Shostack + Friends Blog Archive

More on LLPersonae, Identity Oracles, and RCSL

Adam: But applying for a job is exactly what you describe, “organizations
with whom you don’t have a lot of history and interaction.” For an
awful lot of people, they apply for jobs broadly. One cashiership is
as good as another.
And there are a lot of places where I’d like to protect my privacy.
The Red Cross requires your SSN if you want to volunteer. The DMV
wants it. Dave Birch talks about brands and reputation in a comment,
and I think that there are a lot of places where The Presentation of
Self in Everyday Life really comes into play, and you want to present
one front or another.
How could we extend LLPersonas to reduce the demand in these other
situations?
Bob Blakely: think an Identity Oracle – not an LLP – is what’s desired when applying for a job. It would work like this:
I apply for the job. The employer alerts me that background investigation will be required. I direct the employer to my Identity Oracle.
The employer says to the Oracle “For this job I require a US citizen with no felony arrests, no felony or misdemeanor convictions, a valid motor vehicle operator’s license, and a good driving safety record. Is this applicant eligible for the position?”
The Oracle then uses its information (and perhaps supplementary information it develops through investigation; this information will be protected under my existing personal information protection contract with the Oracle) to answer the question.
[At which point I jump in by assuming RCSL is primarily about reputation. Bzzzt!]
Chris: What work have you done regarding the Relational Continuity Sockets Layer?
There has been a ton of work done regarding reputation management among
autonomous agents, protocols for distributed reputation systems, etc. Can
you provide an example of the RCSL in action that shows its distinctive
properties, and why they are important/useful?
Bob Blakely: [I]t’s important to say that RCSL isn’t really a reputation management system. It’s a way to build relationships which are scoped in time and also in committment of resources. A useful analogy here is that of a card game; at each round one antes a set of resource to qualify for the round, and one obeys certain rules for the hand. In an analogous way the RCSL enables creation of predefined relationships which have both rules (for what can & can’t be done in the relationship) and roles (for which participant can do what at any particular time; compare this with a game like Bridge where the roles for the players differ).
The notion here is to design the resource committment rules, rules of play, and roles to limit risk of all types to the various parties.
The game may require reputation as a condition of entry, and it may change reputation as an effect of the outcome, but in this respect reputation is simply another resource – and is not the only kind of commodity which could be used.
[Stay tuned for more on RCSL in the next post from our interview with Bob and Mike]

Originally published by cwalsh on 13 Oct 2007
Last modified on 13 Oct 2007
Categories:   ID Management   Uncategorized