A post on David Weinberger’s blog caught my attention today. In the post, Weinberger addresses some of the problems with tagging to a singular namespace. In slightly human terms, when you tag an item, that item has value in the context of a namespace. For example, my del.icio.us tags are valid in the context of the del.icio.us namespace, David’s blog tags are valid in the Technorati namespace, and so on. Weinberger, who has a healthy distrust of authority, fears what might happen if a particular namespace were to go away.
In the context of this discussion, we can think of a namespace as a website. If you go to del.icio.us/fstutzman/facebook, you get all of the items I’ve tagged with Facebook. In tagging all of these items, I’ve created content that is valuable for myself and del.icio.us. If del.icio.us were to go away, all of that value would be lost. Weinberger’s notion of moving tags to an “open” namespace – one that isn’t under singular control – is a possible solution. However, it doesn’t really feel like a satisfying solution. Opening the namespace just seems to shift the problem around. So what about this for a possible solution: we throw away the namespace.
There are three key facets of a tagging system. They are the user, the tag, and the tagged entity. In a tagging system, I control the user (myself) and the tag, and the tagged entity is simply a reference locator. Now, I could create such a system with pen and paper, or a text file on the desktop – systems like del.icio.us and ma.gnolia.com are just enterprise implementations of this basic concept. Imagine, though, if I were able to extract my tags from del.icio.us in some standardized form; doesn’t it stand to reason I could simply input them to ma.gnolia and keep chugging along? Obviously, there would be technical issues, but at the core, these systems are based on this simple user-tag-entity relationship model.
The notion of user-centricity implies a level of control. User-centric identity lets me control who sees distributed identity facets of my choice. Imagine a user-centric tagging system. In the system I’d have all my tags, reference locators for the tagged entities, and probably a bunch of metadata on the types of tags they are, in what context they occur, and so forth. In this system, I’d have tag “fred”, and there would be two entities tagged fred. The first entity would be a reference to my homepage, and the second entity would be a picture of me on flickr. Both would be tagged fred, the first tag being a url tag, and the second being a photo tag. Imagining that this “tag file” was transportable, I’d be able to stick it into delicious and flickr, the systems would recognize the types of tags, the permissions I’ve set, and display them accordingly.
When we tag, we give tremendous value to websites. The 661 items I’ve tagged in delicious are valuable – they make the site interesting, lively, more rich. They’re also valuable to me – and when del.icio.us goes offline, it hurts. More and more, I’m tagging stuff everywhere; in claimID, in flickr, in citeulike – the list goes on. And all of these tags are doing the same thing – representing a simple user-tag-entity relationship that is value-added by the website. If I were to extract, from each website, this user-tag-entity relationship, it would still have value to me; in fact, it would have a lot of value to me. However, there’s no way for me to do that today.
When we build our tag clouds, we’re just associating records with our identity through a tag. That I’ve got accounts on tagging websites only means I dilute my tagcloud. I’m doing the same process at in each website, but the value isn’t centralized. For an analogy, imagine that you have 5 email accounts on various webmail around the net. People write to you on these 5 different email accounts, but you really would like to have all those messages delivered to your main email account. To do this, you set up a forward, and the mail to the 5 webmails gets sent to your main account. In your mail account you now have a centralized record of your mail – no need to check 5 webmails anymore.
What if you wanted to do this with a tagging system? Sure, you could parse the xml output of the tag systems into a single xml file – but that would take a lot of work because taggers haven’t agreed on a standard (nor do they really want you taking your tags out of their systems). Furthermore, if you wanted to push this centralized tagging file into one of the services, would it work? No. (But if you wanted to do it with your centralized email accounts, that would work because all of those systems read mbox files.) There’s no reason why an analogous situation can’t work for tagging.
Imagine another scenario, where you and 5 friends were tagging items on the web for a project you’re working on. In a system like del.icio.us, you’d have to use an obscure tag to make sure that no one out of your group has co-tagged items. In a user-centric tag system, you’d be able to specify what entities have what tags for what contexts. It’s a fine granularity, but it is very important in the context of collaborative tagging. I keep thinking how tagging fits so well into the YADIS model; that we could control our tags just makes sense as a next-step for standards bodies. I should note there is a standards body exploring tagging – the i-tags project fits quite logically into this concept of user-centricity (though I make no claims to have any mastery of the standards, so I could be understating or just plain wrong).
If we are to make tags truly user-centric, how would we deal with the fact that we’d need tag repositories? We might want to follow the email model – our tags would be kept in stores (just like mailstores) that we could forward as necessary to sites. The real answer, though, seems to be tie-ins to user-centric identity. If I can maintain my tagcloud as part of my identity, a tag provider would broker certain tags-in-context to websites. Flickr could see all my tags, some of my tags, or none of my tags. Flickr’s system could maintain my tag contexts; it could know I was tagging photos for a group; del.icio.us could as well. When I left the site, I’d be able to take my tags with me – sharing them with my group at a (de)centralized location where we agree to meet up. There’s no logical reason why I can’t do such a thing.
Tags are new. Ninety-nine percent of the world’s population isn’t tagging. However, for those of us who have seen the light about the power of folksonomy and tagging, it stands to reason that this space will grow. If we’re to be tagging “things” in multiple locations, doesn’t it stand to reason that one day we’ll realize this is inefficient? I don’t want to check email in 5 different locations; why are tags any different?
Tags: Del.icio.us, folksonomy, tagging
Fred Stutzman is a doctoral student, researcher and teaching fellow at the University of North Carolina at Chapel Hill's School of Information and Library Science. He studies how people use social media.
this is pretty interesting. It reminds me of FOAF, and you’ve addressed an issue that I’ve read very little about but am increasingly interested in. Everytime a new service or capability pops up 1800 groups produce their take on it, and then in trying them we find that our work gets scattered across domains. My mail and loose paper work is sitting on a coffee table, a counter, a dinner table and the floor. Gosh I wish I had ONE PLACE for it. I suppose you could invent a standard for centralizing these tagging efforts, but it would only be used by those people who so heavily tag, that the pain of the disorganization is in some way overwhelming.
The reason my mail is on the floor is that when I dropped it I wasn’t thinking about my mail, I was thinking about something else, and the reason I have accounts at del.icio.us and ma.gnolia and g.odkn.ow.swhatelse.com is because they didn’t come along in order, but in the middle of doing something else. Now I’m realizing the disparities, and frankly I’d rather be doing things than cleaning my desk, or “organizing”
I don’t think my opinions detract or in any way denigrate your idea, it’s cool, I’m just offering something to consider along the way. I don’t think del.icio.us would be as cool as it is if not for the invention of the bookmarklet.
rick spencer
okay – so why not take this a step further. why not think about it in terms of developing user-centric tagging not as a standalone application but as one expression, one feature, of an entire user-centric online identity.
reading your post about the social networking sites brought back some of the thoughts I had when Friendster came out – I adopted early and, despite the frustration with the site, developed an extensive network/online persona there. I also do have a MySpace, Orkut and Facebook account, but those are significantly less-developed networks (for a variety of reasons; my co-hort for the most part graduated college well before Facebook). And there’s (some) overlap of network connections on those, but it’s always been frustrating to me the repetitive/redundant work of doing what is, really, the same thing several times over. And for what? Resolving e-mail, IM contacts; same. And when I do blogging, or other online writing projects – same thing.
All this work and capital and ME is stranded out there, somewhere.
So – claimID is definitely a logical step towards staking a central pivot to what’s already out there, and that’s good. I think, tho, that going forward it might be useful to think of these systems not as individual components but as modules to the unitary online experience: e-mail, IM, blogging, commenting, tagging, social networking are all of a piece.
The hard part, of course, is getting a whole lotta people to agree on a single standard that would then be interoperable across a whole range of sites/features/modules, and getting people cool with the idea that the primary owner and mover of the information is not the site owner/developer but the user, who can with little or no sacrifice move their usage elsewhere if the features are better. As you point out, right now both user and site owner/developer lose value if the user leaves; making it easier for a user to leave with the value of the site experience (e.g., their tags) intact would, I think, actually also end up benefitting the sites. Because an easily portable and interoperable user identity could then be plugged right back in – with no wasted effort on either side – to a site if it became better (i.e., added/gained features that were a proximate cause for the user leaving in the first place).