Last October I suggested what the web needed were keyword tags. These would be imbedded instructions in any post or image or file of any kind that would allow the author to indicate to the world that the file relates to "Concept X" - whether that is the topic or the author or anything else. This would be transmitted by RSS to a central respository of aggregation organized taxonomically rather than chronologically and would be available to all.
Well, they have been announced today through Technorati with the handy dandy "rel=" tag. I am not sure how many browsers will be able to read the tag for what purposes but the first intended purpose is the categorization of web content, the holy grail that many have said is impossible - mainly due to a lack of imagination on their part. What will occur now is a move to a standard taxonomy, an index of ideas so that if you want to include the "rel=" tag in anything you post it will aggregate to a place on the web with everything else having that tag so you can actually go read things about the things you already know you are interested in.
In a sense this will become open source Google and a challenge to that ediface of the internet. In another it will not or rather will do more. Google is like a index at the back of the book - you have to know the word for the concept you are looking for. What can be created through the "rel=" tag is a table of contents, a hierarchy of concepts and sub-concepts and sub-sub-concepts which lets you drift around the keywords, showing their inter-relationships. It will allow more complex and more intellegent searching - kind of like the Tree of Knowledge I wrote about in October 2003. Not peer to peer but centralized.
My prediction: whoever comes up with the best taxonomy and more useful tree of knowledge interface will be the richest person in the world in 2010. Think of the advertising value.
Comments
alfons - January 14, 2005 10:30 AM
Hey, these are mine pictures!
This is something like syndication really, and I guess that Technorati is really doing a lazy job because a page's tags could be just gathered by reading the page's text.
Alan - January 14, 2005 10:32 AM
But through the tag the author gets to signal the important issue rather than the autobot. That is the ting that will allow the real categorization of content. It can also serve to be included in every post or comment or photo about or by me so I can be my own caetegory. Hmmm...is that good?
alfons - January 14, 2005 10:44 AM
In my opinion the bots should figure out what I mean.
Probably the next interesting thing to see in the near future is search engines to summarize pages, which is hard, but doable. Something like googling "summarize:http://www.genx40.com/archives/2005/january/keywordtags" giving something like "keyword tags Technorati".
Better yet, interpreting the accompanying photos: "with photos from alfacar's natural spring near the grave of Lorca". But that's probably wishful thinking.
Alan - January 14, 2005 10:46 AM
There is a true programmer speaking: are you going to launch the "Let The Bots Decide 2005" campaign against human intrusion?
alfons - January 14, 2005 10:47 AM
Oh, let's help Technorati a bit! :-)
<a href="http://technorati.com/tag/Alfacar" rel="tag">Alfacar Fuente Grande Lorca grave natural spring</a>.
alfons - January 14, 2005 10:49 AM
No thanks, we were just saved from Skynet.
Alan - January 14, 2005 10:50 AM
One thing the bots will be able to do is winnow out multiple references to the same link so that posts about posts will be handled differently by identifying the same link within the content. You might see a display that states the core file and then commentary on the file.
David Sifry - January 14, 2005 11:43 AM
That's why we put the rel="tag" attribute: So that you can make the semantic difference clear between whether you're tagging a post (use the rel="tag" attribute) or you're just referring to the tag (no need for the rel="tag"). This was a great lesson we learned from our 2005 resultions page.
We're tremendously excited about Technorati tags, and please consider this a beta release - we know there are still some bugs (mostly in making sure that we're picking up posts and tags from posts correctly), so please be kind, but it was so useful to us internally that we just had to release it to the world.
Dave
Alan - January 14, 2005 11:52 AM
Thanks for posting Dave. One of the elements I saw was necessary in such a proposal was that it would be voluntary to the author - especially important when you are thinking about tagging by the author and the potential for aggregations creating privacy issues.
alfons - January 14, 2005 12:15 PM
Ah, the privacy issues! But why then allowing google (and lots of other) bots at all? :-)
Alan - January 14, 2005 12:22 PM
There is not permission required to google bot. But creating a seachable index of taxonomy for individual humans which links all posts and web 911 directories, etc., which then has adds on it would likely infringe Canadian privacy law and likely EU law as well. Firms setting these tools up will have to take that into account when aggregating. A consent based system which requires the author to include the "ref=" tag would go some what to overcome that.
Robert - January 15, 2005 11:19 AM
Can anyone say, "Semantic Web?"
Alan - January 16, 2005 9:57 PM
I have been playing with this with posts from the beer blog with great success. I add a beer tag to every post and ping technorati when I do with these sorts of results:<p><center><img src="images/2005/techbeer.JPG" vspace="10" border="1" rel="beer"></center>
alfons - January 17, 2005 9:30 AM
Can't beat you on liquid pleasure, however, I own:
http://www.technorati.com/tag/lorca
http://www.technorati.com/tag/alfacar
http://www.technorati.com/tag/lebuinus
And last but not least
http://www.technorati.com/tag/pow
Eat your heart out! :-)