del.icio.us changed the /url pages again today.

parser has been updated and should function normally.

Ian Davis talks about how tagging is too expensive and will not, in the end, be quite the revolution…

I’ve come to the conclusion that this simple technology is hobbled by being expensive. This worries me. I’m not convinced that tagging will persist in the long term and I have a feeling that five years from now we’ll be looking back at the fad that was tagging and shaking our heads over the vast inpenetrable databases of tagged content. Then we’ll get back to Google and to our librarians and get on with finding the things that matter to us.

The advantage classification holds is that you can lookup the classification you need and be confident that you have found all that is available. With tagging there is no way to exhaustively search all the possible tags that people might have used, in all possible languages and spellings.

But, it’s not about exhaustivity.

It’s about the ease of finding enough to satisfy the searcher’s need for information. If the need for information is very deep, then it’s up to the searcher to continue her research. If the need for information is fairly shallow (or rather, anything other than deep), the searcher can feel fairly confident that others have been interested in the topic area before and done enough prior heavy lifting on the classification and categorization of resources to satisfy her search.

‘All that is available’ seems to be a strawman in this discussion. Very very few searches could ever truthfully be exhaustively answered. In fact, it’s the professional researcher’s opinion we usually trust when it comes to whether something is related or not. With raw information, as opposed to physical artifacts, I’d even venture that an objectively exhaustive search result (regardless of how it was organized/returned) is impossible.

I think Clay has made a very strong argument and I think he’s hedged his statements where appropriate. Tagging has shifted the cost of classification and categorization. The tools for elegant retrieval aren’t here yet.

But they’re coming.

That’s what I hear him saying.

I also think, in the long term, professionals will still be professionals. They provide information on the resources available and how best to use them. Before, this consisted of the actual sorting and managing of the hierarchy. In the future, this will be much more based in managing the many voices and scrubbing the messiness that bubbles up from below.

Embrace the messiness.

Date Zooming has been added.

You can now zoom in on a section of a particularly interesting graph. This should alleviate some of the forward-moving pressure whereby older/established sites will eventually become boring and flat.

One nice effect is the ability to go ‘back’ in time to more precisely determine where and when a site has ‘leveled out’ and become ‘mature’. Over time, I would expect some statistics to begin to fall out concerning how many taggers are required before a site’s Tag Cloud matures.

Some of this, of course, is determined by the community doing the tagging. But over time, as more of a cross-section of society is using these tools, the true averages will begin to show themselves.

Made a small parsing change today.

del.icio.us /url/ pages changed from having the date_added listed for each entry to grouping the entries under a single date_added.

Edit: 6/27 - del.icio.us changed again slightly - ‘theday’ -> ‘date’

So, two is a trend, right?

Flickr: Photos tagged with cloudalicious

The second was Nicolas Hoizey.

More discussion generated by pretty pictures. This is kind of fun.

• Andy Baio’s Waxy.org Links - June 3, 2005 - Sixth from the top.
• Two cultures of fauxonomies collide… - Tom Coates thinks there might be three reasons Tag Clouds change. The site content changes, the vocabulary concerning the content changes, or a sense of group interactions and self-awareness change the way tags are used.
• Ross Mayfield’s Flickr Photostream - A Flickr Tag Cloud Graph

del.icio.us moved servers this afternoon and updated their code at the same time.

The parser has been tweaked and caching is now set at 24 hours. If you graph the same site as anyone else in the past 24 hours, there should be no call (and no waiting) on del.icio.us. Any delay is due to graphing lots and lots of little points.

The regular expressions got a boost. Entering only ‘domain.tld’ in the box above will now work as expected. Note that graphs with and without the ‘www’ will be different.

The greasemonkey script was also updated to parse correctly. It is now at version 2.

As I come up with them, I’ll add more stats about the graph to the lower right.

For now, Tags/Author and Total Authors.

Behold… http://cloudalicio.us

I’ve moved the site to this new domain - to make it more memorable, as well as more long-URL friendly (it’s shorter).

All existing incoming links should still work. If this is not the case, please let me know.

The Bookmarklet and Greasemonkey scripts have been updated as well.

So, having asked for suggestions, I’ve now got plenty for the next few days.

Pietro has a whole list (thanks, I think).

I’ve removed the WordPress dependency for drawing graphs. The only excuse for that was speed of site launch.

The data and graphs are currently being cached for 12 hours (update: 24 hours now). I’m still messing with the caching logic (update: done) and might need to change a few things if/when the date based zooming works.

The ToDo list now has comments enabled. Be nice.

The possibility of graphs based on the (number of tags / author) is interesting and I’ll try to think through getting the x-axis to reflect authors instead of dates. It will allow for more granular analysis for sites that don’t yet have a lot of tags.

Having a log scale for the y-axis at this point does more damage than good for being able to see the trends. The characteristic “cultural changes” are much less pronounced (the diagonals aren’t as strong) and the rest of the data begins to take over the aesthetic. Maybe that’s not the point. Still working through this.

I do think that having a normalized y-axis would be interesting (all weights sum to 1). A side by side comparison to the current graphs might yield additional insights.