Pymmetry has a Maximum Flow (Ford-Faulkersson) implementation
at its heart, which is used to provide an easy means to evaluate
Trust Metric Certifications, in python.
Trust Metric Evaluations, essentially, are Cascading (or,
Hierarchical) Access Control Lists. Pymmetry provides an
easy way to integrate Trust Metrics into any project.
Practical applications for Trust Metrics include Domain Name
Registration protection, Web-site Forum / content access
rights and privileges evaluation, and any other potentially
hostile environment in which it is necessary to distinguish
trusted from untrusted entities, taking into consideration
the opinions of entities in the field.
[The full article is available as the README at
the project has only just been set up on sourceforge, so it will take
24hrs for the dns to be updated for pymmetry.sourceforge.net to work.]
The original code for Pymmetry
has gone via
from mod_virgule, by Raph Levien <email@example.com>. Raph studied
Trust Metrics, and wrote mod_virgule - a community site
forum engine - as a social experiment which has proved highly
successful and effective at its job. Namely, it promotes and
protects a site's purpose from unsolicited interference by
empowering the users, in a hostile environment [e.g. the
internet], to select those people that they trust within
their community to remain honour-bound to the charter of
the site they are using, or risk the wrath of their peers -
ultimately expressed by the revokation of the Certifications
their peers gave them, with the inherent loss of access rights
such revokation implies.
Raph's original Network flow simulation code was written
in Java. For mod_virgule's purposes, he rewrote it in c.
Pymmetry is a python implementation, which is considerably
more flexible, cleaner and easier to understand than c.
The original purpose of Raph's Trust Metric code was to fulfil
a similar aim to that of Keynote. Namely, that in a large
hostile environment, you have to know who to trust (and the
example in his paper was the issue of Domain Name registration).
When you have a chain (or web) of trust, you have to be able
to evaluate that chain, and be as certain as possible that
the web has not been compromised, in order to make decisions.
Keynote focusses, effectively, on "a means to securely evaluate
Digitally-signed Logical Expressions". Whilst Raph's work is
known, because of http://advogato.org,
for focussing on the web of trust and its evaluation, Raph's
original paper does cover "Logical Expressions" as well,
of which - it turns out - Keynote is a superset implementation.
There has been much discussion recently about Trust Metrics -
most notably on http://advogato.org,
the original Open Source Advocacy site, set up by Raph.
It is hoped that Pymmetry will add fuel to the flames :)