Do not expire single use tags on Beta sites

Question

Tags that have exactly one question are automatically pruned after six months, leaving their questions untagged, and their wikis orphaned. This is problematic on Beta sites, where it's not uncommon for valid tags on niche topics to not get more than one question in a while.

Two Beta sites are trying to deal with the issue creatively right now:

• Let's save our tags! - Biblical Hermeneutics
• Let's save our tags! - Board and Card Games

While both efforts might generate more questions for the sites, and that's good, the communities shouldn't really worry about losing valid niche tags. A third site this has come up (in chat) is The Workplace, where we'd like to use locale tags to identify questions that are depended on local regulations or culture.

Turning off the auto-pruning is already an option, as it's turned off on AskPatents and Arqade.

Related requests:

• Stop removing tags that are only used once in the last 6 months - Meta Arqade
• Do not expire single-use tags that have a tag wiki
• Add a review queue for new and for rarely used tags
• Automatic tag clean up and the long tail - Meta SF&F

Answer 1

As noted in comments, tags are sometimes used for classification (including as cross-references from other tag wikis). These tags are valuable even if they are sparsely-populated.

I recognize that removing the pruning introduces the risk of garbage tags. One of the things beta sites should be doing is working out rational tag sets; this means frequent review of the tags list and tag wikis. I do this on sites where I'm not even a moderator; I had been assuming I wasn't unusual in that regard, but I have no data.

If we are still concerned about garbage tags, then we should do one of the following:

• don't prune tags that have wikis (already separately requested - now implemented!)
• notify moderators of impending tag purges to give them time to review (retag, add a wiki, ask more questions)

Cleaning up the "untagged" bin is a chore. It may not be a large chore on all sites, but it's unnecessary custodial work at a time when sites should be building content.

Answer 2

I fully support this for beta and other sites with a small user base, though I do think the auto-clean script should continue to be run on SO.

Here's the issue:

SO has 1.5 million users. When a tag doesn't get used in six months, that means the tag is niche and a tag is probably not necessary.

Beta sites often only have a few hundred users. For instance, Biblical Hermeneutics has only 761 users as of today. If you factor in the size of the user base, that means that a tag used only once on BH is equivalent to a tag being used 1,971 times on SO in six months.

The problem is not that the tag is a niche topic, but that the user base of the site itself is small. We should encourage the use of tags in order to attract a larger audience. (If you're not aware, tags are useful in SEO because they appear in the page title, and thus appear higher in search results for that term).

You could argue that the tags are niche because the site topic itself has a smaller audience, but this logic is wrong. Are there really only 761 people out there interested in the Bible? This site, which I have no affiliation with, along with many others have a very large potential. Anything we can do to increase the user base on small SE sites should be encouraged.

Instead, I would support a review queue for infrequently used tags on smaller sites. This allows a community to prune its own tags at their discretion.

Answer 3

I think this is a very bad idea, and I do not support it.

However, I think that ...

Do not expire single-use tags that have a tag wiki

... is a very good idea, and you should convert your votes to that feature-request, because it achieves the same goal with better outcomes. If you believe in a tag that can only muster one lone question in the last six months, you should believe in it hard enough to make sure it has a tag wiki.

Incentivizing people to create tag wikis for things they care about (and honestly, the best solution of all is still to produce more questions on the topic, as we are a system of Q&A, not a system of single-question trivia, or a system of writing tag wikis) has no downside and should serve to limit the 1 question tag preservation to tags that someone cares about, at least.

Answer 4

Opposed.

I personally have no issue with tag pruning. If a tag is very niche, then its loss is not great - and it can be renewed if the topic ever gains even a little steam. I have more issue with poorly-considered tags tacked on by enthusiasts during a beta, which don't get used (or shouldn't be used...) and should be pruned.

For example, what precisely is the downside to pruning the book of esther from Biblical Hermeneutics? If somebody comes along with a new question on the book, will they have trouble asking it? Trouble finding it? Trouble finding the old question? Not really. All that's happening is that the tag system is slightly less accurate than it could be, since for topics that were initially devoted little attention, a couple of questions might be incorrectly excluded from the tag. (See also Tony Meyer's similar comments on Meta.SF.SE, much in the same vein.)

It would be nice to see what tags are coming up for pruning, and see if any of them need saving - and maybe even allow some to be "protected" by the mods or community, if they're valid niches but don't manage to prompt questions on demand. But if you just disable pruning entirely, then I'm concerned that the result will just be two layers of tag cruft - one layer of valid-but-obscure tags, and one of tags which are simply bad and unwanted.

EDIT: However, I think Monica Cello's suggestions in her answer are a much better solution - they make it easy to preserve tags worth keeping, and also don't impede pruning out the cruft.

Answer 5

There should be a setting enabled that the moderators can set that will allow single-use tags to persist. So any important tags can be appealed to the mods, who are usually deeply involved in the beta sites anyway.