I just read an interesting article about a computational model which tries to capture the evolutational aspects of trust within neighborhoods [1] and trust between strangers. Despite some flaws (I just don't like the idea of inventing a computation model to simulate human interaction) it raised the interesting question how neighborhood can be defined, and how this concept conflicts overlaps with the often used term social network. We are usually talking about geospatial information, and that's why we have to be careful applying our (old) view to the nowadays prevailing web-based technologies.

In its usual understanding, neighborhood is determined by spatial proximity. You're neighbors are living in the house next door, or are countries having a border with your country. Your real-life social network on the other hand is not necessarily constrained by distance. Most of your friends probably live close-by, but you might have friends or other contacts actually everywhere. Real neighborhood is defined by spatial proximity, real social network by .., well, that's up to you. Real social network is heavenly influenced by spatial proximity, most contacts are also spatially close.

The internet is challenging this view, and interesting examples exist which support this. Last.fm is a website which allows you to upload (scrobble) the music you are at listening to (at the moment). The website then generates, based on your history of music,  suggestions (music you might like) for you. One feature is the creation of a list of people having similar music taste (which makes it possible to scan through the favorites and discover new music). This list of people is you neighborhood, because their interest music is similar to yours. The thematic distance is minimized here.  Last.fm provides also the option to identify friends, which means to create your own social networks. Experience has shown (and this is reflected in my Last.fm contact list as well) that online social networks are usually overlapping with real life social networks (you tell your buddies about this great new platform, they register, they tell their buddies, and so on..). As already said: your social network consists mostly of friends living close-by. Online neighborhoods are defined by thematic proximity, online social networks overlap with real social networks. And real social networks are highly influenced by geospatial proximity.

Now a special case: online platforms which handle information which is, in some way, related to real objects. Example: experience/rating sites like Dooyoo allow for commenting on restaurants. Before rating a restaurant, one should (preferably) have visited it. And one can assume, that mostly people from the real (spatial) neighborhood of the restaurant are among this list. So, the neighborhood of such online platforms overlaps with your real neighborhood, because the theme here is something within the geographic space. People with same interests (the thematic neighborhood) are mostly people coming from the same region (the spatial neighborhood). Inferring, for example, trust from neighborhoods has to have this in mind before applying any computation models to websites.

[1] The Evolution of Trust and Cooperation between Strangers: A Computational Model, by Micheal W. Macy and John Skvoretz