in reply to Password cracking algorithm

gents

some quick facts & openings re: this stuff:

for your algorithm, you would typically want to mimic what elcomsoft do with their office password recovery products, namely:

start with a dictionary attack.

proceed to a dictionary attack with smart mutation enabled (trying all uc & lc combos, other digit substitutions etc.

browserUK has a point re: time in an abstract sense: the following excerpted from the elcomsoft site

even if the password contains just small and capital letters, and the length is 12, the total is 52^12 = 390,877,006,486,250,192,896. Even if ***** will be able to test a million passwords per second (actual speed is lower), it would take more than twelve million years to find the correct one. Well, if you're lucky enough -- just six million years ;)

HOWEVER
certain block & stream ciphers (ie RC4 stream c. used in office) have smaller key lengths which enable effective brute force attacks against them. the maximum time against RC4, for instance, given by the (reliable) source above is 13 days.

regards,

wufnik

in the world of the mules there are no rules

Replies are listed 'Best First'.
re: password cracking algorithm - 14 chars in 14 secs?
by wufnik (Friar) on Jul 23, 2003 at 07:15 UTC
    hmmm. replying to one's own post... something i have sadly done before, but, but...

    it seems 14 characters can now be dealt with (lanman & mswin) in *14* secs

    cannot resist pointing out to those interested the timely & academic link from the politech list, mailed this morn:

    http://lasecwww.epfl.ch/pub/lasec/doc/Oech03.pdf. imho a great example of how hash statistics/methodology can be leveraged. ). if nothing else the article should emphasize the importance of good hash design...

    wufnik

    in the world of the mules there are no rules
[reply]
      Yeah, but if you read the article you notice that the set of characters is limited (no difference between upper and lower case), and the 14 chars are split into two sets of 7 chars before encrypting - both halves can be attacked separatedly.

      This means that the key space of the domain tried is about 0.03% than that of Unix passwords, if we restrict ourselves to alphanumerical passwords, like the article does. The precalculated data used in the article fits on 2 CDs. Assuming it scales lineary, for an attack on alphanumerical Unix passwords, you'd need about 12 million CDs (the keyspace is 3000 times as large, and there are 4096 seeds).

      The orginal poster asked about 16 character passwords, including "special" characters. If we assume the special characters are all printable ASCII characters that aren't letters or digits, we have a key space of 95**16. Compare this to the keysize of 36**7 of the article, the former is a tad more. If we scale the 2 CDs of the article to the problem of the OP, we'd need more than 10**21 CDs. And that's assuming you need the same amount of bytes to store a password, or crypted password, which seems unlikely.

      If the OP has a billion computers, each of them capable of checking 2**32 (4G) keys per second, it would take the OP almost 325 thousand years to exhaust the key space.

      Abigail

[reply]

In Section Seekers of Perl Wisdom