On Tue, Apr 5, 2016 at 1:15 PM, Nathan Andelin <nandelin@xxxxxxxxx> wrote:
Whether "your"
algorithm can be cracked or not depends on a lot of factors. So I reject
the notion that ONLY published algorithms should ever be used.
That's fine. I don't think it's worth debating that particular point
anymore. There is nothing *mathematically* preventing the existence of
an algorithm developed in secret which is at least as secure as any
public algorithm. The issue of whether such a thing could exist in
practice, and how likely that is, amounts to a philosophical issue.
If you're comfortable with your secret algorithm, fine, use it. If you
believe it's stronger than any public algorithm, then great, use it
smugly.
Others may not be comfortable about the cryptographic strength without
that public review. And of course most would not have access to your
secret algorithm, nor would they have the expertise or resources to
develop their own secret algorithm that is verifiably as secure as a
proven public one.
The fact that you "hide" your keys should be viewed as
"security by obscurity", under the definitions and parameters discussed in
this thread.
No, it shouldn't. Did you read any of the stuff people have cited?
The parameters that many of the folks in this thread are using are the
ones which come up when discussing Kerckoffs's principle. I'll provide
the Wikipedia link (again):
https://en.wikipedia.org/wiki/Kerckhoffs's_principle
It would be meaningless to even talk about encryption or cryptography
without ANY hiding at all. It's right there in the word: "crypt" comes
from Greek and means "hidden" or "secret".
So let's dispense with willfully being difficult regarding
terminology. Secret keys are not "security by (or through) obscurity"
in the way that we are discussing.
Kerckhoffs (the man) was specifically talking about military ciphers.
He gave six design principles, a lot of them having to do with ease of
use. Most of the original six principles are obsolete today, but the
one that stuck is
"It should not require secrecy, and it should not be a problem if it
falls into enemy hands."
The "it" there specifically refers to methods (or devices) related to
encoding and transmission.
The point being, the "expensive" part of the system should be designed
such that *even if* it falls into enemy hands, our information
security is preserved. If the enemy does manage to steal a key, that
will be of limited use because we will simply change our key. (We
won't have to design and build a whole new encryption device.)
The Kerckhoffs article even has this to say about its application:
"In accordance with Kerckhoffs' principle, the majority of civilian
cryptography makes use of publicly known algorithms. By contrast,
ciphers used to protect classified government or military information
are often kept secret (see Type 1 encryption). However, it should not
be assumed that government/military ciphers must be kept secret to
maintain security. It is possible that they are intended to be as
cryptographically sound as public algorithms, and the decision to keep
them secret is in keeping with a layered security posture."
In other words, the stuff your dad worked on is ALSO *not* security
through obscurity (at least not by design), in the way we are talking
about.
Finally, immediately following the above quote, the Kerckhoffs article
even has its own section on security through obscurity, as well as a
link to an entire article on the subject:
https://en.wikipedia.org/wiki/Security_through_obscurity
To sum up:
1. No one believes we should *rely* on obscurity.
2. Obscurity could, in principle, provide *another layer* of security
for a cryptographically sound algorithm.
3. Most of us cannot know how sound a secret algorithm is.
4. Most agree that the surest way to achieve the highest possible
*cryptographic* soundness is to employ public review.
5. Some are more comfortable with a single layer of proven
cryptographic soundness than with an obscurity layer plus a layer of
unproven cryptographic soundness.
John Y.
midrange.com
