How long and complex should a password be? At what point is it effectively uncrackable?
Time out: Look at that opening paragraph. Its 87 characters long, but it could be your password to your Windows system. Yes, even with the spaces in it. Technically, this has become known as a “passphrase.”
Robert Hensing, a member of Microsofts Security Incident Response Team, has written in his blog that you shouldnt use passwords anymore for Windows systems—you should use passphrases. The blog entry has generated a lot of interest on security lists. Many agree with Hensing, and theres a lot I like about the idea. The discussions also raised my awareness of cracking tools for Windows passwords that do things you might not believe possible.
My opening lines might not make a good passphrase because its not very memorable. But lets say youre a Dead Head: “Its just a box of rain, I dont know who put it there” is a very strong passphrase, and it might be easy to remember. It has upper and lower case, punctuation and 58 characters. The downside relative to a more conventional password is that it has upper and lower case, punctuation and 58 characters. Its going to take a while to type, and youre more likely to make mistakes on it.
As Hensing points out, Windows has supported passphrases of up to 127 characters since Windows 2000. But boilerplate password advice from people like me has always focused on bizarre words that we kid ourselves as being easy to remember, like “Ih8m0d3rnART!” (“I hate modern art”). Take a phrase you can remember and distort it into a password. Hensing asks why not just use the phrase?
In fact, the “Ih8m0d3rnART!” example is instructive in another way. While it looks long and complex, and is relatively impervious to certain types of attacks, its only 13 characters and is therefore vulnerable to a weakness in Windows 2000 password hash methodology. Ill get into it more in a future column, but if you have local administrator access to the system, its possible to reverse-engineer Windows passwords up to a particular length. As I understand it, this problem has been eliminated in Windows 2003 domains, but remains in Windows 2000 for reasons of backward compatibility with third-party programs.
Next page: Malwares embedded dictionaries.
Page 2
Hensing also raises the problem of the large number of malware programs that carry embedded dictionaries of common passwords to try on systems they attack. Many have hundreds of passwords, most of them real sucker material like “password” and “asdf.” But over time youd expect these to get better and for brute-force cracking programs to be able to try more possibilities.
And not everything is as security-conscious as Windows (yes, my tongue is in my cheek). Barnes and Nobles password policies require you to have a six- to 12-character password composed of “letters, numbers, or Shift/numeric characters only; spaces cannot be used.” But the message is getting around; I just set my Amazon.com password to a 129-character passphrase with punctuation and mixed case.
Whether long and effective passphrases would be more acceptable to end users is a matter for research. But if longer, more complex passwords are better, surely passphrases are better than passwords, right? I saw two basic arguments against this in the discussions below Hensings blog and another on the Full-Disclosure list.
The first counter-argument says that if brute-force password crackers work by trying combinations of characters, a passphrase cracker would work by trying combinations of words. I have a hard time believing this could be a practical method of cracking, especially if you consider the possibilities for mixed case and punctuation.
The second argument is related to the first but raises the issue of “entropy” of the password, which refers to the randomness of the bits in the password. I dont fully follow this argument (especially the incoherent ramble I just linked to). Im more persuaded by Hensings position that the greatly increased length of a good passphrase trumps any weaknesses in the randomness of its bits.
Im on board with this, and Ive already begun to move my own passwords over to passphrases, but its going to be a tough sell to non-professionals. Will the only people willing to use passphrases be the ones who were willing to use complex passwords?
Security Center Editor Larry Seltzer has worked in and written about the computer industry since 1983.
Be sure to add our eWEEK.com security news feed to your RSS newsreader or My Yahoo page:
More from Larry Seltzer