private key protection

Jerome Baum jerome at
Tue Oct 18 15:08:15 CEST 2011

On 2011-10-18 14:48, Peter Lebbing wrote:
> On 18/10/11 14:36, Jerome Baum wrote:
>> * I'm going to take the word to mean what it says: "key", not what I can
>> flexibly interpret it as: "encrypted key".
> One of those metal things in my pocket? What good are they for encryption? Even
> if you manage to read it in, it still has way too little entropy...
> Or in other words: the word has multiple meanings. If you want to discuss stuff,
> you need to determine which of those meanings you're talking about, not attempt
> to constrict the definition of the ambiguous word. That'll only lead to
> bickering about which definition is the correct one.
> Peter.

Makes sense if there's no context. But there's context here --
"cryptography". In that context, key means something specific.

Say you're discussing search trees (the data structure) and someone
comes up and starts talking about how binary trees are so efficient.
Then I come along and say "hold on, binary trees aren't necessarily
balanced, so the search time can even be linear". What's ambiguous here?

Now someones comes along and says "that's just stupid, obviously a
binary tree is a balanced binary tree, and if you meant a binary tree
that could be balanced or unbalanced then your statement is trivial".

In the context of the discussion (computer science), the "binary tree"
isn't a piece of wood with leaves [that someone cut in half -- "binary"
:)]. Even if we take "binary tree" at face value. Just like "key" in the
context of cryptography doesn't mean a piece of metal, even at face
value. (A physical key would usually be a "physical key" or something of
the kind.)

It's one thing to be picky when it adds to the discussion proper. That
would be the case when we're distinguishing between the key as it is
stored on disk (encrypted, inside a key-file/-ring/...) and the key as
it is stored in memory (unencrypted). That distinction is important when
considering your attack vectors.

But the distinction between a physical key and a cryptographic key isn't
adding value to the discussion proper. It's being picky for the sake of it.

PGP: A0E4 B2D4 94E6 20EE 85BA E45B 63E4 2BD8 C58C 753A
PGP: 2C23 EBFF DF1A 840D 2351 F5F5 F25B A03F 2152 36DA

More information about the Gnupg-users mailing list