Re: Free Random Password Generator

From: Nick Maclaren (nmm1_at_cus.cam.ac.uk)
Date: 10/04/04

  • Next message: David Douthitt: "Re: inetd.conf security" 
    Date: 4 Oct 2004 21:49:15 GMT

    In article <cjsfol$i55$1@nntp.itservices.ubc.ca>,
    Bill Unruh <unruh@string.physics.ubc.ca> wrote:
    >
    >I think what the person really wanted was a password generator for
    >passwords that had some vague chance of being memorable. That means that it
    >has to produce stuff which looks something like English (or whatever
    >language) words. I once wrote such a generator, and it really worked pretty
    >well. The stuff produced looked like it could be a bunch of words, even
    >though they were not (I did double letter analysis. Ie, go through a whole
    >bunch of text, and calculate what the probablity of any letter following
    >another-- then make up the words randomly choosing from those
    >probabilities. Ie, choose the first letter according to the probability of
    >of the letters in some huge English text. Choose the second according to
    >the probability of letters following that first, the third according to the
    >probability of following that second etc. ) The entropy was about .3
    >letters per letter, password of say 25 letters would have the entropy of 8
    >bytes. But the passwords were at least somewhat memorable-- they looked
    >like English.

    I wrote (and use) one that has a dispersion of about 31 bits, and
    produces words that are vaguely pronouncable. However, I am a bit
    of a philophile, and that includes many of the constructions that
    are likely in English transliterations of Dutch, Czech and other
    such languages.

    What is needed is the ability to accept complete phrases, where it
    is easy to get dispersions of 60+ bits and still be memorable.

    Regards,
    Nick Maclaren.

  • Next message: David Douthitt: "Re: inetd.conf security"

    Relevant Pages

    • Re: Free Random Password Generator
      ... ]>of the letters in some huge English text. ... ]>the probability of letters following that first, ... ]I wrote one that has a dispersion of about 31 bits, ...
      (comp.security.unix)
    • Re: On Bayes
      ... Can I compute from this the conditional probability P(doc having 27 ... to specify the probability of getting b 'bye', h 'hello', and e ... 'english' in a document of type X, ... distribution if the distribution is already given ...
      (sci.math)
    • Re: On Bayes
      ... Can I compute from this the conditional probability P(doc having 27 ... to specify the probability of getting b 'bye', h 'hello', and e ... 'english' in a document of type X, ... you could also get by approximating the distribution ...
      (sci.math)
    • Re: On Bayes
      ... Jim Ferry wrote: ... Can I compute from this the conditional probability P(doc having 27 ... 'english' in a document of type X, ... distribution if the distribution is already given ...
      (sci.math)
    • Quantcast