METHOD & CALIBRATION

The method is the product.

Every remote-viewing protocol shares three controls: blind tasking, a committed reveal, honest scoring. rv.studio enforces all three in software. Below: the full blinding model, what a target can be, and the statistics — including the ones that refuse to flatter you.

01 / LINEAGE
Where the discipline came from

Real history. Real critiques.

The modern discipline traces to Ingo Swann and physicists Russell Targ and Hal Puthoff at SRI International in the 1970s — two decades of research inside the U.S. government program later declassified as STARGATE, and the origin of Controlled Remote Viewing (CRV). Context, not a house method — rv.studio is method-agnostic.

The critiques are part of the record. The 1995 AIR review — Jessica Utts and Ray Hyman reaching opposite conclusions — informed STARGATE's closure; the 1974 SRI paper in Nature drew published methodology critiques. Both sit side by side in the library.

02 / PERCEPTION
Why we say remote perception

It is rarely a picture.

"Remote viewing" is the historical name and a known misnomer. What viewers report rarely arrives as a finished picture — it arrives as impressions across many channels: texture, temperature, sound, scale, emotional tone. The Princeton lab (PEAR) called its program Remote Perception for the same reason. We instrument every channel.

This is orthodox CRV, not a softening: Stage 2 is textures, temperatures, colors, smells, tastes, and sounds; later stages add size, shape, mass, and the site's emotional impact as first-class data. A picture is one output, not the medium.

The perceptual channels.

An impression can land in any of these. You record whatever arrives.

Visual

Form, color, light, contrast — the only channel the word "viewing" expects.

Auditory

Sounds, tones, voices, mechanical or natural noise.

Olfactory

Smells and scents at the site.

Gustatory

Tastes — often metallic, mineral, or organic.

Tactile / textural

Surface texture and feel — rough, smooth, gritty, wet.

Thermal

Temperature — touch, and the ambient warmth or chill of the place.

Kinesthetic / motion

Movement, force, flow, vibration — the body's sense of a scene.

Dimensional / spatial

Size, scale, shape, volume, mass; open versus enclosed.

Emotional / aesthetic

The site's affective tone — awe, calm, dread — recorded as data, not mood.

Not everyone "sees" — and that's not a miss

Viewers have different dominant channels. Some see; others hear it, feel it, or simply know it — conceptual knowing, no sensory carrier at all. Forcing every impression into a picture suppresses the channels a viewer is strongest in, so rv.studio records whatever arrives, in whatever modality.

Channel set drawn from CRV Stage 2 sensory data, Stage 3 dimensionals, and Stage 4 aesthetic/emotional impact.
03 / BLINDING
One session, end to end

Blindness is a ladder.

The classic double-blind keeps the viewer and the monitor blind. rv.studio extends the firewall to the analyst who judges the work and the coordinator who sets it — then measures how far up the ladder each session actually reached.

STEP 01
Task

A target is drawn from a pool and sealed behind a coordinate — a content-free reference tag.

STEP 02
Blind session

You work the coordinate in your own protocol — sketches, descriptors, stages — seeing nothing else.

STEP 03
Record

Impressions are committed and timestamped before any reveal.

STEP 04
Reveal

The target is disclosed.

TARGET
STEP 05
Judge

A separate, blind analyst — human or AI — scores against the target.

STEP 06
Calibrate

The result rolls into your calibration record — in its own tier.

The role ladder.

Each role sees only what its job requires. The server enforces the boundary — hiding something in the interface never counts.

Viewer

Blind to: the target

Works only a content-free coordinate — sketches and descriptors against a target that stays sealed until impressions are committed, so the work can never reason toward a known answer. In associative remote viewing (ARV) — each outcome of a real-world question paired with a different image — the viewer is also blind to the pairing.

Monitor / interviewer

Blind to: the target

Guides the session in live two-way contact with the viewer — the biggest channel for cueing — so the monitor is blind too. Viewer plus blind monitor is the classic double-blind: no one in the room knows the target.

Analyst / judge

Blind to: the question & outcome

A different person from the viewer and coordinator — blind to the question, its resolution, and (in ARV) the pairing — judging only which target the impressions fit better. Scores lock before any outcome is revealed.

Coordinator / session-runner

Blind to: the target — as far as possible

Runs the session and must not be able to cue it. The server draws the targets and seals the image–outcome pairing — no one, not even the question's author, knows which image stands for which outcome. The question is always written by a human tasker outside the viewing party and kept from the session-runner until the outcome is known. The top rung: every role a separate person.

The session earns its tier.

Blind means the viewer has no information about the target. The tiers count how far up the ladder the firewall reaches. The platform doesn't take your word for it: each session's tier is derived from a who-saw-what ledger.

Tier
Who is blind
Counts toward your verified accuracy?
SoloViewer only
The viewer is blind; you score your own work. Good practice — the most bias-prone configuration.
No — practice only, tracked separately
MonitoredViewer + monitor
A blind monitor runs the session; an independent analyst judges it.
Tracked as its own tier, shown alongside
Double-blindViewer + monitor + analyst
Everyone with the viewer is blind, and a separate analyst judges blind to the question and outcome — the strongest control against cueing and judge bias.
Yes — counts toward your verified accuracy
Triple-blind+ externally-tasked, coordinator blind
A blind monitor, a blind analyst, and a coordinator blind to the question — written outside the viewing party and kept from the session-runner until the outcome is known. With the pairing sealed, no one present knows the question or what any image stands for. A hit here can't be explained by cueing, leakage, or a judge who knew the answer.
Yes — counts toward your verified accuracy
Calibration is isolated per tier

The honesty guarantee: your verified accuracy — hit-rate, Brier score, hit-rate over base rate — counts only double- and triple-blind sessions. Solo and self-scored work reports in its own bucket and can never inflate it. And since the tier comes from the ledger, a session run without a separate analyst is downgraded automatically — the claim can never outrun what happened.

Every tier keeps its own count, hit-rate, Brier, and base rate — less-blind tiers sit beside your verified accuracy, clearly labeled.

Integrity markers.

Two records make the blindness auditable rather than assumed.

The blind ledger

Append-only: who saw what, when, in which role — written at every step, never edited or removed. It stores no secrets, only proof of what each party could see. The achieved tier is computed from it.

The compliance certificate

A permanent record, written at resolve, of the controls actually met — separate analyst, blind monitor, sealed pairing — and the tier earned. The receipt that the session ran at the tier it claims.

None of this proves remote viewing works — it is rigor, not a claim. What it guarantees is simpler: whatever your numbers turn out to be, they are not an artifact of who knew what.

04 / TASKING
What you can be sent, and where you can work it

A target isn't always a picture.

Most practice runs against images — thousands of curated ones, or pools you build yourself. But a target only has to be something sealed that you know nothing about, and some of the best-documented operational work was never visual at all.

Image pools

Curated collections and your own uploads, drawn server-side and sealed behind a coordinate. Build and curate a pool without ever seeing what's in it — the interface keeps your own pool blind to you.

Written taskings

A written question in place of a picture. You write the questions yourself, so the set is no secret — but you draw one at random, so you never know which one you're working. The pool is the envelope. Eight curated collections ship with the studio, and you can seed your own envelope from any of them.

Four ways a written tasking ends.

A question has no decoys and no ground truth at draw time, so nothing here is scored. At the reveal you answer one thing — did my reading answer the question? — and the envelope reacts.

It answered the question

Your reading covered what was asked. Out of the envelope and into your journal — it won't come up again.

It answered part of the question

Some of it landed and some of it is still open. The question is resealed and the reading is kept, so the next pass starts from what you already have.

It didn't answer the question

Nothing in the reading speaks to what was asked. The question goes back in the envelope and this reading isn't carried forward.

I don't want this question

Nothing to do with the reading — you would rather not meet this one again. It stops coming up in your draws, everyone else's envelope is untouched, and you can put it back whenever you like.

Past, present, or something that hasn't happened yet.

Envelopes state which way they point, and the studio says so before you draw and again at the reveal.

This one hasn't settled yet

Future work generally comes through less cleanly than historical work, and a reading can land on a version of events other than the one that arrives. That's the working expectation among people who do this rather than a settled fact, and we don't try to explain why here.

Judge whether your impressions answered the question — not whether the future has agreed with them.
Written taskings are not scored — on purpose

A question has no decoys, so there is no chance baseline to beat. That means no hit-rate, no Margin, no Luck odds, and nothing reaching a leaderboard. It is real work rather than calibration, and inventing a number for it would undo the honesty of everything on this page. Earlier readings are surfaced only at the reveal, never at the draw — handing you your own notes beforehand would have you reading your transcript instead of the target.

Alone, together, or in a room with no screens.

Practice does not have to happen on a screen, and it does not have to happen alone.

Solo

Draw, work the coordinate, lock, reveal, score. The whole loop, on your own schedule.

Monitored

A live monitor runs the session with you — and is blind to the target, because the room is the biggest channel for cueing.

Hosted, including in person

Host a target event for a group: a sealed pool target, shared or one each, your choice of decoys, and one judging method locked for everyone so the results stay comparable. Run it in the room on paper and upload the readings after.

How an in-person event stays honest

For a shared target, the draw is sealed when the event is created — the host can't re-roll to fish for an easy one, and it isn't disclosed to the host either, even when the host is the judge. Locking the event is the upload cutoff, and the lock always happens before the reveal, so a reading can never be authored after the target is seen. Attendees with no account can take part: the host adds them by name and uploads their reading.

Offline and guest results count inside the event and stay off the public boards — nobody outside the room can audit what happened in it.
Reminders that bring a target with them

Set the days and the hour and the studio can nudge you, open a session with your pool already chosen, or draw and seal a target before the reminder is sent, so a coordinate is already waiting when you open it. The reminder never names the pool it came from — naming it would narrow the field before you start.

05 / SCORING
How scoring stays honest

The scoring rubric.

Commit before reveal

Impressions are timestamped and locked. Scoring runs only against what was recorded before the target was known.

Judged blind by someone else

AI scoring against the actual target — plus a separate human analyst, blind to the question and outcome, for double- and triple-blind work.

Rolled into its own tier

Every scored session lands in its own blindness tier — verified accuracy grows only from fully blind results, never self-scored highlights.

06 / EVIDENCE
The statistics, and the bar they have to clear

"Was I on target?" is the wrong question.

Pick an image at random and you'll land on the target sometimes. Against four, chance alone puts you on it once in four tries. So the only question worth asking is whether you were on target more often than chance would have put you there — and how confident anyone should be in the answer.

Your dashboard answers it with two numbers per row, in plain words rather than notation. Margin is how much you beat chance by. Luck odds is how often chance alone would match you — written as odds, because "1 in 215" is read correctly on sight and "p = 0.0047" is read as "small number, good?"

Judged by Panel On target You Luck Margin Luck odds
Blind analyst 1 in 4 9 / 24 38% 25% +1.4 1 in 12
Blind analyst 1 in 2 3 / 4 75% 50% +1.0 8 more to tell
You (self) 1 in 4 31 / 62 50% 25% not blind
Illustrative — sample data. On the real dashboard these rows are yours, and the third one is the one most products would have quietly counted.

Margin — how much you beat chance by

0 is level with guessing. +1 is a good stretch. +2 starts to look like something. +3 and up is hard to put down to luck. Margin tells you how big the gap is; it does not yet tell you whether to trust it. Statisticians call it the z-score.

Luck odds — how often chance would match you

1 in 5 means a guesser does this well one run in five — nothing much. 1 in 200 is a different conversation. When several rows are on screen we tighten the bar, because comparing five things gives chance five chances to look impressive. Statisticians call it the p-value, Holm-corrected.

Four rules we don't bend

  • Each panel size is scored against its own chance line. Against one decoy, chance is on target half the time; against three, one time in four. Blending a 1-in-4 result with a 1-in-2 result produces an average nobody actually worked against, so the rows are never pooled.
  • No verdict below 12 scored viewings. Three on target out of four is a huge margin and means nothing — four attempts can fall any way at all. Under 12, the row still shows, with how many more it needs; the odds stay blank because there is no honest answer to give yet.
  • Every displayed row is Holm-corrected. Showing six rows is six chances to find something impressive by luck, and the one that gets there is the one that gets screenshotted. The correction is applied across exactly the rows on screen.
  • Self-judged work is shown and never scored. Judging is the step where someone decides whether a description corresponds to the target. Doing it for your own session means you already know what you wrote, so a correspondence can be found after the fact with nobody meaning to bend anything. We still show the rate — it is real practice. It just isn't proof, so we don't present it as proof.
Where this currently stands

Nobody on rv.studio yet has enough blind-judged work for any result to reach statistical significance. The dashboard says so, per row, rather than hiding it behind a headline number. Including self-judged work would make everyone's figures look considerably better; we exclude it. Offline and guest work is kept off the global boards because no one can audit another group's room. None of these choices flatter anybody, which is the entire reason to trust the ones that eventually don't.

07 / CALIBRATION
Calibration over claims

We show the curve even when it's flat.

ILLUSTRATIVE — sample data

Calibration works the way honest forecasting does: stated confidence against achieved accuracy, computed separately for each blindness tier. A calibrated viewer sits on the diagonal — events called 70% happen about 70% of the time. If your hits are real, the curve shows it. If they aren't, it shows that too.

0.182Brier score
+6.4%Hit-rate over base
241Scored sessions
0.71Calibration slope

Reliability curve · confidence histogram · hit-rate sparkline — every figure stamped illustrative.

What the numbers mean

  • Brier score — a strictly proper scoring rule; lower is better. Rewards confidence that matches reality, penalizes over- and under-confidence alike.
  • Hit-rate over base rate — accuracy above what chance alone would produce for the same task. The honest measure of whether there is any signal at all.
  • Calibration slope — how closely stated confidence tracks observed accuracy.
What we claim / what we don't

Remote viewing is a contested area of parapsychology. rv.studio makes no claim of proven psychic ability. We provide the protocol and the calibration math, and we show your numbers honestly — including when they're flat. Whether the signal is real is a question your own calibration record is built to answer.

Sample figures on this page are stamped illustrative; live data is shown only to signed-in viewers.

Begin a calibrated practice.

Blind targets. Honest scoring. A calibration record you can trust.