I am looking for a method of making random numbers that is off-line and is also independent of specialist equipment like a Geiger counter.

  • 3
    $\begingroup$ a deck of cards $\;$ $\endgroup$
    – user991
    Apr 23, 2014 at 8:23
  • 2
    $\begingroup$ Dice are specialist equipment? $\endgroup$
    – Thomas
    Apr 23, 2014 at 8:47
  • 1
    $\begingroup$ Flip a coin? Cards makes the most sense - although make sure to shuffle them sufficiently many times using a riffle shuffle (other shuffles are either equivalent or worse) to ensure propper randomisation $\endgroup$ Apr 23, 2014 at 8:55
  • 1
    $\begingroup$ What kind of number? 0/1? 1..6? π? $\endgroup$
    – CL.
    Apr 23, 2014 at 9:20
  • 1
    $\begingroup$ Do you expect cryptographically secure randomness, or merely some pseudo-random numbers? Do you assume the human brain to be “specialist equipment” or not? How will you solve the issue that an averagely trained human brain can not remember a sequence of 100 random numbers? Fun aside – your question is a mixture between “not a real question” and “too broad”. (I’m not sure if we even handle “lack-of-creativity” at Crypto.SE.) In the end, I decided to flag your question for being too broad. $\endgroup$
    – e-sushi
    Apr 23, 2014 at 19:33

1 Answer 1


Do a long jump, measure the distance in millimeters, take it modulo 10.

Repeat 100 times.

Another idea: use clock timer that shows milliseconds, start it and then turn off after some time, look at the last digit.

  • $\begingroup$ Too easy to measure incorrectly - people are bad measurers, they tend to cheat, round off, etc. Besides, we are cryptographers - have you ever seen one of us jump? Removed downvote, probably a joke :) $\endgroup$ Apr 23, 2014 at 11:21
  • $\begingroup$ Why to cheat here? And it is not hard to measure with enough precision. $\endgroup$ Apr 23, 2014 at 11:33
  • $\begingroup$ You may not intentionally cheat, but just be compelled to round off to e.g. half centimeters. If you do a lot of measuring, you will notice that this is quite a strong tendency. Besides that, a long jump requires a lot of equipment to be in place, and 100 jumps are not healthy. Gosh, you were not joking. $\endgroup$ Apr 23, 2014 at 11:36
  • $\begingroup$ @owlstead Well, OP did not specify a time limit, so he could do, say, five jumps a day and be done in under two weeks. But yes, this kind of sampling is really not guaranteed to produce a uniform distribution. $\endgroup$
    – Thomas
    Apr 23, 2014 at 11:43

Not the answer you're looking for? Browse other questions tagged or ask your own question.