Passphrases You Can Memorize That Even the NSA Can’t Guess

Re:Passphrases You Can Memorize That Even the NSA Can’t Guess by bradleyjg on Friday March 27, 2015 @03:16PM (#9278448)

This is the more usual objection that I was referring to with Kolmogorov complexity. Forty random digits has very decent entropy, 40 digits of PI, though they look very random, has much less. From that you can see how it's really hard to quantify exactly how much entropy there is an song verse that exists somewhere on the internet.

If you wanted an absolute ranking of password strength, you'd first want to construct a universal brute force dictionary of passwords. Clearly the password 'password' is going to be near the beginning of the list and a password consisting of 48 randomly picked characters from [a-zA-Z0-9!@#$%^&*()] is likely to be somewhere out beyond the number of particles in the universe. But where exactly is "When you ain't got nothing, you got nothing to lose" going to appear? The first 10 million? billion? 10 billion? trillion? I really have no idea.

[ Parent | on HN ]

Re:Passphrases You Can Memorize That Even the NSA Can’t Guess by abandonliberty on Friday March 27, 2015 @07:58PM (#9279826)

We should be able to at least estimate an upper bound.

If wikipedia has 2.5B words, let's generously assume 10B words. Let's allow phrase lengths of 1-10 words and arrive at 100B combinations.

Optimistically we're at 37 bits, not even a 7 character [a-zA-Z0-9!@#$%^&()].

And unlike the [a-zA-Z0-9!@#$%^&*()], we could madly cut that down based on frequency and popularity.

Therefore existing phrase passwords are security through minority. If they are adopted, they provide at best an order of magnitude or two over simple dictionary words.

But we're kidding ourselves, most people still use 'password'.

[ Parent | on HN ]