Tasteweb

Webs of Trust, (or networks of endorsement) are a common social technology with many useful properties; they can grow quickly, they can support a blend of shared structure and local structure, and they can incrementally self-correct with minimal labor. Despite being fairly common in the online world, we identify many still unrealized applications for webs of trust which we expect would greatly empower grass-roots organization of information, news systems, and public dialog. The main obstacle to most of these new functions turns out to be the performance scaling limits of today's graph databases. We've identified indexes and algorithms that would allow us to transcend those limits. The project aims to implement fast shortest path indexes (eg, Contraction Hierarchies, BatchHL+), and "sparse query" indexes (novel) (dynamic unions, or dynamic cache placement), for open source graph databases, to enable several new critical functions for webs of trust: Globally inclusive networks of endorsement, exclusive claims, news discovery, and subjective filtering. Once implemented, we plan to make this functionality available to emerging open source social network protocols and social computing frameworks.