I was reading about random numbers and their usage in security. I'd like to verify with more experienced people how good this method of generating random numbers will be.

Human beings are very hard to predict, their movements and behavior. When we combine that with different kind of sensors and different factors in the nature (environment) we can get very good random number, almost unpredictable.

What do you think, how strong and unpredictable this random number would be? Is this something that can give totally different and random every time? I guess when I combine all of those factors mentioned above probability of hitting the same number is very low since each person is different and is in different environment with different behavior and movement.

  • $\begingroup$ Your Question is not very clear , can you be more specific with you question ...Thanks $\endgroup$ Feb 8 '19 at 19:19
  • $\begingroup$ Considering: person, environment he'is in and sensors in his device, the numbers we can generate from those different states, will this result in generating random numbers with low predictability? $\endgroup$
    – Misko Mali
    Feb 8 '19 at 19:23
  • $\begingroup$ Welcome to crypto.stackexchange - "what do you think" type questions are not an appropriate format for our site. We require questions to be objectionably and precisely answerable. $\endgroup$
    – Ella Rose
    Feb 8 '19 at 23:38

Most computers already have a number of sensors that feed data into the system's random number generator. The timings of IO operations and thermal noise from system thermometers already give plenty of data. Once a random number generator is seeded with enough random data (256 bits of high-quality randomness at the very least, or some higher amount of lower-quality randomness), no more random data is required for the system to be able to securely generate practically unlimited unpredictable random numbers. This problem is pretty much solved already.

You could hook up more inputs to feed into the system's random number generator, but it would be more useful and easier to just feed raw video data of a person (or of anything, even of a dark room) into the random generator instead of parsing the video data to understand a person's movements and input those into the random number generator. The amount of entropy across the video's entire pixel inputs is likely to be higher than the amount of entropy of a person's movement data. If you think a person's movement data is unpredictable, well a video of that person moving plus all the random noise in the video will be even more unpredictable.


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