Tonight I attended this month’s SDForum Search SIG meeting titled “Search Different: Tagging, Social Bookmarking and Sharing”:
The session will focus on “alternate” search mechanisms that have been developed over the past 12 months such as bookmarking, tagging, rating, sharing and any combination thereof – generally in a social context. These concepts have grown to being at the core of many Web 2.0 companies, and offer a different way of finding or discovering “stuff” online – as a complement or alternative to traditional search engines.
Panel Discussion
Charlene Li from Forrester Research moderated a panel of company reps including:
- Joshua Schachter – Del.icio.us (now part by Yahoo!)
- Kevin Rose – Digg
- Manish Chandra – Kaboodle
- Michael Tanne – Wink
Similar to the first Search SIG meeting, we had a moderated panel discussion, followed by demos from the main participants, followed by some ad-hoc demos and announcement. This is a pretty good flow, but tonight there was so much to cover it didn’t finish until about 10:00pm.
Tonight’s was one of the better moderated panels I’ve seen. Charlene was a great moderator and came well prepared with a short list of questions and discussion starters.
Some key points or take-aways I found:
- “Social search” is important and unique because everyone searches differently; by aligning or searching along with a trusted group or peers, one can hopefully find relevant results (Charlene gave the example of searching for a laptop sleeve, fitting in to the Kaboodle model pretty well).
- Q: “How can we trust what’s out there?” A: look at the motivation behind tagging; self-productivity (tagging things for yourself to remember later) can provide value beyond the individual. Tags measure what user’s think; page rank measures what web masters think — key difference.
- There didn’t seem to be much worry about coming up with a standard vocabulary or taxonomy; instead, there is a natural clustering around common tags. “Lots of reasonable data vs small amount of accurate data.”
- Q: “What about spam?” A: the advantage is that people are involved, not just algorithms; most sites have user-driven or moderated features (similar to how Craigslist users can flag improper postings).
Predictions from Charlene for the social search/tagging space for 2006:
- There will be a whole ecosystem created in this space, monitoring, tools, ranking, etc.
- Predicts that one or more social disasters coming along the way
- There will be a much different cast of characters – right now, innovative smart companies, but larger players later
Del.icio.us Demo
The “popular tags” are calculated over a short (couple of hours) interval, but what’s interesting is the composition of the list doesn’t really change much over time. You can tell from the popular tag list that the current user base is very technical (design programming software blog reference web tools music css web design).
Joshua also mentioned that when they analyze tag popularity over time, everything declines in proportion because the spread of tags used widens (when the data is normalized to account for the overall growth). To gauge interest level or importance, they use an algorithm that captures recent changes in popularity rather than just the sheer numbers.
When asked about advice for startups trying to scale, he suggested getting a colo provider with a real UPS! When asked about how to grow, he recommended relying on word of mouth and what he called the “bored-at-work” network.
Digg
I’m not as familiar with Digg, but will be playing with it after seeing tonight’s demo. I especially liked the new Digg Spy feature which shows updated activity for both the front page and digg queue.
Later I thought of a good question about Digg (and similar sites): is there a good way to handle different news sources covering the same story (e.g., techdirt, cnet, yahoo). There probably isn’t any automated way to deal with such duplicates, but I wonder if the Digg users naturally moderate up some sources and down on other?
For more from Technorati: searchsig