DarrenBarefoot.com

You do realize there’s ‘beta’ in the section link for Technorati Blogs, right? :P At this writing, I’m in the top 10 for listing of weblogs that are about sex, which should be another data point to determine the functionality’s current quality.

You can actually “tag” words you normally use in the course of writing by linking them to any URL that ends in that tag (and not just technorati.com) as long as you have rel=”tag” in the HTML. Also, it should pick up categories also as long as the URL in question ends with the tag. Well, it makes for a good theory, but the practice is spotty.

(BTW, the tags you have at the end of the article may not make the article show up in the tag listing, because they don’t appear in the RSS feed, which is evidently what Technorati checks to make that determination. That said, I bet nobody has thought to tag stuff inappropriately so that it can appear on pages that have high search engine ranking. Okay, that last sentence was sarcastic.)

You say that it’s 2005 and computers still don’t do a good job of automatically classifying information, when we used to have perfectly good humans that did the job. We called those people librarians. Boris recently wrote about tagspaces, and if I read him correctly, tags only work well for individuals who know what they will tag something with. That is, if I need to find a URL or know I linked to something that’s related to something else, then I search for a tag that, based on what I know about myself, I would use to categorize it. Aggregating individuals’ tags to determine what the category something belongs to so far hasn’t delivered on its promise, which is a “wisdom of crowds” of categorization. I trust myself not to spam tags, and trust those I know not to spam tags (and know when someone is spamming tags to make a point, as in the case of this article), but other than for individual use, I don’t really think that tagging is the tool that is going to save us.

[Reply]

purple monkey dishwasher! HAH! Will anyone get that?

[Reply]

You’ve started the folksonomy backlash Darren! As a Librarian, I say nice work. :-)

What Richard said. Spot on.

[Reply]

I have to respectfully disagree with you about this, because I use a plugin that uses Yahoo term extraction to link a set of Technorati tags on the end of each post I make on the blog, dynamically. Check out this post on singly linked lists and then take a look at the tags:

sll, null, reference, linked list, linked lists, head, datastructure, containers, element, imagine, and memory

Those are (mostly) decent tags for this post, especially things like “sll,” “linked list.” and “datastructure.”

[Reply]

I think tags and folksonomies have their place, but I think they are more powerful and meaningful when you own the tag - not a third party like Technorati.

What I mean by that, is tagging specifically for Technorati is stupid. I realized this a while ago and implemented my own tagging system on my blog. All Technorati should do is support the rel=”tag” microcontent format, which they do. The problem is that most bloggers link directly to Technorati.

The other powerful thing about tags is that when used correctly, they can aggregate data from other sources, like pictures, or video, etc. Linking just to Technorati doesn’t really do that. You’re at the mercy of whatever Technorati wants to put on that page. Even if someone solves the problem of a computer understanding what you’ve written, you then need to have the computer understand what’s in the picture, audio file, video, etc, in order to be able to say “this picture is somehow related to this blog post.” Tagging, imperfect as it may be, provides us that capability right now.

[Reply]

The main message here is that Technorati’s approach sucks for indexing fast-changing blog content. And I agree. I see better results from PubSub, IceRocket, Feedster, Bloglines, and other services, and I don’t need to tag artificially for them.

[Reply]

The disappointing reality, one that everyone here missed, including Darren in his original post as well as the commenters that followed, is that the technology cannot fail where it cannot ever be expected to succeed.

The fault here lies not in the failings of technology, but rather in the unresolvable inconsistencies of written languages. Consider English, a language so filled with overloaded words, obscure grammatical rules, etc., with usage so warped that actual humans can’t be relied upon to interpret it properly, and here we have people like Darren suggesting that computers be instructed to understand something that even now humans do not. The computer is a machine, one that can only follow the instructions it is provided, and given the complexities of encoding such instructions into the computer, even the implementation and application of explicit instructions can be an immensely complicated process.

On the other hand, we cannot expect all of the contributors on the internet to become librarians. Formal classification systems work in narrow problem domains such as the library organization of books (yes, it is a narrow domain: don’t kid yourself, Dewey doesn’t classify information, it classifies papery things tied up with bits of string) only because the domain is narrow, the classification system well-established, and the users classifying content limited in number relative to the users of the resulting classified materials. On the internet, virtually everyone is a creator of content, the problem domains are unlimited, and new content is being added at a rate that in many cases likely exceeds the rate of consumption of that very content. So, to suggest that everyone becomes a librarian would be essentially as untenable as to suggest that computers be taught to do something that even humans cannot.

So, while creators and users of formal classification systems go off and work to improve the value that can be provided by bringing their systems to bear on an ever-widening range of problem domains (and trust me, they are working very hard on this - just as the RDFers the effect this kind of activity has had on their Standard) and while automated classification systems are continually improved by their developers, we still need a solution that has a chance of working for the problem at hand.

Enter tagging. Folksonomies. Whatever you want to call them. However, whatever you want to call them, remember this: They are not temporary. They are not a kludge. They are the only known tractable solution to this problem. Formal classification systems can’t be applied to this space, and the automated ones have a hard enough time understanding _code_ let alone the rules of human languages that would have to somehow be embodied in code.

Tools may be created to help us evolve and manage these “folksonomies”, but they are most definitely not going away, and you’d be doing yourself, your readers, and fellow netizens a dis-service by discounting the very system that has been applied so dramatically succesfully. The only known system that can.

[Reply]