Experiment III · protocol
The Psychic Self-Test
Five cards, one chosen at random, and a sealed commitment shown to you before you guess — so you can prove afterwards that the answer was never changed to suit the result.
This is a protocol, not a result
Collection has not opened, so there is no data on this page and no counter to watch. What is published here is the plan — filed in advance precisely so it cannot be adjusted once results start arriving. The charts below describe the null hypothesis, computed from the design itself.
The wager
Five symbols, one drawn at random: chance is one in five, 20%. The claim under test is that some people identify the card more often than that. The null says the rate sits at 20% forever, however long anyone plays.
Fixed in advance: two-sided exact binomial against p₀ = 0.20, α = .05, with reads at 10,000, 50,000 and 250,000 trials, and five Holm-corrected per-symbol comparisons.
Why you can trust the scoring
Every online psychic test has the same flaw: the server picks the card and also marks your answer, so nothing stops it picking after you guess. A good score proves nothing to anyone, and a bad one is always suspect.
So before you see anything, the server draws a card and a long random nonce, and shows you the SHA-256 hash of the two together. That hash is on screen while you choose. Afterwards you get the card and the nonce, and your own browser recomputes the hash and tells you whether it matches. If it does, the card was fixed before your guess existed — the server could not have found another card and nonce producing the same hash.
What this does not prove is that the draw was fair. A server could commit honestly to a loaded deck. That is why the dealt-card frequencies are published and checked against a fifth apiece: a loaded deck shows up there, and the protocol names it as a prespecified check rather than a defence invented later.
A score in a run of 25
| Score | Probability | Roughly |
|---|---|---|
| 9+ | 0.0468 | 1 in 21 |
| 10+ | 0.0173 | 1 in 58 |
| 12+ | 0.0015 | 1 in 649 |
| 15+ | 1.4e-5 | 1 in 73,720 |
Scoring 9 or more out of 25 — which feels remarkable — happens about once every 21 runs by chance alone. At 5,000 runs, roughly 234 people will manage it with no ability whatsoever.
The filed protocol
SHA-256 5bd550b2e8d7102f649c26338774c53172e531a1d81a8929f371088a3eb1627f
Known ways this could mislead
- With enough participants, extreme individual results are guaranteed. In a run of 25 trials, scoring 9 or more happens about once in 21 runs by chance — so a few thousand runs will produce hundreds of apparent psychics. No individual score is treated as a finding.
- The claim that scoring 9 of 10 is roughly one in three thousand is wrong: it is one in 238,186. The one-in-3,125 figure belongs to five consecutive hits, a different event, and the two are routinely conflated in the direction that flatters the scorer.
- The commitment proves the card was not changed after your guess. It does not prove the draw was unbiased — that is what the published card frequencies are for.
- People stop playing when bored or losing. Because the card is sealed before the guess, stopping cannot select for favourable outcomes, but per-participant rates are shown with their trial counts so a short lucky run cannot pass as a high rate.
CHANCE LINE 20%