PKI: Only Mostly Dead

Peter Gutmann pgut001 at cs.auckland.ac.nz
Sat Jun 1 04:18:20 EDT 2002


>Peter Gutmann should be declared an international resource.

Thankyou Nobody.  You should have found the e-gold in your acount by now :-).

>Only one little thing mars this picture.  PKI IS A TREMENDOUS SUCCESS WHICH IS
>USED EVERY DAY BY MILLIONS OF PEOPLE.  Of course this is in reference to the
>use of public key certificates to secure ecommerce web sites.  Every one of
>those https connections is secured by an X.509 certificate infrastructure.
>That's PKI.

  "Opinion is divided on the subject" -- Captain Rum, Blackadder, "Potato".

The use with SSL is what Anne|Lynn Wheeler refer to as "certificate
manufacturing" (marvellous term).  You send the CA (and lets face it, that's
going to be Verisign) your name and credit card number, and get back a cert.
It's just an expensive way of doing authenticated DNS lookups with a ttl of one
year.  Plenty of PK, precious little I.

>The truth is that we are surrounded by globally unique identifiers and we use
>them every day.  URLs, email addresses, DNS host names, Freenet selection
>keys, ICQ numbers, MojoIDs, all of these are globally unique!
>"pgut001 at cs.auckland.ac.nz" is a globally unique name; you can use that
>address from anywhere in the world and it will get to the same mailbox.

You can play with semantics here and claim the exact opposite.  All of the
cases you've cited are actually examples of global distinguisher + locally
unique name.  For example the value 1234567890 taken in isolation could be
anything from my ICQ number to my shoe size in kilo-angstroms, but if you view
it as the pair { <ICQ domain>, <locally unique number> } then it makes sense
(disclaimer: I have no idea whether that's either a valid ICQ number or my shoe
size in kilo-angstroms).

(This is very much a philosophical issue.  Someone on ietf-pkix a year or two
 back tried to claim that X.500 DNs must be a Good Thing because RFC 822 email
 address and DNS names and whatnot are hierarchical like DNs and therefore
 can't be bad.  I would suspect that most people view them as just dumb text
 strings rather than a hierarchically structured set of attributes like a DN.
 The debate sort of fizzled out when no-one could agree on a particular view).

I think the unified view is that what you need for a cert is a global
distinguisher and a locally meaningful name, rather than some complex
hierarchical thing which tries to be universally meaningful.  Frequently the
distinguisher is implied (eg with DNS names, email addresses, "for use within
XYZ Copy only", etc), and the definition of "local" really means "local to the
domain specified in the global distinguisher".  I'm not sure whether I can
easily fit all that into the paper without getting too philosophical - it was
really meant as a guide for users of PKI technology.

Peter.


---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at wasabisystems.com



More information about the cryptography mailing list