We bring three old instruments into a new frame: a symbolic draw protocol, hardware randomness seeded by atmospheric noise, and the cortical drift of an operator in ganzfeld conditions. We treat each as a formal device that perturbs the observer's prior, preregister every question, and publish the trials that find nothing, which is most of them.
An iconographic deck used not to predict the world, but to perturb the operator. Cards map deterministically into pre-declared outcome bins; we measure the shift in the operator's prior, never the card.
A matchbook-sized hardware RNG seeded by low-frequency atmospheric background noise. Operators are asked to hold a question in mind while the device emits a single stream of bits.
A receptive state under uniform visual and auditory field. We record 32-channel EEG and look for drift patterns that correlate with target selection, blind to the operator.
A binary or k-way question with a specified outcome time is filed to the timestamp registry. No new question may be substituted once the trial begins.
The operator, spread and instrument each have explicit numeric priors. The priors themselves are auditable and have been calibrated on previous trials.
The trial is run: spread drawn, RNG bits collected, ganzfeld session recorded. The operator does not know which instrument will be consulted first.
Readings are transcribed by a party who does not know the question. Scoring is mechanical: a pre-declared table maps draw bins and bit patterns to the outcome bins named in step 01.
The prediction is written, signed and timestamped. Once sealed, it cannot be edited. No one on the team sees it until the outcome time passes.
At the outcome time the world is checked. If the outcome is ambiguous, the trial is voided, and that voiding is recorded too.
The result (hit, miss, void) is logged in the registry. Trials without findings are published with the same care as trials that succeeded.
Priors are updated. The calibration is preregistered for the next cohort. We have never yet recalibrated in a direction that made us look better.
We claim that an operator who performs a draw makes demonstrably different decisions, and we study those decisions.
We claim that bit-stream statistics measured during a held question sometimes deviate from the null. We do not yet know why.
We claim it is a specific sensory regime that produces reproducible cortical drift in some operators, and we are curious what that drift correlates with.
They have survived the first replication cohort. That is the only claim we make about them. The registry is public. Come break them.