-1

I've noticed that some sites publish very detailed password requirement statements to users.

Here's an example screenshot (not singling out this site - just a random image that came up when googling 'password requirements'):

enter image description here

In this case, a bad actor would know details about the password including:

  • Minimum password length.
  • Mandatory characters
  • Values that are definitely NOT part of the password

I'm curious -- is this level of detail in describing password requirements actually harming the security/integrity of the password strength? I suspect that the complexity and permutations of the characters required here outweigh the potential risk of having "weak passwords" automatically excluded from cracking attempts.

That said, are there clear standards/controls that advise one way or the other?

2
  • And then what would you say to the poor user who keeps trying to enter a password that keeps getting rejected? Password requirements are detailed if the required password has complex requirements.
    – schroeder
    Commented Sep 11, 2020 at 7:07
  • 2
    Not quite what you asked, but users hate those kinds of password requirements. An alternative is to just have a minimum length and check if the password is in any known beach (using haveibeenpwned or similar)
    – paj28
    Commented Sep 11, 2020 at 8:02

1 Answer 1

1

The entropy (number of possible passwords) you lose to those requirements is trivial compared to the number of people who would otherwise use one of the 100 most common passwords out there. So no, there's nothing wrong with publishing the requirements in detail, not if they're any good.

Of course, that last clause matters. Most of the requirements in that list are worthless. The only ones, other than the length (which is on the short side) that I'd consider really good are the "... or commonly used passwords" and maybe "... commonly used password character patterns" and anything involving names. Depending on how those rules are implemented*, they might actually enforce decent passwords. Otherwise, the most common password on that site is going to be "Password1!" (10 characters; one each of upper, lower, digit, and symbol; no long strings of the same character; no names). Next most common will be various common words (or names) "encoded as text/1337", as a colleague once described it (consider "$tackOverfl0w" as a password for this site, which also potentially fits all the requirements in your picture, or the initial example in this XKCD.

The requirements for different cases, numbers, and symbols are particularly egregious, because they make it hard to use passphrases (see Diceware or the solution XKCD suggests instead in the link above). For very short passwords (less than 12 or so characters) they might do more harm than good just by slightly reducing the likelihood of people choosing the same substitutions, but those substitutions add very little security even if chosen at random (and they're much worse than random; in practice people use the same ones) so they're much worse than just using a better way of generating a password to start with.

* I once saw a service - whose entire purpose was doing secure logins for other sites - that claimed to offer a "prevent the use of the most common passwords" feature but failed miserably at it. Their "most common passwords" list was too short - only 10,000 passwords - and it wasn't even the 10,000 most common passwords that fit their other requirements. In fact, not a single password on the list fit their other requirements so the entire feature was worthless as it would never rule out a single password candidate.

Not the answer you're looking for? Browse other questions tagged .