Scryway

Remote viewing guide

Remote viewing practice asks you to describe a target you cannot see, then compare what you recorded with the target afterwards. These tools are for structured practice and honest self-review. A session result is not scientific evidence of psychic ability.

Session types

Single blind target

When you begin, the server picks a hidden image and gives you a reference number such as RV-4821-9037. The number is random and tells you nothing about the image. You record impressions, lock them, and only then see the target. You then reflect on the correspondence in your own words. There is no accuracy percentage: judging resemblance between a free description and a picture is subjective, and a number would suggest more precision than exists. You may add an optional 1–5 self-rating, which is labelled as your own judgement made after seeing the target.

Multiple-choice practice (2 to 5 images)

The target is assigned before you record anything. After you lock your impressions, you see the target among one to four decoys, each a different kind of subject, in random order. You pick one, optionally state how confident you are and rank every image, then the choice is locked and the answer shown. Guessing alone is right one time in N (50% with two images, 25% with four). Short runs swing widely either way; a handful of hits or misses says very little.

Ranked

Ranked sessions always use four images, require a confidence, and count an abandoned session as a miss, so results cannot be cherry-picked. Each one moves a private rating (see below). There is no public leaderboard yet.

Daily target

Everyone taking part on a given day (UTC) works with the same pair of images. Each person’s target is one of the two, chosen at random for them, so a more appealing picture cannot favour one answer for everyone. After you choose, your session is sealed; results for the day are released together at 00:00 UTC, with an anonymous count of how everyone did. Each day’s images are kept out of all other sessions until then.

Precognitive option

For single-target and multiple-choice sessions you can choose to have no target at all while you record: the server draws it at random only when you lock your impressions. Precognitive results are listed separately in your statistics.

Recording impressions

  • Write down raw impressions first: colours, textures, temperatures, sounds, shapes, lines, sizes, how things sit in relation to each other.
  • Keep guesses about what the target is separate. The worksheet has its own field for analytic guesses, so you can see later which kind of note matched.
  • Sketch shapes and layout rather than trying to draw a finished picture. The sketch pad has colours, three line widths and a text tool for short labels such as “water?”.
  • The timer is optional and imposes no limit.
  • Drafts save as you go. Locking is final: the submission is stored as it was and cannot be edited. Later thoughts go into dated reflections beside it.

The worksheet is a general structure for notes. It is not Controlled Remote Viewing (CRV) or any other named protocol; named protocols will only be added after their sources and requirements have been reviewed.

Reviewing a session

  • Compare specific elements: which impressions correspond, which do not, which are ambiguous.
  • Notice when an analytic guess led you away from what the raw impressions described.
  • Keep misses and abandoned sessions in view. Abandoned sessions stay in your history with their status, and their targets are not shown.

Confidence, scores and statistics

  • Confidence is your stated probability that your choice is the target, from the chance level up to 100%. Single-target sessions can record, before feedback, how well you expect your impressions to match (1 to 5).
  • Brier score: the average of (confidence − outcome)², where the outcome is 1 for a hit and 0 for a miss. 0 is perfect; always stating the chance level gives the reference score shown beside yours, and the skill figure compares the two. Being sure and wrong costs more than an honest middle answer.
  • Calibration groups sessions by stated confidence and shows how often you were right in each group.
  • Ranking score: when you rank every image, the target’s position becomes a score from 1 (first) to 0 (last); chance averages 0.5.
  • Against chance: hits are compared with the number expected by guessing, and the chance of doing at least this well by guessing is calculated exactly (sessions with different numbers of images are combined correctly). It is only shown after 12 sessions, and it is not proof of anything: among many people practising, some will score well by chance.
  • Rating (ranked only): starts at 1000, where the expected result equals chance. After each session it moves by K × (result − expected), with expected = 1 / (1 + 3 × 10(1000 − rating)/400) for four images and K = 32 for your first 20 ranked sessions, then 16. Deleting a session does not undo its effect on the rating.
  • Categories: after feedback, results are also grouped by the broad label on each target image (water, structure, vegetation…). Free descriptions are never given an accuracy percentage.

How targets are kept hidden

Targets are chosen and stored on the server before you record anything. Their images, titles and descriptions never reach your browser until the feedback stage. Image files have random names and stripped metadata, and the image store refuses requests for a target until your own session has reached feedback. A refresh, a retry or opening the session on another device shows the same target. Because the target stays on the server, blind practice needs a connection.

The target pool is about 1,300 openly licensed photographs from Wikimedia Commons’ featured pictures, across water, coast, mountains, landscapes, plants, structures, interiors, cities, vehicles, boats, animals, people, objects, sky and ice, each credited at feedback. Every one was checked before use: images with legible text, signs or watermarks were left out, as were weak or cluttered scenes and near-duplicates. New sessions prefer images you have not seen before.

Descriptors, decoys and practice settings

Every target is tagged with a fixed set of fifteen descriptors (water, land, sky, natural, man-made, structure, people, animal, plants, motion, tall, enclosed, dark, light source, cold) and a main colour. They are used in three ways, all of them optional except the first:

  • Decoys. In choice sessions the decoys are picked to differ from the target in as many descriptors as possible, so that choosing correctly means telling genuinely different scenes apart.
  • Descriptor checklist. Switched on in the practice settings, it asks you to tick what you sense before you lock. After feedback it is scored against the target, and set beside the number of matches you would get by chance with the same answers, worked out from how common each descriptor is among the possible targets.
  • Frontloading. Off by default, so every target is possible. Switched on, you can limit the draw to chosen categories or to targets with chosen descriptors (“water”, or “man-made with people”). That is a hint about the target, so frontloaded sessions are labelled, scored against chance within the narrowed set, and kept apart in your statistics. Ranked sessions, the daily target and shared targets are always fully blind.
  • Beginner targets. A smaller set with one clear, bold subject each, for getting started.

Start a session