I have lately found my own opinions to conflict with those of some others regarding the FOAF (Friend-Of-A-Friend) RDF profile spec. I've held that once it matures a bit (today, one index reports 156,912 separate semantic web term definitions), and once social networking platforms begin to consolidate, FOAF will be an immesurably helpful technology for transmitting profiles. Perhaps also as a sort-of social vCard in the interrim. What I haven't expected thus far has been widespread, unfettered application of FOAF by social networking platforms eager to let users export entire profiles and relationship webs to the newest network on the block. Intuition suggests that profiles alone cannot fuel a leading social network, but it also makes sense that smart companies will protect their profile databases as intellectual property. In turn, it seems not to make much sense for companies to knock down exit barriers and empower users to hop seamlessly between platforms. This all goes double for established platforms such as MySpace and LiveJournal whose databases hold the most information worthy of protecting.
These dynamics might be different, I have added, from those of RSS syndication feeds offered by almost every blogging platform today. Yes, competitive pressures forced the widespread adoption of RSS once Blogger began to offer the service for free. RSS is compelling for blog authors because it makes posts more accessible to blog readers, but that technology doesn't make it any easier for either to switch platforms entirely. FOAF import/export features clearly do, and so we should expect social networking platforms to avoid fully implementing them.
Or so I have argued. And the jury is still out here: MySpace, which WikiPedia reports having had more than 20 million users as of June, 2005, doesn't use FOAF.
As it has happened, though, a number of big players have begun to experiment with the technology. LiveJournal has published every one of its profiles as publicly-accessible FOAF since early 2004, for example. Ecadamy has followed suit, and TypePad has even gone so far as to export its FOAF profiles to the now-defunct Plink. (Check out Alex's comments on Plink in response to a previous post). Typepad still generates FOAF profiles automatically for each of its users. Just this past week, Tribe has announced that it's taking the next step: full FOAF import/export functionality. Now, according to Wikipedia, Tribe had about 300,000 members as of June, 2005. LiveJournal reports 7,732,113 users as of today, 295,310 of whom have updated their profiles in the past 24 hours. Clearly, each of these companies holds a wealth of profile data. So what's going on here -- have they decided that profile data are worthless to them, or do they simply not expect FOAF to trigger a mass user-exodus?
One striking anecdote may shed some light here. In analyzing a 2004 FOAF-web crawl (actually called a "scutter dump" with respect to FOAF and other RDF docs) John Paolillo and Elijah Wright from Indiana University found that despite a reasonable initial distribution, the scutter was drawn to a disproportionate number of LiveJournal FOAF profiles. The dump was crawled and recorded by Jim Ley, by the way, and has been used for a number of SN analyses. It's still available for download in MySQL format at http://www.jibbering.com.)
Here is how it worked out:
| LiveJournal | Other | |||
| Date | Visited | Not Visited | Visited | Not Visited |
| 3-Mar | 663 | 0 | 1607 | 0 |
| 4-Mar | 13940 | 121408 | 160 | 23 |
| 5-Mar | 2810 | 17648 | 1279 | 130 |
| 6-Mar | 11782 | 60844 | 0 | 6 |
| 7-Mar | 4347 | 22650 | 0 | 1 |
Even though the ratio of LiveJournal to 'Other' profile hosts began at 663/1607 in the first round, it had risen to 26,997/1 by the fourth round. John and Elijah hypothesized that "a large social networking site like LiveJournal presents a kind of "black hole" from which the scutter will have little chance of escaping. They even generalize their understanding to cover user behavior, suggesting that "large social networking sites effectively control their members' social capital." Because LiveJournal controls users' interface with FOAF, it still strongly biases them to connect with other LiveJournal members despite the notionally open FOAF implementation. So LiveJournal users have stuck around for some reason other than proprietary "profile-lock".
Even in light of the above example, the question remains to what extent LiveJournal users have used FOAF export features to add their own profiles and relationships to other social networking sites. While that's an important question, the anecdote does make a strong case for other assets' resiliency to FOAF. It suggests, for example, that the eyeballs and updates of users may not be as capricious as are profile data themselves. Established social networking sites also retain the promise of potential new connections, especially along dimensions such as education or musical interest which aren't covered by the core FOAF spec. Access and privacy controls may also prove important here. Platforms who decide on behalf of users to make profiles publicly accessible as FOAF open themselves to the sort of privacy-invasion allegations that sank Plink.
It will be fascinating to see whether full-featured FOAF implementation mestastisizes beyond isolated examples and cascades throughout the social networking world. I still suspect the reach of FOAF as designed today will be limited, and we may even see some feature backtracking as users begin to learn about the perils of an uncontrolled spec. What do I know, though? If I am entirely wrong in this regard, and FOAF spreads like wildfire, I think it's clear that established platforms will be able to maintain their leads for a while. The value of assets beyond simple profiles will start to become even more clear in that case.
Comments