Re: draft paper: Deploying a New Hash Algorithm

2005-08-06 Thread John Kelsey
From: Steven M. Bellovin [EMAIL PROTECTED]
Sent: Aug 5, 2005 12:04 PM
To: Steve Furlong [EMAIL PROTECTED]
Cc: cryptography@metzdowd.com
.Subject: Re: draft paper: Deploying a New Hash Algorithm 

...
I'd have phrased it differently than Perry did.  I'd say
that the attackers are often cleverer *about security* than
protocol designers, because insecurity is their specialty.
Ordinary protocol desingers are good at designing those
protocols, but they haven't been trained to think about
security.  

Yes!  I've noticed that it's really common for me to work on
a project for a very short time (like an hour or two), and
start noticing all kinds of security holes, including a lot
of stuff with nothing to do with cryptography.  I'll still
be asking very basic questions of the other people on the
project about how things are *supposed* to work, but be
pointing out attacks they never thought of at the same time.
I think this is just a different way of thinking.  Attackers
and security people do this all the time.  Most normal
people never do--it's like once they've got the rules in
their heads, that's what's possible, and they don't even
think about it.  

How many times, working on security for some system, have
you pointed out an attack, only to hear some variation on
but who would think of that?  And you can see the same
thing happening in discussions of homeland security and
counterterrorism stuff.  It's like most people look at the
national guardsmen in the airport, and say whew, I feel
safer, rather than what the heck are those guys supposed
to do to stop hijacked planes crashing into buildings? 

I like your starting points, but I think the real approach
to thinking about this is a bit broader.  It has to do with
understanding the rules, and trying to ask, for each one,
and what makes me obey that rule? or what would happen if
I didn't do such and so?  

   --Steven M. Bellovin, http://www.cs.columbia.edu/~smb

--John Kelsey

-
The Cryptography Mailing List
Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]


Re: draft paper: Deploying a New Hash Algorithm

2005-08-05 Thread Steven M. Bellovin
In message [EMAIL PROTECTED], Steve Furlong writes:
 [Moderator's note: ... attackers are often cleverer than protocol
 designers. ...

Is that true? Or is it a combination of

(a) a hundred attackers for every designer, and
(b) vastly disparate rewards: continued employment and maybe some
kudos for a designer or implementer, access to $1,000,000,000 of bank
accounts for an attacker


I'd have phrased it differently than Perry did.  I'd say that the 
attackers are often cleverer *about security* than protocol designers, 
because insecurity is their specialty.  Ordinary protocol desingers are 
good at designing those protocols, but they haven't been trained to 
think about security.  Here's how I put it in my talk at the IETF 
plenary last night:

\ns{Patterns of Thought}  
\item   Serial number 1 of any new device is delivered to your enemy.
\item   You hand your packets to your enemy for delivery.
\item   Your enemy is just as smart as you are.  If we haven't seen
a given class of attack yet, it's because it hasn't been necessary;
simpler attacks have worked well enough.  (Besides, how do you know
if you'll actually notice it?)
\endns


--Steven M. Bellovin, http://www.cs.columbia.edu/~smb



-
The Cryptography Mailing List
Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]


Re: draft paper: Deploying a New Hash Algorithm

2005-08-04 Thread Alex Alten

Steve,

At 05:34 PM 7/29/2005 -0400, Steven M. Bellovin wrote:
In message [EMAIL PROTECTED], Alex Alten 
write

s:
At 08:12 AM 7/25/2005 -0400, Steven M. Bellovin wrote:
In message [EMAIL PROTECTED], Alex Alten
write
s:
 Steve,
 
 This also seems to be in conjunction with the potential switch over from
 RSA et al. to
 ECC for PKI, etc.
 

Yes, Eric and I have been talking about that, and we'll add some
discussion of that to the next version of the paper.

Variable output is really needed too, say 16, 32, 64, 128, 256 and 512 bits.
And on the wishful side, the ability to optimize compression across
multiple CPUs.


That's completely orthogoal to what the paper is about.  We're talking
about how to convert to *any* new hash algorithm; we're not concerned
with which is chosen.  (I confess, though, that hash outputs of less
than 128 bits don't strike me as cryptographically useful except for
HMAC and the like.)


Sorry for going off on a tangent.

Actually 32 (or even 16) bits is really useful for retrofitting old 
insecure protocols where you
don't want to alter the header size, you only need access control, and the 
packets only exist

for less than 100 msecs.

- Alex

--

- Alex Alten


[Moderator's note: I have to strongly disagree. 16 bits is rarely, if
ever, of any use in authentication in a modern system. Even if you
think something can't live long enough to be spoofed, it usually can,
and as it turns out, attackers are often cleverer than protocol
designers. Crypto is too brittle to play such games with it. --Perry]
-
The Cryptography Mailing List
Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]


Re: draft paper: Deploying a New Hash Algorithm

2005-08-04 Thread Steve Furlong
 [Moderator's note: ... attackers are often cleverer than protocol
 designers. ...

Is that true? Or is it a combination of

(a) a hundred attackers for every designer, and
(b) vastly disparate rewards: continued employment and maybe some
kudos for a designer or implementer, access to $1,000,000,000 of bank
accounts for an attacker


SRF

-- 
There are no bad teachers, only defective children.

-
The Cryptography Mailing List
Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]


Re: draft paper: Deploying a New Hash Algorithm

2005-07-25 Thread Steven M. Bellovin
In message [EMAIL PROTECTED], Alex Alten write
s:
Steve,

This also seems to be in conjunction with the potential switch over from 
RSA et al. to
ECC for PKI, etc.


Yes, Eric and I have been talking about that, and we'll add some 
discussion of that to the next version of the paper.

--Steven M. Bellovin, http://www.cs.columbia.edu/~smb



-
The Cryptography Mailing List
Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]