Isn't the reason tagging works better than rigor categorization because of the sheer number and variety of terms people can comes up with?

Are people scrutinizing tags a little too much? I've seen people stripping out various (IMO good) tags to enforce their own sense of what's SO tags should be like ...

I think the SO engine should be a little more cunning in how it deals with tags re-orging, if too many people becomes a "tags dictator" then it reduces/defeats the purpose of having a tagging system in the first place.

I've read a few crowdsourcing book, including Clay Shirky's and that's what I think about when I see people scrutinizing tags.

What do you think?

I propose that what tags a question should have be calculated the same way an ownership of a CW question is calculated, i.e. in relative percentage to the overall efforts.

Here's my opinion on these cases:

Case 1: "pyton3.0" and "python3" ... If there are people who use "python3.0" to tag questions in the first place, why wouldn't there be people searching for python 3.0 using the tag "python3.0" as well? Removing all the "python3.0" and replacing it with "python3" defeats the purpose of having many eyeballs tagging questions in the first place.

Case 2: Removing a tag does not equal more search hits! For example, imagine a case where a benevolent dictator comes and retags all the career-related tags as "subjective" would that increase search hits? No, of course! Having a variety of tags to describe a question is what make it "better" for search.

  • 1
    -1 for near violation of Godwin's law. I'd give you -0.5 instead if I could. – Andrew Grimm Nov 3 '10 at 2:45

Tags aren't for searching.

They are for browsing - they are like hyperlinks, not search keywords. The idea is to group questions into categories by human editing. The categories are typically accessed by clicking on the tag buttons, not by searching.

Each question can have a maximum of 5 tags, so having too many synonyms uses up the tags quickly. If every Python question has to be tagged python, python3 and python3.0, there's only two tags left to indicate any other aspects of the question.

I do quite like the suggestion of having lists of alternative synonyms for each tag, so that Stack Overflow could use them as search keywords to assist in searching. Any question tagged python3 could automatically have the search keywords "Python", "Python 3.0" added for search engines.


Tags are serious business. Think about tags in terms of how they're intended to be used. They're here to help people find them from the Internet and to determine which questions are interesting and which are not. If tags are too specific their usefulness is lost because odds are fewer are searching for the uncommon terms. So the way I see it, it's OK to scrutinize tags.

  • What I propose is (Removing a tag == less search hits), not more... think about it! – chakrit Jul 31 '09 at 12:03
  • If you wanna help people find the question, what you do is add a tag that is relevant to search, not scrutinize existing tags!! – chakrit Jul 31 '09 at 12:04
  • 3
    @chakrit: But replacing an excessively specific tag [geant4.9.2.p01] with a more general one [geant4] improves searchability... – dmckee --- ex-moderator kitten Jul 31 '09 at 14:10
  • @dmckee: Exactly! – Welbog Jul 31 '09 at 15:14
  • @dmckee I probably needs to improve my English... Why you need to "replace"? just add the [geant4] tags along with the existing one! That'd improves searchability even more. – chakrit Jul 31 '09 at 17:34
  • Tags aren't for searching, they are for browsing – MarkJ Nov 23 '09 at 12:50

I think that because bigger tagging overhauls are done by moderators, and they are beforehand discussed (and voted on) here on meta (see especially The Great Tagging Reorg and other questions tagged retag-requests), excessive "scrutinization" usually is not a problem.

If some user (manually) edits tags in a way that most others consider harmful, we can surely revert these changes and tell that user to reconsider (e.g. by leaving comments in the questions.)


I propose that what tags a question should have be calculated the same way an ownership of a CW question is calculated, i.e. in relative percentage to the overall efforts.

On other hand, it might indeed make sense that the tags of a given question would somehow better reflect the collective thinking of the community, and not just that of the latest editor. I.e. each user could tag a question in a certain way, and some kind of aggregation of that data would be shown on the question. (Something a bit closer to how tagging works e.g. on last.fm or Amazon.) But while this is interesting to think about, I'm not at all sure if such fundamental changes to the tagging system should really be made. Especially as it (IMHO) works exceptionally well as it is.


Here, I suggested created an alias mapping for tags that are being renamed that are like python3.0 and python3. I would suggest that if this mapping were used, it could also allow for searching for either tag to yield the same results. Obviously this only affects internal/tag based searches, I have no idea how much of an influence tags play in searching from google, bing, etc.

  • +1. Stack Overflow could automatically pick up the aliased tags and display them somewhere unobtrusive in the page, and in the meta headers, so that Google etc. use them as search keywords. – MarkJ Nov 23 '09 at 12:52

Eyeballing all these tags helps define cases where there are too many tags created with too wide a net for classification.

Take the deal with python and wanting to change python3.0 to simply python3.

Clean up of tags like this help the community get right to the question at hand without having everything so fragmented.

Every now and then, yes, new tags will be created and will live on. But then you have to prune and look at the branches every now and then to make sure you're not letting the tags get too out of hand, too needlessly narrow that they defeat the purpose of having tags in the first place.

  • But then there are people who will also search for "python3.0" instead of "python3" for the very same reason that there is a "python3.0" and "python3" in the first place, right? ... Who defines what is the best tag for the community to "get right to the question"? And based on what? – chakrit Jul 31 '09 at 12:01
  • 1
    Discussion and consensus defines what's in the best interest. But do you want to fragment so much as to have each tag for each version point number? "python3.0" was around before "python3.1" came along and, as noted in the linked discussion, the need/want of having that extra tag or combining all "python3.*" began a discourse. – random Jul 31 '09 at 12:10
  • @random I propose that a new solution be used to tag scrutinizing, not just leave everything fragmented... – chakrit Jul 31 '09 at 12:15
  • Say one set tag, "rouge" while another tag, "red", but they're talking about the same thing. You want more relevant search results, not necessarily just more search results. Removing disused, incorrect and obsolete tags would increase relevant search hits. – random Jul 31 '09 at 12:34
  • @random Are you saying... If people tags it "rouge"+"red", it should be more relevant to just the people searching for "red"? People who search for "rouge" are simply using a disused, incorrect and obsolete tags? – chakrit Jul 31 '09 at 13:09
  • 1
    Yes, yes they are – random Aug 1 '09 at 2:02

I now have such a long list of 'uninteresting' tags that I have to scroll a browser window to get to the bottom. Why? Because of the proliferation of unhelpful 'splitter' taxonomy. If I could specify patterns for interesting and uninteresting tags, I wouldn't need a giant list. If moderators could create a hierarchy of tags (python contains pythonX, ie contains ie5, 6, 7), that would help too.

My goal is to visit the site in a moment of spare time and rapidly discover questions that I have some chance of answering. The site navigation assumes that adequate answers will be upvoted and/or accepted. Usually, they are not. So I need to have a precise tag-based filter to find them.

I can, further, only click on one tag at a time. If the questions on some topic are randomly distributed into a collection of dueling tags, I can't collect them in one place very easily.

If I could ask the site to show me only truly unanswered questions, or if we could improve the frequency of acceptance or upvoting, the tag clutter would be less of an issue.

Further, the existence of some tags only leads new users to do a bad job of tagging. On some days, it seems as if you can type any sentence of English and find that all of the words (including 'not') are valid tags. This does not enhance the chances that the OP will think about the tags. Further raising the bar for tag creation might help this. Or making the decision to add a tag downvotable.

  • 2
    You can use wildcards on your interesting and ignore tags. So you could set up "python*" (note the asterisk) and that will ignore "python3", "python3.0" or even "python-4000" for example. – ChrisF Nov 22 '09 at 21:59
  • Oh, yippee! Thank you. – Rosinante Nov 22 '09 at 22:27
  • Glad to be of service. – ChrisF Nov 22 '09 at 23:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .