What is considered incorrect tagging?

Question

Let me start by saying that I know that there is gamification in receiving tag-badges. Users* want the badges to show they have skills in certain areas.

I am active the sql tags and I have seen an increase in what I would call bad tagging. Examples of bad tagging, include adding select, MySQL error message tags, query to basic SQL questions.

Now, I know these tags are related, but they seem to be adding unnecessary noise, so I was looking for an official stance on how to handle this and I did not find a clear answer. A few of the posts send mixed messages:

• Should I retag a question with a tag that is based on the answer and not the question?

The question happened to be about SQL tags and the accepted answer basically says there are five tags so use them if the tag is relevant.

The next question:

• Should adding redundant but related tags be encouraged or discouraged

Had a great answer from Jeff Atwood

No, you should not add tags that are a superset of the question.

Those two answers seem to disagree with each other. Then I found the How do I correctly tag my questions with a great answer by Joel Coehoorn that says:

Re-tagging

• Do not re-tag a question if you are not going to add value to the question information by doing it.
• Do re-tag questions to use well-known and popular tags that are appropriate for the question.

What is considered appropriate tagging? Are these edits incorrect? Should they be rolled back?

I don't want to get into editing/rollback wars with people, and I was hesitant to even ask a question about it, but how should these edits be handled?

^{*Even high-rep users}

Answer 1

In my opinion, appropriate tagging refers to adding tags that are not redundant and are relevant to the information in the question. If I see a question that doesn't follow this guideline, I would modify (or rollback if appropriate) so that the tags are relevant and are not redundant. Since this topic is sql oriented, I will be speaking from that position.

Example 1

Question with a sql tag, is implying a platform agnostic query. The query tag is not necessary here, since query is part of the SQL acronym, and implies that is the nature of the question. Since there are very few questions that are actually platform agnostic, the sql tag should really not be used because it's ambiguous.

Example 2

Question with a sql-server tag, implies the SQL Server platform. From there, the question should state the nature of the question (query or performance). If the query tag exists, there is no need to list the DDL components of the query (UPDATE, FROM, WHERE, JOIN). Of course, if the query has an ORDER BY and it isn't behaving like it should*, then it should be included.

Example 3

Question with a mysql tag, implies the MySQL platform. Let's say the OP states that it is giving an error and does not mention what that error code is, but a mysql-error-1064 tag appears. This tag does not follow, just because the user editing guesses that is the error being thrown.

In summary, I would modify Joel's use of "appropriate" and state the following:

Re-tagging

• Do not re-tag a question if you are not going to add value to the question information by doing it.
• Do re-tag questions to use well-known and popular tags that are appropriate relevant to the question.

^{*Inside joke}

Answer 2

It would be nice if we could just apply the simple rule "use any tags that exist and match the question." Alas, this doesn't really work on SO (it does work reasonably well on some other SE sites) because of the huge number of crap tags.

The questions I ask myself when deciding whether a tag really is useful are:

• "Could someone want to follow / ignore questions with this tag?"
• "Could someone want to search for this tag?"
• "Could someone be an expert on this tag?"

and, specifically when deciding how to tag a post:

• "Would someone following this tag be more likely to know the answer than a random user?"

and also:

• "Is this tag redundant to others on this question?"

---

So, for example, let's take FreshPrinceOfSO's example question about UPDATE queries in MySQL. What tags should it have?

• mysql — sure, that's kind of obvious. It's a popular tag favorited by many users, with a well defined scope.
• sql — technically, that applies too, but it's kind of redundant to mysql. I wouldn't use it here, but I wouldn't remove it either.
• sql-update — are there really experts specializing on UPDATE queries? I kind of doubt that, and the fact that the tag only has 12 followers (which, on SO, is statistically indistinguishable from zero) pretty much confirms it. Useless.
• mysql-error-1064 — um... nope. Just nope.

Note that the redundancy criterion is not absolute — there's plenty of personal judgement involved. I wouldn't normally tag a MySQL question with sql, since anyone watching the sql tag will almost surely know that we also have a mysql tag. However, if the question wasn't really restricted to a specific SQL dialect, or if it was about some obscure dialect that most people had never heard of, I'd definitely include the generic sql tag.

Answer 3

The only question you need to ask yourself is: Is the tag relevant to the question?

If a question has tags that aren't directly relevant to the question, go ahead and remove them. (Example: Questions tagged with objective-C are often also tagged with xcode even though they have nothing to do with the development environment.) If a question has room for more tags and you think adding a relevant tag will be useful, go ahead and add it.

There's no need to concern yourself with finding the single most relevant tag. If there are multiple tags that work, there's no harm in attaching both even if one tag implies the other(s).

I know these tags are related but they seem to be adding unnecessary noise

I disagree with your premise that adding sql and query to the same post is a problem, and the tag summary for query seems to support my position:

This tag can be used to refer to SQL queries against a database (SQL-Server, Oracle, MySQL, etc.) or for filtering and manipulation of memory data structures (jQuery, Linq).

I'm not sure what kind of "noise" you're thinking of. Tags exist mainly to help people find relevant questions and answers. If the question at hand would be useful to someone searching for questions about queries, then by all means use the tag. If it would also be useful to people searching for questions about SQL, add that tag too. There's no "noise" to speak of unless one of the tags does not, in fact, relate to the question.