[pkng] Notes from the informal lunch in Hirohshima

[Paul Hoffman <[e-mail]>]

Very informal (and very delayed) notes of a very informal PKNG lunch
2009-11-11, Hiroshima

The lunch was organized as a last-minute affair, spurred by Leif Johansson. ISOC (through Lucy Lynch) kindly found us a room and paid for box lunches.

Those in attendance were:
'Bob' Morgan
Carl Wallace
Jeff Hodges
Karen O'Donoghue
Leif Johansson
Lucy Lynch
Masaki Shimaoka
Paul Hoffman
Peter Saint-Andre
Russ Housley
Ryu Inada
Sam Hartman
Sean Turner
Stephen Farrell
Tim Polk
Yasuo Miyakawa

These brief notes were just snippets of the conversation. The following should make it clear that there are many ideas floating around for this work, and some need to be fleshed out more in this RG.

People who made statements that sound like something below that want to expand or correct them should do so in a separate thread.


Leif: Let's turn the certificate structure inside out
Sam: Is this an xmf with some public keys?
Leif: A trust infrastructure with entities can be in multiple rings
Stephen: Our work might include symmetric keys
Paul: How? This is about public keys
Russ: Enrollment approaches that create a binding 
Sam: Wants terminology for following bindings
Sam: In SAML, hard to follow chain of trust and consequences of running the protocol
Carl: Define migration from current certs, then come up with another term
Bob: Don't start with glossary, start with a model
Jeff: What are our use cases?
Lucy: Start with box of TinkerToys
Peter: Make the Internet a higher trust environment
Peter: Want more use of certificate auth
Leif: Your relationship with Wikipedia is different than employer; use different keys
Leif: But there is one identity holder
Sam: But there are privacy concerns
Leif: There is a difference between your credential and what you send to the relying party
Stephen: Maybe you only send the key, and the other party gets the auth in some other way
Russ: Try to bind attributes to different contexts
Yasuo: There are federation back-end databases that are not just identity
Russ: SPKI uses the public key is your identifier: what happens next is up to the relying party
Russ: In HIP, you have an easy proof that the public key is yours
Ryu: A device has security and identity; maybe we should have a PKI version of DHCP
Ryu: But it must be easy to provision
Leif: Is naming in scope? If we extend far, how will it affect our mission?
Sam: We'll have to pick some form of naming
Sam: Some choices will impact what we can do with this work
Sam: People have different trust in attribute names and attribute values
Stephen: We can look into embedding names in keys
Jeff: Understanding proper terms: names, identifiers, identities
Lucy: We can pass information about where something succeeded or failed
Lucy: Also, can glue information about interactions
Peter: How will our technology change so that people can use this stuff?

--Paul Hoffman, Director
--VPN Consortium