Fun with GUIDs *** Geek Topic Advisory ***

by Elsewhere 22 Replies latest jw friends

  • Pole

    ::I think the gaussian distribution is the same as the normal distribution mentioned in the thread.
    C'est vrai, but I've already said that, my friend ;-). BTW, are you a French, or Dutch speaking Belgian?
    ::This gaussian distribution is observed in nature a lot. Especially in quantum physics.
    And in social sciences. Except that perfect gaussian (normal) distribution is rarely to be found in practice.

  • DannyBloem
    C'est vrai, but I've already said that, my friend ;-). BTW, are you a French, or Dutch speaking Belgian?

    Oups did not see that. I am a dutch one.

  • Euphemism

    Sorry, Elsewhere, but I have to nitpick.

    What this means is that we would need a battery of computers working 24/7 generating 71.9 quadrillion GUIDs per second for the entire life of the universe before we would find a duplicate GUID.

    That is not actually correct. First off, as others have already pointed out, GUIDs are not actually generated randomly. However, even if they were, the amount of time needed to reach collision is different from the amount of time needed to generate the entire number space.

    Theoretically, with random generation, a collision can happen after only two numbers have been generated. The odds of that, however, are 1 in 2^128. After that, with each new GUID that is generated, we have a minisculely increasing probability of a collision. With each new sample generated, the probability of a collision increases proportionally to the number of samples already generated.

    It's similar to the famous question of how likely it is, in a given group, for at least two people to share a birthday. The rate of increase of probability given an increase in the size of the group is considerably higher than linear.

    Still, the numbers are such that this is a moot point from a practical perspective. I would happily bet my life on the avoidance of a collision.

Share this