Ideograms & gestalts
Write the target code and continue directly into a fast, automatic mark. Decode only basic motion, feel, and broad gestalt. Repeat roughly 3–6 times.
Double-blind experiment · FLOW-informed sessions
Describe first.
Look later.
A five-target blind test with session structure adapted from Daz Smith’s FLOW method: raw data first, interpretation clearly separated, and no feedback until every session is locked.
5sealed targets
The protocol
Five steps.
One reveal.
The aim is to reduce information leakage and unconscious influence. The images are selected and committed in advance, while the viewer sees only random codes during the sessions.
- 01
Prepare
The script selects five new images from Wikimedia Commons or a local candidate folder.
- 02
Seal
Each image receives a random code. The mapping is stored locally and kept hidden.
- 03
Session
Run the FLOW-informed sequence below for each code. Lock every page before continuing.
- 04
Reveal
Reveal only after every session is complete. The script verifies the commitment first.
- 05
Match
Compare every session with all five images and score without favoring the true pairing.
Daz Smith’s approach
Let the data build.
FLOW grew from Controlled Remote Viewing (CRV). Each stage widens the “aperture”: start with fast, low-level impressions, then add sensory, spatial, and conceptual detail without rushing to name the target.
Sensory data
Record brief clusters for touch, smell, sound, sight, and taste. Add basic dimensional qualities, then note your aesthetic impact (AI).
Sketch & probe
Sketch shapes, scale, density, and spatial relationships. Label parts, probe them for fresh sensory data, and use movement prompts when useful.
Data matrix
Organise the flow under S2, D, AI, EI, T, I, AOL, AOL/S: sensory, dimensional, impacts, tangibles, intangibles, and overlays.
Interrogate & mind-map
Select an important intangible or AOL and break it into smaller data clusters. Daz’s FLOW adaptation uses a mind map to expose the elements beneath a label.
Model & summarise
Where useful, combine dimensional impressions in a simple 3D model while recording new data. End with a concise summary and your strongest sketches.
Describe; do not identify
Prefer “tall, hard, cool, echoing” over “church.” Naming the target too early turns description into analysis.
Declare analytical overlay
When a sharp image, noun, or guess appears, mark it AOL, pause briefly, and return to raw descriptors.
This is an independent experimental adaptation, not official training material. Primary reference: Daz Smith, FLOW — formerly Open Source CRV.
Before reveal · visible
Codes and commitment only
The image content and pairings remain unknown.
TARGET 1JW7-718
TARGET 2EF2-972
TARGET 3TL2-268
TARGET 4JL1-337
TARGET 5UW1-882
SHA-256: 23de4a5a1730286b…64e1bf
Before reveal · concealed
The secret mapping
Files receive generic .bin names, preventing thumbnails from accidentally exposing their content.
JW7-718
EF2-972
TL2-268
JL1-337
UW1-882
sealed/f7a39d…b812.bin
Local script
Two commands.
Python 3 is all you need; no external packages are required. The script creates a hidden .rv_test folder next to itself.
$ python rv_test.py prepare → record the five codes → complete and lock all five sessions $ python rv_test.py reveal
Prefer your own images? Use python rv_test.py prepare --local /path/to/images. Provide at least five candidates.
After the reveal
Score as a matrix.
Compare every session with every image. The diagonal matters only if the true pairings receive the highest independent score more often than expected by chance.
| Session | Image A | Image B | Image C | Image D | Image E |
|---|---|---|---|---|---|
| JW7-718 | 8 | 2 | 1 | 3 | 2 |
| EF2-972 | 1 | 7 | 3 | 2 | 2 |
| TL2-268 | 2 | 1 | 9 | 2 | 1 |
| JL1-337 | 3 | 2 | 1 | 6 | 2 |
| UW1-882 | 1 | 3 | 2 | 2 | 8 |
Protocol hygiene
Four hard rules.
A clean procedure makes the result easier to interpret. Always record deviations; never hide them after the fact.
No interim reveal
Do not open any image until all five sessions are complete.
Lock every transcript
Save the date, time, and unedited text before the reveal.
Score blind
Ideally, the judge should not know the correct mapping.
Pre-register repetitions
Choose the number of rounds and scoring rule before seeing results.
Important: one striking round is not evidence of a paranormal effect. This design reduces some forms of bias, but reliable conclusions require repetition, pre-registered analyses, and an adequate sample size.