An unresolved problem here on community.lsst.org is how to make tags useful.
A useful tagging system should have a relatively small, controlled set of tags. Tags are best when they’re grouping posts that might be hard to otherwise group or organically search for. Tags aren’t useful when they’re only a proxy for highly technical nouns. It’s far easier to just search for those types of words.
The problem with Discourse’s implementation of tags is that you can’t pre-formulate a controlled tag vocabulary. Tags can only be created in conjunction with the existence of a topic using that tag. As a stop gap, I’ve allowed anyone to create a new tag in conjunction with that post. The assumption was that I could later go back and re-organize these organic tags into a consistent, useable set.
I’m realizing that this model will actually fail. Anytime I, as a moderator, edit tags on a post, the entire topic is considered ‘edited,’ thus bumping it back up on people’s feeds. This discourages me from addressing tags because I don’t want to create artificial topic list churn.
My proposal, which I’d like feedback on, is that we
- Turn off permissions for regular users to create new tags for their topics,
- Create a wiki post containing the canonical list of tags
- When a topic appears that could use a new tag, a moderator (such as myself) will add a tag from the canonical list to the topic. If the tag has already been used previously, of course, the ‘regular’ user is free to use that tag normally.
Is this a good model to overcome the technical limitations of the tagging plugin? If there is support, I (or anyone else) can create the canonical Tag List wiki topic and begin designing the tag system.