At 11:25 PM 3/5/00 -0500, Phillip Hallam-Baker wrote:
>
>I note that you are not signing your emails with PGP, wheras
>I sign every one of my messages with S/MIME (except for those
>I send from the PalmVI or RIM which I don't yet have an S/MIME
>client for).
You can sign a message with one click, with Eudora and PGP.
It would make deniability tougher. No one is currently spoofing anyone's
posts here, so there's little need.
>I sign every one of my messages because S/MIME makes that easy.
Your signature links some private key to the message, but
there's no person necessarily associated with it. You may
be the nym used by a committee, or different people over time,
or a bot. Signing a message doesn't change this.
>PGP is in my view popular with people who want to have absolute
>control over their environment - even if that is at the expense
>of security. To use PGP securely, one pretty much has to only
>use keys signed by people you know are meticulous in checking
>credentials.
No. To manage your trust you must trust introducers.
That is social. How you trust your associates is your problem,
always has been. PGP lets one nym introduce
another, but what you trust these nyms for is up to you.
Nyms have reputations. PGP lets nyms identify their own
speech. Any meat-legal structure you layer on top of it
is necessarily locale-specific.
What is a 'credential'? You mean signed statements like,
"This here character who calls himself 'Joe' and uses key <fingerprint> did
a good job painting my house. My nym is Jane Doe 876 and you can check my
reputation here..", right, not the numbers the government brands you with?
Which government?
Why trust them? (Perhaps particular governments will only
use their force if you transacted with nyms they know about.
Relying on meat-force is your choice. Nyms only have location
if they chose to.)
You can have multiple identities in different contexts.
Linking them is your choice. You don't need meat-identity
to do business in the future.
In my case that means I only use keys signed by
>Jeff Schiller. Now I have the advantage of actually knowing Jeff,
>but for the life of me I can't see the scalability in that
>solution. What do I do if I want to speak to someone who hasn't
>yet met Jeff - buy them an air ticket to Cambridge MA so they
>can meet him?
If you want to speak to Alice who doesn't know anyone in common, exchange
public keys when you next meet.
Or, more conveniently, plaintext email the keys, and then call her and do a
voice authentication and verify the fingerprint (or read the whole key).
Or, if you like getting encrypted spam, post your email address and public
key on a server.
Since you probably have *several* public keys under *several aliases*, post
'em all. Linking any of these to any physical
meat identities is an orthogonal option.
>PGP is unfortunately not scallable to commercial usage.
You forgot the smiley.
> Unfortunately
>PGP is only about privacy. PGP does not provide any meaningfull
>or usefull statement about identity.
Correct. It does one thing well. What you call 'identification' is not
what it is for. Linking atoms to public
keys is not what its does, but since you can sign things, you can
use PGP to help you link to meat, if you want to.
Speaking privately is not related to knowing who you're speaking to.
The integrity capabilities
>of PGP are as a result not usefull if one wishes to provide any
>degree of assurance with respect to the enforcement of digitally
>signed contracts.
To link atoms to keys (for legally binding the atoms
to actions performed with the keys) you bring the atoms to a mutually
trusted party and demonstrate that both the atoms and the key-holder agree
to the binding. You needn't tie any of
this to government ear-tags. The mutually trusted party
performs the domain-specific authentication of the atoms
and confirms that the meat agrees to what is being signed
with the key. Any group of mutually consenting people can
agree on a meat-verifier suitable for them. You can use
PK methods to sign these certs but its not a PK problem.
dh
