Security through obscurity

I know that it is pretty much accepted by the security
community that security through obscurity is useless. There
are countless references to this, and it is a strong
argument in favor of publishing algorithms. 

From time to time, however, it is healthy to question policies
such as these and ask ourselves whether or not this is a good
idea. It seems that Denning has changed her mind on this by
advocating the secrecy of the skipjack algorithm. 

All of the known plaintext attacks on algorithms such as
DES, that involve exhaustive key search are based on knowledge
of the algorithm. Wouldn't keeping the algorithm a secret
render such chips, as the one presented in Eurocrypt '93
useless?

I agree that analysis is more difficult when you don't know
if an intruder has compromised your algorithm, but if it were
my data that I wanted kept secret I wouldn't give the
cracker a head start by publishing my algorithm. Clipper has
proven how difficult it is to reverse engineer an algorithm.

Let me just say that I know I have violated an accepted
doctrine of computer security, but I think it's okay to
question even the most accepted ideas every once in a while.
I thought this would be a bit provocative.

Wonderer
-------------------------------------------------------------------------
To find out more about the anon service, send mail to help@anon.penet.fi.
Due to the double-blind, any mail replies to this message will be anonymized,
and an anonymous id will be allocated automatically. You have been warned.
Please report any problems, inappropriate use etc. to admin@anon.penet.fi.

an41418@anon.penet.fi (wonderer) writes:
> [regarding security through obscurity and the clipper/skipjack stuff] 
> 
> All of the known plaintext attacks on algorithms such as
> DES, that involve exhaustive key search are based on knowledge
> of the algorithm. Wouldn't keeping the algorithm a secret
> render such chips, as the one presented in Eurocrypt '93
> useless?
>
> [...] I wouldn't give the
> cracker a head start by publishing my algorithm. 

You are not going to be able to keep your algorithm secret, period.
Those who are determined enough will be able to dig it out of any
programs or chips you use to implement your algorithm.  Security through
obscurity is stupid because no matter how smart you may think you are
in hiding your method, there is always someone smarter who will dig it 
out and changing technology constantly lowers the barrier of how smart 
people need to be to dig information out of old locks using new tools.

The problem with security through obscurity is that if it is set up as
a part of a system people have a tendency to think that they can use
the obscurity to hide weaknesses in the system.  Once your system or chip
is out there in quantity there will be nothing on it you can hide, and any
weaknesses will be impossible to fix without a recall (severely damaging
trust and reputation with your users...)

If the algorithm is truly secure, then very little will be gained by
letting everyone know the method used, and by letting others examine and
test the system it is possible to have others establish the strength or
weaknesses of your method for you.  I do not trust you, I do not trust the
NSA, I do trust the cryptography community in general because they have
nothing to gain by misleading me.

> Clipper has
> proven how difficult it is to reverse engineer an algorithm.

Clipper has proven nothing because _no one has had a chance to reverse
engineer the algorithm_.  Give me clipper in software or give some of the
people on this list the clipper chips and see we'll see how long the method
remains secret.  So far clipper/skipjack is vapor.  When I can hold one in
my hand and tweak inputs and outputs then we shall see how difficult it is
to reverse engineer an algorithm.

jim