / partial / miss call, so it can never promote a miss to a hit.
It never touches your statistics; your own call is the only input to every number. A referee, not an oracle. The judge is enrichment, not authority — Pro, advisory, on supported devices.
Privacy
Rangefinder runs entirely on your device — no accounts, no analytics, no servers. What you record stays yours. Optional Pro iCloud sync goes through your own private iCloud, and nothing else.