The Evolution Of Tagging
Second-hand details on some reasoning behind the logistics of my tagging system.
Web 2.0 brought rounded corners, gradients, violators/badges, 'AJAX', and the advent of tagging. I have resisted most of the Web 2.0 gunk as much as possible, but am guilty of a few things as you can see around themolotov.net, mainly: tagging.
Tagging is a great idea. How tagging seems to be implemented though, is horrible. A guy named Tim made a great post about the breakdown of tags. He considers that his site has many more tags than Amazon.com. He discusses many things, but I think one point that he drives is that you can't get quality from quantity alone. If a thousand people tag a single item with 'bananas', by the numbers the tag 'bananas' will get the largest 'tag share' (as he calls it). I'm pretty proud of the fact that I realized this long ago when implementing the tagging system here on themolotov.net.
Each tag that I use gets assigned a completely subjective number, almost like a percentage. The tags are then weighted, so that when sorting by tag or searching, the most relevant results will appear at the top, as per my opinion. Of course, my opinion rules with an iron fist in this scenario and you might just be out of luck if my setup can't get you what you want or need (that's a different point altogether though). The tags 'relevance' (as I call it) is different from a typical percentage in that physical things like heterogenous chemical mixtures can be thought of in percentages, but subjective abstracts cannot. If I read a book on Amazon and give it three tags, I'm assuming each one of my tags has the same weight and will be factored based on social comparison results (i.e. if 999 other people used the same tag I do, it weighs more). This just seems patently unfair.
It will be neat to see how tagging systems evolve, and how people evolve with them. Unfortunately, with any socially-dependent rating or categorization system, you'll have false-positives and people deliberately trying to skew the results. Weightsm 'relevance' or 'tag shares' wouldn't alleviate the problem of people being idiots. Even Tim laments, "It seems the naughty teenagers and the pharmaceutical spammers have discovered Amazon tags!"