How aggressively should we curate tags?

Question

I suggested an edit changing the tag of the only digit-recognition question to ocr (it has 11 questions). In my opinion, there's no use in such tag fragmentation. I've also found methods, it is not descriptive of the problem at all and features 7 questions.

Should we try to avoid (i.e. edit to remove) the creation of new tags when it could be incorporated into a larger tag? When finding small, non-descriptive tags, should it be immediately removed or a specific meta question is due, like in StackOverflow? Or more broadly, how aggressively should we curate tags?

Answer 1

This is a great question! Thanks for bringing it to meta.

Question(s):

I am going to summarize your question into two major points.

1. Should we try to avoid creation of tags if another tag will work?
2. How aggressively should we curate tags?

TL;DR:

1. Yes, if you are contributing content, then try to avoid creation of tags if another tag will work.
2. You should not aggressively curate tags.

Why?

I put in the TL;DR because I am afraid I will be somewhat long winded here. Please bear with me while I do some exposition to build context, and hopefully establish why I think the above is true.

Who needs to find the questions via tags?

Primarily answerers. Non-answerers are most likely to find the the questions and answers they are looking for via a web search engine. But answerers, the bread and butter of stack exchange, will often follow tags to help find questions to answer. So removing a superfluous minor tag is not nearly as important as getting an appropriate tag affixed. For the same reason, an incorrect tag, should be removed so as not to provide a false positive to those same answerers.

Things to be aware of when editing:

1. Edits without the edit privilege (suggested edits) require someone else’s attention. One of the reject reasons for suggested edits is:

   no improvement whatsoever This edit does not make the post even a little bit easier to read, easier to find, more accurate or more accessible. Changes are either completely superfluous or actively harm readability.

2. Edits drive posts to the front page.

   So while edits are good, annoying answerers with a lot of churn on the front page is not good.

These two things mean;

Try to make your edits count

Generally try to fix any problems with the post when you are editing it. This is especially relevant if you come to the post to edit the tags. Don't simply replace a tag and move on. Take a moment to cleanup anything else that the posts needs.

On sites in which I have the edit privilege, I will freely make very small edits to new posts, but as the post ages, will need to see a better reason than something small. But this also means that for posts that have been recently answered, or had other activity that drives them to the front page, I might then take the time to fix smaller things, since they have already been bumped.

Why tags as an organizing tool?

Tags are very flexible and easy for the user to use. They are easily created in a distributed fashion, and have been shown to generally converge toward consistent meaning within a community.

There are however a couple of disadvantages, from Wikipedia:

... the resulting metadata can include homonyms (the same tags used with different meanings) and synonyms (multiple tags for the same concept), which may lead to inappropriate connections between items and inefficient searches for information about a subject.

Synonyms

Jeff Atwood had a blog post a few years back describing the Whys and Whats of the tag synonym system.

The official page has this:

Users with more than 1250 reputation and a total answer score of 5 or more on the tag, can suggest tag synonyms. Users with a total answer score (total upvotes minus total downvotes) of 5 or more on the tag, can vote for tag synonyms. Suggestions will be automatically approved when they reach a score of 4, and automatically deleted when they reach a score of -2.

You can suggest tag synonyms at 1250 reputation (2500 on graduated sites) if you have an answer score of at least 5 in the tag. But we don't have a lot of users that meet this criteria, so to date this has been a moderator (Sean) activity.

When to use Meta?

Meta is about giving the community a place to ask and answer question about the site. Via voting on these things, it gives the community an opportunity to weigh in on things.

Meta questions on SO for tags are really about largely deployed tags, when they need to be made more specific or maybe have attracted lots of misuse so should be edited out and then have the tag wiki give advice on what the tags should be used or for...

Homonyms, once they get to be a big enough problem, should likely be dealt with in Meta. These sorts of discussions would likely benefit from the community getting a chance to weigh in on how best to deal with it.

One more thought about removing small tags.

I personally think trying to remove small tags (that are not obviously in error) that have been used more than a few times, is fighting against the tide. Some things to consider:

1. All tags started with a list of 1 post.
2. Someone thought the tag relevent (someone else may also).
3. Since we are somewhat catering to answerers, and they are following tags, small use tags are likely not a big enough problem to warrant editing a post.