Global trust metric

From TrustLet, a free, collaborative project for collecting and analyzing information about trust metrics.

A global trust metric is a trust metric where trust values for nodes are not personalized, i.e. they do not depend on your own position in the network. With local trust metric, the trust values you see for other nodes depend on your own position in the network.

Examples of global trust metrics are PageRank, Ebay, Advogato (at least with a fixed sets of seeds), and actually most of the metrics used in real systems.

[edit] Popularity

Global trust metrics may be the most popular in use because they are significantly simpler to compute compared to local trust metrics. A global trust metric tracks one trust value for each node on the network. This means as you add more nodes, you add the same number of trust values. The growth in trust data is linear in contrast to local trust metrics where the growth in trust data is exponential to the growth of nodes on the network.

[edit] Ebay

Ebay feedback is an example of a global trust metric. After each transaction both buyer and seller can leave positive, neutral or negative feedback. These scores are totalled and presented as a measure of "trust". For example, a seller with 50'000 feedbacks which are 99.9% positive would be considered highly trusted. While a seller with 10 feedbacks which are only 40% positive would be considered untrusted.

This is a global trust metric because the feedback score for a user is global, it's the same for everyone on the system. All users will see the same feedback score for User A. While in a local trust metric each user sees a different score for every other user.

[edit] Reputation

Global trust metrics are essentially "reputation systems", they track an entity's global reputation. They provide a sense of what "the whole community" thinks of an individual node.