Trust, a word we use in place of trustworthiness comes in many flavours which have a bearing on how we calculate and quantify it.

Trust can be divided into two ‘top’ catagories.

• Global trust- this is the opinion of the network as a whole as to the trustworthiness of the user. Global trust ratings are not context specific; every user sees the same rating
• User trust is the relationship of a specific user to another users. The trust value will depend of the context it is assessed in. e.g the details of the relationship and transactions engaged in, levels of risk and the trust network paths between the users.

Global Trust

In the trust machine this has three components

• Reputation, calculated by the EigenTrust algorithm (more in another post) this is a measure of how well connected a user is in the trust network – their trust centrality
• Provenance is a measure of how confident we are that the user is who they say they are, I.e.how much we know about the individual behind the user, how transparent and accountable they are. A user with low  or no provenance could easily have been set up specifically for to hide a poor trust rating form prior behaviour. Provenance trust is obtained using the APIs of other well established social network providers and is higher for high centrality accounts which have required time and effort to build. Only one trsut machine user can ever be linked to each account.
• Badges are awards made by application providers to users. Applications providers may choose to award local badges which will affect trust ratings within their applications, or global badges which conform to Trust Machine guidelines and affect Trust Machine ratings across applications

User Trust

This comes in two varieties:

Direct Trust is derived from direct experience of interactions between users. A simple rating system can be used to rate posts and transactions

Indirect Trust, where users have not interacted is inferred by their position within a trust network graph. Graph is a directed network with users as nodes joined by edges which are direct trust relationships. each edge has a label which quantifies the relationship. This is calculated by algorithms which analyse paths between users.