[Update: More recent Triple Tag post over here] 

Users are great and smart and do cool stuff, you can’t stop them. Therefore it’s no surprise to see them constantly pushing things to the limit, and this is why we’re seeing the start of curious tagging methods.

Let’s look at the obvious one that I deal with…

geotagged
geo:lat=53.1234
geo:long=-2.5678

The second two tags have three parts; its namespace ‘geo’ and a key/value pair ‘lat=53.1234′. Because most tagging systems don’t allow us to search with wildcards, you can’t find all items tagged with ‘geo:lat=*’ for example, we also use a declarative tag ‘geotagged’ which can be used for searching.

Anyone can follow this [namespace]:[key]=[value] convention. If I wanted to sell my bike, I may add a photo of it to Flickr and add the tags:

selltagged
sell:price=79.99
sell:currency=dollar

There’s the declarative tag and two TripleTags. If it gained traction and enough people did it, then it’s easy enough to build a website or services that handle the searching and tagging of photos of items people want to sell. Once they’ve be sold the tags are automatically removed.

Combine these with geotags and you can find items near you that are for sale.

The Berkley Research crew are doing this with celltagging, more on that in another post. Look here for an extreme example of TripleTags.

In the case of geotags I tried to follow the W3C standards for the geo namespace. I think there’s a lot to be said for following standards or microformats. However there’s nothing stopping people from tagging however they like, and if enough people do it becomes its own useful standard. There are many places around the internet where you can read about why user created tags and ‘folksonomies’ are good, I’ll not go over them again here.

Why do we need TripleTags?

Tags are great there’s no doubt about that, but key/value pairs are great too and have their own slightly different uses. The Windows Registry uses key/values, html markup uses key/values <img xsrc=’blank.gif’> src is the key, blank.gif is the value (converted to TripleTags is ‘img:src=blank.gif’, ‘imgtagged’). If you look at adding items to Google Base they let you pick a category, then add your own key/value. There’s a good reason why Google chose to do that, and it’s something you can’t do with simple tags. The above example can be tagged as location:town=london, location:street=oxford_st and of course locationtagged.

The fairly obvious problems with TripleTags

First off, we’re making tags hold more information than they were really designed to do and that breaks things, such as tag clouds. If you had 200 celltagged images, your tag cloud would have one giant ‘celltagged’ tag and polluted with 200 individual, one use only tags. The more people celltag, or geotag (or selltag, or locationtag) the more polluted these tag clouds get.

Also the display of tags for individual items becomes less user friendly. TripleTags are normally used for systems to read and use, not people. If your items has several TripleTags it devalues the rest of the tags, they become harder to read for the end user.

There are a couple of possible ways that this can be controlled, more probably.

A site that implements tagging can either just let it continue. If a set of TripleTags start to appear but not become widely adopted then it’s up to the user to elect to pollute their own tag clouds and interface. The chances are there’ll be a quick bloom in those tags which will rapidly die off as they don’t reach critical mass. Those that do reach critical mass and then start to impact on the system could then be adopted and absorbed into the system and therefore removed from the tagging system.

The problems is that if the tags go in a direction different to that of the site. In our example of selltagging, if it became adopted by thousand of users on a site, but the site didn’t want to absorb such a tag into its inner workings into the system, then you’d end up with a rampant set of TripleTags floating around. Maybe that’s just fine and at the end of the day it’s the users choice what they decide to do and how far they are willing to go to stop their own tagging system working for them.

An alternative for the system owners is to continue to allow the TripleTags, or even encourage it as it’s adds extra functionality to tags but hide any tags with the format ‘*:*=*’ errrr… ^(\w+):(\w+)=(.+)$ (thanks Simon) from tag clouds. Then move any TripleTags with their declarative tag out of the list of tags, or at least to the end or a separate location.

A more extreme solution would be to ban TripleTags from being submitted altogether, which shouldn’t be too hard to implement. Unless the users decide on a different format geo}lat}53.1234 and so on. And if you enter that cat and mouse game with your users then what the users are trying to do is probably important enough for you to use one of the two suggestions above.