A few days ago Marc Canter sketched the idea of a technology to interconnect social networks together, via a post on the Pho list, and it caught my eye. “Instead of a giant centralized social network with 1,000,000 members, we’d prefer to see 1,000,000 social nets with 10-25-150 members each”, he wrote, referencing work done by his company Broadband Mechanics.
That got me thinking, and this posting is an attempt to capture some of the thoughts triggered by that vision — even if some of them are tangential to what Marc is trying to build.
It was an online community that first got me connected to the Internet about 14 years ago: the Hicom conferencing system, for people working in Human-Computer Interaction. The shift in terminology from the ‘online communities’ of that period to the ‘social networks’ of today reflects a change in the character of group affiliations online.
The early online groups earned the tag of ‘communities’ as, to take Hicom as an example, the only means I had to get online was via Hicom, and over 95% of my communications were with other Hicom members: this gave the community a clear identity and made it more close-knit, even allowing for the fact that many of its members were from universities and also had access to the JANET network for email among themselves.
When the Internet took off outside academia, the number and range of online forums increased dramatically. I think of the effect of this in terms of entropy: with more spaces to gather, and more choice in affiliations, the ‘energy’ of group discussion became more dissipated; instead of a few ‘hot spots’, there were a large number of ‘tepid’ areas. The term ‘social networks’ captures this nature of more permeable social groupings, with looser affiliations.
If this metaphor has any power, then it means there is no point in trying to fight entropy across the board — the laws of physics forbid it — by creating a “giant centralized social network”, but what Broadband Mechanics offer is a means of creating some local order across multiple networks.
The name they give to the technology for doing this is Digital Lifestyle Aggregators. The concept seems to be a set of standards and open source APIs to support aggregation of person-related data across social network sites, blogs, personal media collections (like photos in Flickr), electronic and mobile communications. There are some further details here and here, though I wish there were some more straightforward use cases of DLAs in action.
In the absence of such use cases, I don’t pretend to understand DLAs fully, and what follows may therefore miss that target. But the gist of machine-readable aggregation of personal communications and content leads me to three related questions.
The first is about the circumstances under which people are willing to have lots of personal data about themselves distributed ‘frictionlessly’. Many people want to keep a degree of separation between their professional social networks and their personal ones. They want to control how they present their identities in public. If people are cagey about sharing the titles of music they listen to, how much more sensitive will they be about sharing their lists of contacts, or enabling all their contributions to different social networks to be aggregated?
The technology may include options to protect privacy, and I guess people could maintain different online aliases to separate different aspects of their identity. But both of these steps would undermine, to some degree, the core purpose of the technology.
The second question I have is about the (lack of) richness and utility of the machine-readable data. Take the example of Friend-of-a-Friend (FOAF), a vocabulary that gives “a basic expression for community membership: describing people and their basic properties such as name, e-mail address, and so on”, and is cited as one of the building blocks of DLAs. As a member of the Ecademy social network, a FOAF file of my Ecademy contacts is generated automatically. Some of these contacts are people I know quite well, others I’ve exchanged no more than a brief “thanks, but no I’m not interested in your service” message — the FOAF file has no means of distinguishing between these. So how useful is it to anyone? Again, the richer and more useful the data was, the greater the privacy concerns that would be raised.
The third question is how people-centred data can be integrated with data-centred data (i.e. metadata). The spread of entropy means that many of the same topics are being discussed separately in different social networks. If I read an interesting message on a social network, I might be interested to find out more about the writer, but I might be equally or more interested to find other opinions on the same or related topics on other networks. Aggregation would be helpful if it could help me find those other opinions — much as services like PubSub help me track all mentions of particular terms on blogs.
Something tells me this stuff is still some way away. After all, these few half-baked thoughts started off with a message on a ‘social network’ that runs on a good old-fashioned email list.