What are reasons to remove (or "burninate") a tag?

Question

I have seen multiple discussions here about if this or that tag should be removed as it has outlived its utility.

I understand that the objective of tags is to help attract expertise to a certain question area. It is not clear to me what the downside is for having any given tag. I think there is no phyiscal limitation in an electronic medium as there is in a print or physical medium, and with ranking and filtering, the cost of having a long tail of special interest items is effectively zero.

So, the first part of my question is, what are the benefits of culling tags at all? The second part is, if there are real benefits, what are the criteria used to decide if a tag should be removed or not?

(I apologize if there is already a post on it -- I could not find it searching for reason + burninate, or reason + remove + tag).

Answer 1

It's not useful

We have in the past removed tags like jungle that were covered better and entirely by a combination of other tags - in the case of jungle it was terrain and some more.

It's attracting bad tagging

That's what at some point threatened the tag wall - it was attracting many questions and answers on it were not actually about the wall-spells but the physical walls and such. Instead of outright burning the tag, its contents were culled. Would it have been the other way around and been about physical structures, it might have undergone burnination or re-definition!

It's misinterpretable

We once had a tag named adventure. It was burned because people couldn't decide if it was for writing such or published ones or making stuff up on the fly or any combination thereof.

It's duplicative

We have had tags that were totally duplicative and in a Venn diagram would have 95%+ overlap. For example adventure-design got axed and merged into adventure-writing, as discussed here.

It's something one can't be an expert in

Once we had a tag 3rd-party to demark content for a game not by the main publisher. It faced the Inquisition, just like its previous heresy 3pp which was exterminated. Why? Well, you can be an expert in publishing, or in the products of one publisher (we currently have only very few publisher tags - there seem to be about 3), but you can not be an expert in "3rd party publishers". Maaaaaybe you could be one in being a publisher of 3rd party content, but that would make your question fully taggable with publishing and legal.

It encompasses too many things... and gets split off

In some cases, a tag needs to burn (in part) because it became clunky and then got split. In general, this happens only on old tags that accumulated many many questions. This is why we have homebrew-review and homebrew. While we have kept homebrew, we could have also made two different tags and totally got rid of the previous one - like was discussed for time.

Answer 2

It crowds out more valuable tags.

While Trish's answer covers a lot of factors that could cause a tag to fail to provide positive value to the site, there is one way that tags can actually provide negative value to the site (that is, actually be harmful rather than simply irrelevant).

There are two limitations on tag "slots". First, a question can only have a total of 5 tags. If a tag that provides little or no positive value (as described in Trish's answer) gets chosen over some other tag that would have helped an expert find and answer that question, then that tag harmed the question, and by extension, harmed the site.

The other limitation on tag slots is that the tag on the question with the most weight (that is, the tag that has been used in the most questions) gets bolted onto the question title for the purposes of search engine crawlers. If a lower value tag ends up in this position over a higher value tag, it can make it harder for people searching Google/Bing/whatever to find that question, and by extension to find our site. Again, this harms the question's ability to get answers and the site's ability to help people.

In particular, the search engine tag slot may be playing a role in our diminishing number of questions about less-common systems and increasing focus on D&D 5e. Consider these four questions: 1 2 3 4. In each case, when viewing the question you can see the highest weighted tag before the question title, and in each case it's something generic rather than the name of the game system, which would actually help people find the question:

An important aspect of this is that because the search engine slot is held by the tag that appears on the most questions, as a tag gets used more and more its potential to harm the site by claiming this slot goes up. It's easy to look at a tag that's only on 6 questions and say, "Eh, we'll leave it be for now and see how it gets used." A tag that's on 1000+ questions need to be providing a lot more value to justify its existence.