sxsw: leveraging solipsism (Liz Lawley)

Quoting from sxsw: leveraging solipsism (Liz Lawley, who once again took much better notes than I did)

Unfortunately, two of the original three speakers for this panel—Stewart Butterfield and Peter Merholz —couldn’t make it today. Jeff Veen is moderating, and Tantek Çelik, Don Turnbull, and Thomas VanDerWal are the participants.

Jeff Veen starts by framing the context, since the title is…well…somewhat oblique. He points out that tools that help us manage information are becoming more socially aware., for example, which allows you to discover people as well as information, and to discover information based on people rather than simply topics. Last year social networks were all the rage; but he felt that tools like Friendster were like yearbooks—fun and useful for showing off who you know, but that’s a short term activity that doesn’t sustain long term interest. It gains ongoing attraction once you add in the kind of value-added media that tools like Flickr (and, I’d add, provide.

He makes an important observation—what’s most interesting here is the blending of public and private. That needs more elaboration, I think it’s a key concept. He also talks about the need for more interoperability between these systems. Can travelocity, for example, know where he is and share that information in useful ways with other systems I’m on (like flickr, for instance).

Thomas VanDerWal is up first, and discusses personal views of information. Too much online information is ephemeral—so we end up emailing things to ourselves, copy and pasting into new documents and losing context. We need a way to get back to information we’ve seen. (Reminds me of Microsoft Research’s “stuff I’ve seen” approach to searching.)

He says that we “get lost early” in the information around us, and ask how we can get to “findability” in our own information spaces?, for example, allows us to name things in ways that make sense to us. But how do you tie different personalities together? How do we jump between disciplinary vocabulary boundaries?

Our current tools don’t support us well. (His slide is titled “that synching feeling”) Synchronization frequently makes mistakes and overwrites inappropriately. We need a “mothership of information” to tie together our various devices and collections of information.

How do we build a “personal infocloud”? Many requirements. It has to be portable (or ubiquitous), the access appropriate to the context, organized in a way that makes sense to the user in the context they’re in.

External storage and management is important. We need smarter aggregation, attention.xml for everything on your own hard drive as well as the online sources we’re following. What’s important? What should I be focused on? Need standard formats for being able to pull information in and organize it. Aggregation only works when information is in a recognizable format.

(“Unbolding” as a constant activity; great term.)

The next speaker is Don Turnbull from UT Austin’s School of Information. He opens with a great line: “I’m from the university, and I’m here to help.” Launches into an interesting discussion of tagging and folksonomy issues.

Turnbull poses some key questions related to folksonomies:

  • How do you get people to cooperate?
  • How good can the tags be? Can you find things you wouldn’t have found? but more interesting, can you browse through categories you never would have thought of (like the “me” tag, or “whatsinyourbag”)
  • Is there a point where we stop tagging? where we feel we don’t need to tell the system anything else about us? (for example, he himself has tagged thousands of movies on netflix “mostly because I go to a lot of faculty meetings and we have wireless access…”; is there any point in tagging more?)
  • What about changing interests? You buy a gift for someone on amazon, and your recommendations are skewed towards it for a while. How can you tell recommender systems “I’m not interested in that any more?” [my note: handles this pretty well]
  • There are still lots of people not using these systems; this is a small slice of the information world

He raises some issues related to tagging, as well, such as the potential for spamming and gaming, the inherently explicit nature of tags (not always a good thing), and the value of tags being easy-to-parse and analyze plain text.

Then he moves on to social and community issues related to tagging and sharing of data:

  • Who controls the sharing? And who controls those controls??
  • anonymity vs community (and privacy issues related to this)
  • free riders—people who never tag, just browse
  • what constitutes a community? are personal relationships necessary? do they grow out of the information sharing, or define with whom you share information?

(Ack! I want his slides! I’m missing a lot!)

Talks about all the implicit metadata that could be added to explicit tags, such as “i bought this,” “i own this,” dwell time, clicks, chatter, etc.

He ends with the concept of “don’t fence me in” – we need tag mobility across systems, (flickr, email box names, amazon ratings), a common api for tags, and the ability to move between desktop and server-based views of our data.

The last speaker is Tantek Çelik from Technorati. This is a much less theoretical, much more “look at our cool Technorati tags” presentation.

He says “Anybody can be their own delicious.” — But this misses the point, I think. the value of delicious isn’t just your own bookmarks or even your own tags, it’s the collaborative filtering and discovery. He says that technorati’s approach allows you to own your own data—but the user owns his or her own data on server-based sites, too; it’s easy to import/export and backup. The value to me is in cross-user data, and new ways of thinking about things.

A questioner mentions open space technology—how can we do that virtually? How can we extend the conversation in this room beyond the borders. Panel member (can’t see who) says “that’s why I maintain a blog.”

Tantek says that things like using the technorati tag for sxsw2005 in a blog entry provides “unprecedented” aggregation, but this is exactly what trackback provides. O’Reilly did this last year by allowing people to trackback to conference session pages.

A few more questions, and I’m off to eat. I’m starved! More later from the Malcolm Gladwell keynote this afternoon.

(A meta comment about sxsw: it’s hard to get called on to ask a question; that’s where IRC really helps, but it’s surprisingly underutilized here. Too bad.)