One of the ironies of the popularity of the relational data model is that its popular application is often very focused on entities (and, therein, the popularity of entity-relational diagrams). But, one way—the more pure way, to look at it is: there are no actual entities, only relations.

[mashup:] So, there is this interesting new site, Operating Manual for Social Tools (already aka "OM"). It just started, and every post so far is provocative and great.

I came across this site via danah boyd, whose sociability first, technology second is cross-posted on OM. Somewhat in response to this, David Weinberger has posted on OM, Groups as atoms. (Both are great.)

danah and David's posts both pose questions and considerations stemming from observations / assumptions about who artificial social networks (e.g., Friendster) are designed for. Are these designed for individuals, for groups, for . . . ? (This consideration is developed in earlier posts on OM: danah's considering the goals of social network modeling and David's The cross-purposes of social networks.)

Getting back to the relational data model, what I think is of interest in regards to these articles on OM is the relational model's recognition of relationship as the context of data, rather than entity-ness. In other words, the context of data is in relationships (between people, things, other information, etc.) rather than in the so-called entities on which we see the data as attributes.

What I think is one of the issues that danah and David are addressing is that social relationships are appreciated / used in ways which can not be attributed to entities. (The whole is greater than the sum of its parts.) Rather, I believe the value is specifically in relationships.

Artificial social network websites can be seen as environments in which social networks are happening, or are encouraged to be happening. We develop relationships with these environments.

So, social software can be looked at as fitting into a sphere of relationships between people and each other, but also relationships between people and the software (or, environment altogether). And, that total sphere of relationship, inclusive of the relationships with the environment, is the who that the social software is ultimately about.

I imagine this means that no "object" or "entity" (or "atom", to reference David's term) can be seen as the fundamental subject of the design of social software. And, also, there is a liability in breaking the social system down into component parts and trying to analyze them as entities or objects. (See also: Adina Levin's Computer as Door.)

Relational systems have not had interfaces that adapt to all of the nuances of the represented relationships (which is part of the reason why, in practice, they are so much about entities rather than relations). So, all of this is not to say that social software simply need emulate RDBMSs.

But, I think there is a fundamental idea to explore in how one interfaces with an evolving sphere of relations, and that the relational model provides what I think is a useful context of experiments in this regard. And, at the same time, I'm cautious about ascribing too much solid-ness to any entity in social software design.