*PKI Labs Conference Call*
February 10, 2003

*Attendees*

Neal McBurnett (convener) - Internet2

Sean Smith - Dartmouth

Eric Norman - Wisconsin

Peter Honeyman - Michigan
Carl Ellison - Intel
Renee Frost - Internet2

Ben Chinowsky (scribe) - Internet2

*Discussion*

The group reviewed the minutes from the previous call; the references to "cross-site scripting" need to be removed. [AI] Ben will revise the January 13 minutes.

Sean noted a few highlights from the work at Dartmouth:
- Kunal is finding that many previously-successful signature-hacking tricks don't work with Silanis and Office XP. These products turn off an auto-update field; the approach he's taking to overcoming this obstacle is to use a trusted macro to turn it back on.
- In order to get FIPS compliance, Sean has included DSA in his S-BGP implementation. He's looking for a toolkit that does DSA precomputation, as this can provide a significant performance increase.
- Sean noted that he'd had a chance to help "a certified genius" with PGP, and noted this individual's initial confusion between the passphrase used to protect the private key and the private key itself. Carl suggested that this was appropriate for a genius, as the private key is "just a proxy for the password"; Peter replied that this is true locally, but not after the public key gets into databases worldwide. Sean said that it seems like today, for encrypted mail among academic colleagues, PGP is a better choice than S/MIME, assuming there's an out-of-band method (e.g. phone calls) for establishing trust.

This led into a "what is trust" discussion. Carl argued that without out-of-band communication, there is no trust -- the trust relationship is something that happens in the physical world, which then gets mapped into the digital world. Peter offered a definition of trust as "that which you cannot confirm but must assume". Eric suggested that the relying party need not use a method of verification, but must know that one is available. Sean noted that much of the sociological work in this area (e.g. Trust & Trustworthiness by Russell Hardin) seems to suffer from "inappropriate quantifying". Eric asked where private keys fit into the "something you have, something you are, something you know" trichotomy. Peter cited the example of a password -- which if written is something you have, but if memorized is something you know -- to argue that this taxonomy is "something that lives in the social domain, not the technical". Peter said that he'd heard of a company trying to derive keys from thumbprints, and Carl noted that they'd had to increase the false acceptance rate to an unacceptable level to make that work at all.

Thirty papers have now been submitted for the Second Annual PKI Research Workshop. The group brainstormed topics for panel discussions; suggestions included bridges, privacy and anonymity, and digital rights management. There was general agreement that, in accordance with the tradition established last year, the panels should include as wide a range of perspectives as possible, in particular including representatives from the Federal Government along with others offering less traditional approaches to PKI. Peter suggested that the group decide on the panel topics after reviewing the papers, as reviewing the papers will illuminate what the "hot button" issues are this year.

The next meeting will take place March 10 at 4:00 PM Eastern, per the regular schedule.

*Action Item*

[AI] Ben will revise the January 13 minutes.