Toolbox runs in your browser

Index / Chance

Decision Ball

Twenty answers, one drawn at random. Ten of them agree, five refuse to say and five disagree, and the whole list is printed further down so the odds can be checked rather than trusted.

Your question
This box is for you. Nothing reads it: the answer is drawn before the text is looked at, the text is never put in the address bar, and it does not leave this tab.
Not asked yet

Press Ask. Every answer is equally likely.

Asked 0
Chance of any one answer 5%

How this session has gone

nothing yet

Every answer in the ball

20, each 1 in 20
What this is, and how the draw works This is not the toy from 1950 and it is not pretending to be. Magic 8 Ball is a registered trademark of Mattel, and the twenty answers moulded into that die are theirs. The twenty here were written for this page. What is shared is the idea — twenty answers, ten agreeing, five non-committal, five disagreeing — which is a structure rather than an expression, and which is stated here so nobody has to guess.

The question is not used. The answer is drawn first and the box is never read; it exists because writing the question down is most of what the object is for. Nothing on this page inspects the text, no answer is more likely for one question than another, and the text is deliberately kept out of the address bar, because a question worth asking a ball is often not one you want in your browser history.

Every answer is exactly 1 in 20. Not approximately: the draw uses crypto.getRandomValues and rejects any value that lands in the short remainder at the top of the 32-bit range, because 232 is not a multiple of 20 and taking the modulo alone would make the first 16 answers very slightly more likely than the last four. The page says underneath which generator it used.

The tally below is a record, not a mechanism. Each draw is independent of every other, so four disagreements in a row is a thing that happens about once in every 160 asks and means nothing about the fifth. The expected share is printed beside the actual one so the difference is visible rather than felt.

Nothing here is stored. Reload the page and the tally is gone; there is no cookie, no local storage and no server that could have kept it.