[37649] in bugtraq
Re: MD5 To Be Considered Harmful Someday
daemon@ATHENA.MIT.EDU (Paul Wouters)
Wed Dec 8 18:14:11 2004
Date: Wed, 8 Dec 2004 22:05:50 +0100 (MET)
From: Paul Wouters <paul@xtdnet.nl>
To: Dan Kaminsky <dan@doxpara.com>
Cc: Gandalf The White <gandalf@digital.net>, davids@webmaster.com,
BugTraq <bugtraq@securityfocus.com>
In-Reply-To: <41B7699A.5070507@doxpara.com>
Message-ID: <Pine.LNX.4.61.0412082203470.31218@expansionpack.xtdnet.nl>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
X-MailScanner-From: paul@xtdnet.nl
On Wed, 8 Dec 2004, Dan Kaminsky wrote:
>> Can't we just truncate the password to 8 characters like in the old days
>> before doing the MD5 hash? It will greatly reduce the chance of a
>> collision.
>> In fact, I am not even sure my systems don't do this already.
>>
> Actually, this greatly increases the chance of a collision. 123456789 will
> collide with 123456780.
I was talking about limiting the input, not the output. I assumed it would
be more difficult to find a collision when there are less degrees of freedom
in the input.
What I realised after my mail though, is that it is probably pretty easy to
prebuild a dictionary of all 8 character combinations and their MD5 sums.
Paul
