Re: Are PassPhrases Secure Enough?

From: clintonG (csgallagher_at_REMOVETHISTEXTmetromilwaukee.com)
Date: 11/24/04

  • Next message: clintonG: "Re: Are PassPhrases Secure Enough?" 
    Date: Wed, 24 Nov 2004 14:53:40 -0600

    Einstein the genius shows up!
    Seriously, kidding aside thank you for an insightful response.

    <%= Clinton Gallagher

    "Valery Pryamikov" <Valery@nospam.harper.no> wrote in message
    news:umkNo7l0EHA.3832@TK2MSFTNGP10.phx.gbl...
    > Hi
    > The problem with passphrases is that they has near to natural language
    > entropy that is very low. Entropy is a measure of information uncertainty
    > and reverseproportional to information redundancy. Redundancy could be
    > thought as measure of data compressibility (i.e. more compression rate you
    > can achieve on your data - lower entropy of that data). Natural English
    > information redundancy is about 75%. This gives something like 1.25
    entropy
    > of Natural English language. I.e. each extra byte you add to your
    passphrase
    > only increases your password strength for something from 1.25 bits
    > (unlimited length) to 2.9 bits of security (for longer than 10 chars
    > passwords). I don't know why is that so popular to advocate natural
    language
    > passphrases lately ignoring many researches that was done during past two
    > decades that indicated that passwords build from passphrase by taking one
    or
    > more characters from each word would provide better security due to better
    > entropy while as they still quite easy to remember. You only need such
    > password of size 15 chars (to disable LM Hash) and it would be as
    difficult
    > to break as your passphrase of size more than 50 chars. I, for example,
    use
    > one or other line from Russian song or story, take some char from each
    word,
    > write them in Latin alphabet instead of Cyrillic (which means that
    sometimes
    > I have to use two Latin letter instead of one Cyrillic), throw in some
    > symbols and get password that looks like following: "KjbZ-m0d,jDbzRb". It
    is
    > composed from well known child story that I used to read to my sons - I'll
    > never forget it (I actually used it couple of years ago) - but it will be
    > very tough to break. For more information for choosing password - check
    this
    > paper by Ross Anderson (one of the world most acknowledged experts in
    > cryptography and information security) that was written in 2000
    > http://citeseer.ist.psu.edu/yan00memorability.html (click on cached PDF to
    > read the paper). Only note that they are talking about minimum of 8 chars
    > only because UNIX and Linux "crypt" password protection doesn't allow
    > passwords longer that 8 chars (and "crypt" still is used on some
    UNIX/Linux
    > distributions) . If you interesting you can also check my blog where I
    > posted 6 articles about passwords (just search my blog for passwords).
    >
    > -Valery.
    > http://www.harper.no/valery
    >
    > "clintonG" <csgallagher@REMOVETHISTEXTmetromilwaukee.com> wrote in message
    > news:O8V39Sk0EHA.2016@TK2MSFTNGP15.phx.gbl...
    > > I've been thinking of implementing the following and seek comments pro
    and
    > > con.
    > >
    > > When registering, the user provides an e-mail address and creates a
    > > password
    > > phrase question such as "What is your father's middle name?" The user
    also
    > > provides the answer to the phrase he or she has created.
    > >
    > > Future attempts to log in initially request the e-mail address only. A
    > > callback is issued and the login form is updated to display the user's
    > > passphrase "What is your father's middle name?" The user provides the
    > > correct answer and is presumably authenticated.
    > >
    > > That is the gist of this postulation. I am seeking comments regarding
    how
    > > to harden this methodology using hashed values to salt or other comments
    > > why
    > > this type of methodology would not be as secure as any other currently
    > > implemented.
    > >
    > > --
    > > <%= Clinton Gallagher, "Twice the Results -- Half the Cost"
    > > Architectural & e-Business Consulting -- Software Development
    > > NET csgallagher@REMOVETHISTEXTmetromilwaukee.com
    > > URL http://www.metromilwaukee.com/clintongallagher/
    > >
    > >
    >
    >

  • Next message: clintonG: "Re: Are PassPhrases Secure Enough?"

    Relevant Pages

    • Re: Should be in crypto for criminals Re: just stupid?
      ... > brute force the pass phrase at all, you can brute force it to obtain ... passphrases by making it so that only one passphrase needs to be remembered ... passphrases are pathetically low in entropy. ... > That's not the pass phrase generator. ...
      (sci.crypt)
    • Re: Are PassPhrases Secure Enough?
      ... The problem with passphrases is that they has near to natural language ... Entropy is a measure of information uncertainty ... I.e. each extra byte you add to your passphrase ... to break as your passphrase of size more than 50 chars. ...
      (microsoft.public.dotnet.security)
    • Re: Are PassPhrases Secure Enough?
      ... Entropy is a measure of information uncertainty ... > of Natural English language. ... > decades that indicated that passwords build from passphrase by taking one ... If you interesting you can also check my blog where I ...
      (microsoft.public.dotnet.security)
    • Re: md5 collisions and speeding tickets
      ... >>passphrase has roughly 20 bits of entropy. ... >>passphrase will have more than 20 bits of entropy. ... Good pass phrases contain no english words at all. ... a clue which nowadays most computer users ...
      (sci.crypt)
    • Re: hiding encryption keys
      ... >> To have something secure requires secure storage of some kind, ... Each individual has a limit to the amount of entropy they can ... memorize, if each individual can memorize say 120-bits, then each passphrase ... For a single passphrase this would deliver ...
      (sci.crypt)