Last night I attended Berkman special evening event titled “Future of Tagging” with Joshua Schachter, founder of delicious, with David Weinberger, Berkman Fellow, who moderated. (Schachter also gave a luncheon presentation and you can read Deborah Finn's notes and David Weinberger's live blogging of the session.)
The room was packed and it wasn’t because of the free pizza. Clearly the people in the room were del.icio.us groupies, including me. I arrived early so I could get a good seat, which paid off because I was close enough to Schachter that I video taped his elevator speech and got a business card, too! He was pleased again to hear about the nptech, too.
Some interesting chit chat before the session went live over
Weinberger: “What irks you?”
Schachter: “I’m labeled as the Web 2.0 poster child and I don’t know what it means. Oh, maybe I do, a logo with a gradient or diagonal lines in the header and CSS forms. And, talking about Web 2.0 is Web 2.0 is very Web. 2.0. ”
The first ten minutes were a discussion between Weinberger and Schachter and then the remaining 70 minutes was wide ranging discussion and q/a:
Weinberger: “There is huge excitement around the social aspect of del.icio.us and yet as you say the tool is also for an individual to help remember where something is, as a business going forward how will you balance the individual versus the social/group?
Schachter didn’t feel there was necessarily tension between individual and the group. “You get more out of the system when the motivation is selfish. For example, you as an individual find a page you want to remember. You might tag it with “read_later,” so those tags work well for you, but not necessarily the social system. It doesn’t describe the content. The tags you use to describe something should be intuitive so you can recall the bookmark. You can assume, however, that someone will tag the item for how the group does it.”
Weinberger started the next question off with “You’re the poster child for Web 2.0 and folksonomy.” Schachter jumped in with, “I don’t use the word ‘folksonomy. Tagging in delicious is about 1/3 classification and 2/3 functionality. Something easy to do that let's you recall the item. The goal isn’t to classify, it’s to remember.”
Weinberger said, “The other aspect of delicious is that I can participate in a tag stream that is of interest to me. For example, taxonomy. That stream of tags that comes through everyday is very rich and a valuable resource. One of the reasons that I tag stuff is that I want to contribute to the knowledge stream.”
Schachter, “Exactly, think of tags as votes. You’re doing it for yourself, but the good of the group. Delicious is about memory first, discovery second.
Weinberger noted, “Delicious feels so social to us, we want to know who else has tagged the information, we want to know who they are.” Schachter, “I haven’t come out with a pleasing way to display the information. When delicious tells you the number of people who bookmarked, I hate the way it looks. When I took it out, people complained. I replaced it with a bar graph. It shows popularity – large number. If just one other person bookmark, it shows a link. Popularity or lack of popularity lets you know something.”
Weinberger, “One of the problems with tagging is the ambiguity of language that we use for tags. Enterprise can mean one thing in the business community or it can mean Star Trek. What are you going to do as the system gets larger?” Schachter replied, “Clustering the data. One tag by one person is one tag by a human. From there, you do math to make connections to what other people are using with the help of statistics. Show your terms in their language.”
Schachter said there are new features coming that are more social or will be group oriented. The first one will be the ability to pick out people for my network and the other will be private groups or private tagging networks. Both of the features have a chance of altering delicious.
Then it was opened up for questions -- that were as wide ranging as the answers ... Some interesting tidbits.
Schachter watches how people use (and abuse) the system and makes changes based on that.
You can tag something with for:person or delicious account.
Tagging file formats like mp3, video, future might be to tag file types (ical). “I’m not trying to tag every file format or data type, but what’s possible, doable, and useful.”
There will be no handbook or style guidelines for tag syntax on delicious (e.g. no use dot, but not colon). “I’m very hesitant to suggest how people how to do. We get 40 emails a day telling us what should go in “politics” what should go in art. Like, yeah, I’ll get right on that.”
His definition of tag spam:
A bookmark with 1,000 different tags
There 2-3 spam incidents a week. We have an abuse monitor. I let activity happened, and then I figure out the activity and compensate the system. It’s a reactive approach. The type of spam that is happening now is posting and deleting to stay on the front page.
If I quickly block people off, when aggressive spammers come to the system, I’ll make so that site can’t be bookmarked. I block it form everyone.
Why doesn’t spam happen a lot more? Right now, delicious blocks out google page reference and spammers want to get page rank. So, spammers don’t bother.
Ownership of Data
When the user “Types Some Stuff In (TSSI)” and if I could only search as tags, I wouldn’t necessarily find everything. People search differently than they tag. The other part of del.icio.us is discovery.
Bookmarking for a site
O’Reilly now has a link that lets people bookmark an article into delicious (see here for example). But they have gotten too busy to analyze the information.
Technorati Use of Delicious Tagged Items
Right now Technorati picks up the RSS for a tag. For example, we get 15 million hits a day and only 1 million were hits from Technorati. They only drive only 400 hits back). I’m not pleased. I don’t know what to do about it.
Other interesting tidbits
Other interesting tidbits
There is a dating service that uses tags to match people, it’s called Consumating
Are there a large number of people collecting large collections versus using it? There are people using it who aren’t users.
How much of data do you have? Lots and lots, Tens of millions of posts, 100,000 people. 500,000 unique tags.
Can you analyze the speed? He is working on finding early finders. If you bookmark something and than 1,000
people bookmark. If you bookmark
something six months ago and over time more people bookmark it?