Backup Brain recently highlighted one of the problems with the tagosphere - the 'which tag do I use' problem.

I say 'one' of the problems, because there are at least three tagospheric problems that are creating a Tag-Hell:

1. When in content tagging mode which tag should I use for this post or photo? Example by Dori Smith:

Even though a proposed tag was advertised to the MSN Search Champs event attendees, chaos reigned anyhow - that's human nature and not an atypical scenario.

2. When in content search mode which tag should I use to search and find the content I mean to find? Example by Dori Smith:

"Nine different searches, nine different sets of results; but all of them are, at their heart, looking for the exact same thing. That's not working, by any meaning I know for the word."

3. Another tag search problem. If I'm looking for 'foo', I may be looking for 'foo'bar rather than kung'foo' (please - no correction comment: this was poor attempt at play on words that don't exist). Example:

The above tags are similar and they are different. But are they different enough to merit different tags, and are they similar enough to force consolidation? Yes and Affirmative I'd say. Do those who use the 'taxonomy' tag really mean something so different to those using tagsonomy, or folksomomy?  Maybe.

Tag-Hell Solutions

So as a possible solution, do we all delegate some uber-semantic-taxonomiz(s)ation-folkosonimis(z)ation-committee-slash-group to solve it on our behalf?  What would they call themselves, and would we agree to call the group by their chosen label?  And if this elite assemblage of chosen ones could agree a common ontology, how would it be 'enforced'? Would it be used? How could their system let language evolve, let memetic forces do their thing and yet keep our tagging habits in order?

Given that the people who are actually interested in this stuff can't even agree, do we really expect them to solve the problem for us?

Or, do we continue the present course and make a few degrees course correction? If so, what could we do to solve these three tag issues?

The current tagosphere's course could be described as the 'order emerging from chaos' course - let the market decide, the market being us. It's the bottom-up approach.  But as Dori has shown, we need help, clearly.

What's missing from the lists of tags above is the tagosphere's feedback and signals back to the user that can help the user decide which tags to use when tagging content and searching for tagged content.

Tag Data

There are three type of tag data than can act as signals that I'm calling: Aggregate Tag Counts, Related (or Relational) Tags and Social Tags. (I'm sure there are proper folksonomic terms for these but I don't know what they are).

Aggregate Tag Counts

The numbers (that I've made up for illustration) represent the number of posts / items that have been tagged with each tag.  If you saw this metadata and felt that the tags 'tag' and 'tags' meant the same thing (to you), I'd say you'd go with the flow. It's human nature - we like to fit in. So the feedback has a reinforcing effect.

In search mode, this feedback could also help locate where the majority of the content being searched or browsed for exists. Technorati used to provide this metadata as part of tag search results, but this feature has, alas, recently disappeared.  Since I use Technorati to help me decide which tags to use I find this loss unhelpful - I feel I'm groping for the right tag to use in terms of critical mass use by others.

This lack of feedback can only make tagnoise worse.

Related (or Relational) Tags

Another feedback signal, more widely used by tag-aware systems, are related tags. I've explored this topic previously so I won't labour the point other than to say that -

Related tags should be:

  • bi-directional
  • be represented by their relative strength to other tags
  • and be surfaced to the user in tagging and search modes.

In Del.icio.us the same tags 'software' and 'programming' are mostly bi-directional and have relationships to other tags but their relative strengths are not represented.

Social Tags

It might help to leverage the social dimension of tags. An idea proposed by Scott Koon (aka Lazycoder) is that you could use your social circle or reading list / OPML file to navigate tagged content, so that in your aggregator / feedreader / tagware you can either locate content tagged with tags that you've pre-defined, or browse a tagloud scoped to your OPML field / friends list.

It's nice idea and plays squarely in the Attention space.

Let's take it further - could we use the tags used by your reading list / OPML cloud to help decide which tags to use when tagging or searching content. The premise is that if I find you interesting, then I'm likely to find the tags you find interesting, so if I have a choice between 'folksonomy 'and 'tagsonomy' I might choose the former because you use it. In terms of solving Tag-Hell, I'm not sure if this would be as helpful as the other two types of tag data described earlier, Aggregate Tag Counts and Related (or Relational) Tag, but there maybe something in it. Who knows? (....and if your read this far, you might well be thinking, who cares?).

Tagspace Browsers

Quick note on these: Tagspace browsers are nice and fun tag discovery tools but really don't help me decide which tags to use, nor find content.

At last, the last sentence of this post

What I'd really like to see are the Aggregate Tag Counts and Related Tags data surfaced by all tag-aware systems.

Actually, the following is the last sentence of this post, sorry

If these aids became standard UI features tag-aware systems (they do exist in some) we could emerge ourselves out of Tag-Hell.

-

Tags: , , , , , , ,