Re: Publishing password rules

["Basgen, Brian" <bbasgen () PIMA EDU>]

 A key component of password strength is entropy. Running English text has extremely low entropy which is why, for 
example, plain text can be compressed so well. Thus, when considering password strength as a function of bit-strength, 
a long passphrase of simple words is quite weak. Passphrases also have end-user challenges -- namely typing a long 
string of characters the user can't see without making a typo, etc.

~~~~~~~~~~~~~~~~~~
Brian Basgen
Information Security
Pima Community College




> -----Original Message-----
> From: The EDUCAUSE Security Constituent Group Listserv
> [mailto:SECURITY () LISTSERV EDUCAUSE EDU] On Behalf Of Strzelec, Wally
> Sent: Thursday, November 06, 2008 7:16 AM
> To: SECURITY () LISTSERV EDUCAUSE EDU
> Subject: Re: [SECURITY] Publishing password rules
>
> I like the idea of a "password phrase".  Complex passwords are hard to
> type and hard to remember.  A simple silly phrase such as "The cow is
> all red" is easy to remember, type and its 18 chars.  It is also very
> easy to add complexity by simply misspelling a word adding a period
> etc...  I think that when it comes to strong passwords, length is
> better than complexity.
>
> -Wally
>
> -----Original Message-----
> From: The EDUCAUSE Security Constituent Group Listserv
> [mailto:SECURITY () LISTSERV EDUCAUSE EDU] On Behalf Of Geoff Nathan
> Sent: Wednesday, October 29, 2008 9:46 AM
> To: SECURITY () LISTSERV EDUCAUSE EDU
> Subject: [SECURITY] Publishing password rules
>
> A week or so ago I asked for opinions on whether publishing strong
> password standards constituted a security risk.  The background for
> this is that we have just instituted increased strength requirements
> (minimum eight characters, at least one upper case and at least one
> numeral, no obvious matches--dictionary, accessID etc.)  We’ve now had
> to back off a little because of *&#$%&!! Oracle limitations that forbid
> non-alphanumeric characters (well, most of them).
> As part of this we’ve been debating whether we should publish the rules
> or let users play twenty questions.  I personally favor publishing the
> requirements behind some authentication wall, such as the password
> change page.  By a large majority (12-3) the folks who responded to my
> question agreed.  Several pointed out that eight characters was
> probably too weak to make any difference, and, in general I agree, but
> bumping that number up would not fly here at the moment, especially
> given a six-month expiry cycle.
> Many thanks to the following for the responses:
>
> Valdis Kletnieks
> Roger Safian
> Steven Alexander
> Vijaya Sastry
> Adam Nave
> Tim Doty
> Alex Everett
> Bill Terry
> Bob Bayn
> Brian Basgen
> Jack Suess
> Conor McGrath
> Jim Rizzo
> Gary Dobbins
> Joel Rosenblatt
>
> Geoffrey S. Nathan
> Faculty Liaison, C&IT
> and Associate Professor, Linguistics Program
> +1 (313) 577-1259 (C&IT)
> +1 (313) 577-8621 (English/Linguistics)