LSCitter: building social machines by augmenting existing social networks with interaction models

We present LSCitter, an implemented framework for supporting human interaction on social networks with formal models of interaction, designed as a generic tool for creating social machines on existing infrastructure. Interaction models can be used to choreograph distributed systems, providing points of coordination and communication between multiple interacting actors. While existing social networks specify how interactions happen---who messages go to and when, the effects of carrying out actions---these are typically implicit, opaque and non user-editable. Treating interaction models as first class objects allows the creation of electronic institutions, on which users can then choose the kinds of interaction they wish to engage in, with protocols which are explicit, visible and modifiable. However, there is typically a cost to users to engage with these institutions. In this paper we introduce the notion of "shadow institutions", where actions on existing social networks are mapped onto formal interaction protocols, allowing participants access to computational intelligence in a seamless, zero-cost manner to carry out computation and store information.