Alien Math · Hunch Lab
Hunch or luck?
Five cards, face down. One is the lucky one. Pull the first card your gut points at.
Five cards. One is the lucky one. Trust the first pull.
Every score on this page sits next to the range plain luck would give you, and the luck meter swings into the warm zone only past that range. Practice as much as you like. When you want an answer that counts, seal a test: you pick its length first, you see nothing until the last pick, and quitting early stays on your record.
Game 1 · Five faces, one hidden card
Card guess
A card is already lying face down before you choose. Pick the face you think it shows. Luck gets it right one time in five.
Seal a test of this game
The card is drawn by your browser’s cryptographic random generator before you choose, so this is the classic card-guessing setup. The draw happens on your own computer, so anyone could peek at it with developer tools. Treat it as a game you play with yourself.
Game 2 · Four symbols, nothing drawn yet
Next symbol
Nothing is chosen yet. Pick the symbol you think the generator is about to draw. It draws only after your pick is in. Luck gets it right one time in four.
Seal a test of this game
Here the target does not exist until after you choose, so this one is about guessing the future. Same random generator, same caveat: it runs in your browser, so it is a personal check rather than a supervised experiment.
Game 3 · Mind over a coin
Nudge the coin
Light a lantern, high or low, then lean on the coin. It flips 25 times and every flip pushes the marble one step: heads up, tails down. Each flip is drawn the moment it lands, after your aim is set. Luck sends a flip your way half the time, and keeps the marble inside the pale fan 95 times in 100.
Pick a lantern. The coin starts flipping when you do.
Seal a test of this game
This is the setup Helmut Schmidt and later the Princeton PEAR lab used with electronic random generators: a person tries to push a stream of fair bits one way. Here the bits come from your browser’s cryptographic generator, one per flip. It runs on your own computer, so it is a game you play with yourself.
Game 4 · A hidden picture, four to choose from
Picture viewing
Somewhere in a stack of twelve drawings, one is about to become your target. Before anything is picked, describe it: a few words, a few tags, a scribble. Lock your notes, and only then does the generator pick a set of four and one target inside it. You judge which of the four your notes fit best. Luck matches one time in four.
Tags
Sketch
Seal a test of this game
This is the classic free-response setup from remote viewing and ganzfeld labs, shrunk to fit one person and one browser. The target is drawn by your browser’s cryptographic random generator after your notes lock, and you are your own judge. Your words and sketches stay in this page’s memory and are gone when you reload; only the count and any sealed picks are kept. Same caveat as the other games: it all runs on your own computer, so it is a personal check rather than a supervised experiment.
Game 5 · A hundred coins
Lucky streak
Streaks feel like a sign. Make up 100 coin flips that look random, then let luck flip 100 real ones next to yours and see whose longest streak runs further.
A hundred nights of luck
A streak you call ahead of time is rare. A streak that turns up somewhere in a long night is ordinary. Each line below is one night of 100 tries.
Every flip comes from your browser’s cryptographic random generator. The odds in the sentences are exact, counted by tracking the current run length one try at a time, so they do not wobble with the pictures.
Sealed tests
Your sealed record
Every sealed test you start shows up here, finished or not. Each entry is chained to the one before it with a SHA-256 fingerprint, so an edit made outside this page shows up as a broken chain. It all lives in this browser only: clearing site data erases it, and nothing is sent to us.
Reading the meter
Luck is streakier than it feels
The luck range
For 40 card guesses, plain luck lands anywhere from about 8% to 33% right, 95 times in 100. A 30% night is inside it.
Streaks happen
Guess 100 cards and luck alone gives you three right in a row almost half the time. The page tells you how likely your longest run was.
Stopping when it feels hot
Watching a score and quitting on a high turns luck into an apparent gift. Sealed tests remove the option: the length is fixed before you start.
Prior art
Who got there first
Card guessing
- J. B. Rhine and Karl Zener’s five-symbol cards, Duke University, 1930s. The faces in Game 1 follow that set.
Psi games online
- Dean Radin’s GotPsi tests (2000) and the IONS Psi Arcade app ran these games at scale long before this page.
- Russell Targ’s ESP Trainer app does feedback training on four-choice targets.
Pushing a random stream
- Helmut Schmidt’s binary random generator studies at Boeing, 1969 to the 1970s, asked people to push a lamp around a ring one way.
- The Princeton PEAR lab ran high, low and baseline aims on random generators from 1979 to 2007; a three-lab replication in 2000 did not repeat its effect.
- John Walker’s RetroPsychoKinesis experiments have let anyone online aim at prerecorded random bits since the 1990s.
Describing a hidden picture
- Russell Targ and Harold Puthoff’s remote viewing work at SRI, published in Nature in 1974, had viewers describe a distant target and outside judges match the descriptions.
- The ganzfeld studies judge one target against three decoys, so chance is one in four; Daryl Bem and Charles Honorton reported 32% matches in 1994. Game 4 borrows that four-way judging.
- Phone apps such as Remote Viewing Tournament already let you describe, then pick from candidate pictures. I haven’t found anything this game adds beyond the luck meter and the sealed record shared with the other games.
Streaks and the hot hand
- Mark Schilling, “The Longest Run of Heads”, College Mathematics Journal 21 (1990), worked out the longest-run odds and the classroom game of spotting made-up flips.
- NRICH’s What Does Random Look Like? and the Wolfram Demonstration Consecutive Heads or Tails already put the make-up-your-own-flips game online.
- Gilovich, Vallone and Tversky, “The hot hand in basketball”, Cognitive Psychology 17 (1985); Miller and Sanjurjo, Econometrica 86 (2018), showed a subtle bias in how streaks had been counted.
Fixing the length first
- Preregistration is standard practice in parapsychology now; the Koestler Parapsychology Unit has run a public study registry since 2012.