22
$\begingroup$

At present, the reputation threshold for creating new tags is at 300, making this accessible for relatively new and inexperienced users. A lot of new tags created now (2022) are unnecessary, or simply misspellings or new names for existing tags. One of the last ones I nuked was stoch.

A low threshold was necessary when the site was young, to populate the tags! but those days seem past. Some sites have increased the threshold, like

  • Math SE (1000)

  • SO (1500)

while some others keep it at 300 (Tex/LaTeX, MathOverflow).

I propose we ask SE to increase the threshold to 1500.

$\endgroup$
7
  • 10
    $\begingroup$ +1. I like the idea. But 1500 might be a little severe as a threshold. Ideally, any new tag that is applied would go through a quick voting process, expedited by an appropriate queue, so that one or two other people could lay eyes on it and screen out obvious errors and dups. But I don't think the SE machinery supports that. $\endgroup$
    – whuber Mod
    Commented Mar 8, 2022 at 18:54
  • 2
    $\begingroup$ @whuber: That would probably require new implementations ... but if 1500 is too high, what about 1000? (same as mathSE)? $\endgroup$
    – kjetil b halvorsen Mod
    Commented Mar 8, 2022 at 19:03
  • 2
    $\begingroup$ I can see arguments for both higher and lower than 1500. My principal source of unease is that I wonder how much of an effect increasing the threshold would really have. Perhaps many of the problematic tags are created by people not very familiar with English or perhaps don't type well and don't re-read their posts carefully; and none of those characteristics is going to be closely related to how much of a reputation one has. $\endgroup$
    – whuber Mod
    Commented Mar 8, 2022 at 19:14
  • 8
    $\begingroup$ @whuber: I have been following on the new tags list, and my impression (but I did not take notes) is that most bad tags are by <1000 users. While unfamiliarity with English can be reason, I think it is mostly unfamiliarity with SE $\endgroup$
    – kjetil b halvorsen Mod
    Commented Mar 9, 2022 at 14:44
  • 3
    $\begingroup$ I think there's 2 layers of unfamiliarity with SE: generic issues about how to add a tag, their purpose, procedures for discussing/merging/deleting them, etiquette about mass (de)tagging... And a more specific issue: our forest of tags is the result of arbitrary historic decisions (which change over time, eg as we realise some have got overpopulated & should be split up) & it takes time to get accustomed to them. Even a seasoned statistician won't immediately know the best way to tag their questions; if they're frustrated in the meanwhile, it's easy to add a tag that's overly broad or narrow $\endgroup$
    – Silverfish
    Commented Mar 10, 2022 at 0:53
  • $\begingroup$ I've not been active with tagging for a while now, but my recollection is a sizeable proportion of the new ones are not inherently unreasonable - you could imagine another site, perhaps one of a different size, where they would work okay - but it's just "not how we do things around here". Personally I think raising the limit is fine: there's quite a few reviewers who add good tags to questions, sometimes ones that the OP wouldn't have known exist. That will usually work out better than the OP just having a guess and making up their own tag unless they've got familiar with what we have already $\endgroup$
    – Silverfish
    Commented Mar 10, 2022 at 0:58
  • 1
    $\begingroup$ It would be useful to know how the distribution is for good and bad new tags as a function of reputation. At the moment this measure seems to mainly restrict the quantity and reduces the problems of 'too many' new tags (which is a problem that might need to be tackled differently, e.g create better detection systems that currate new candidates) but it is not clear how this improves the quality. Unless we believe that we do not need too many changes anymore and this measure is effectively just a way to fix the current tags $\endgroup$
    – Sextus Empiricus
    Commented Mar 18, 2022 at 14:59

2 Answers 2

Reset to default
5
$\begingroup$

I've raised the reputation required to create tags to 1000 - in general this setting doesn't get customized on a per-site basis. It's pretty much 300 network-wide with the exception of Math and Physics, at 1000, and Stack Overflow at 1500.

It's also somewhat difficult to get data to back up the request because tags are hard-deleted when they disappear from the site, so we can't even look for examples unless you provide them for us. That said, there's not generally a huge amount of harm in changing this provided the community here is willing to retag stuff that may need new tags created.

When it comes to tag management, please also consider whether you're seeing tags created frequently that shouldn't exist - such as the "stat" tag mentioned in the comments here. If there are specific tags that are problematic and keep coming back, we can block them from being created again, without the need to block all tag creation for lower-reputation users.

For more information on the tag blocking process, please see the main MSE FAQ about it - How do we request to have a tag blocked?

$\endgroup$
2
  • 1
    $\begingroup$ Thanks for taking care of this, Catija! Just to be completely clear, what amount of reputation is required to create a new tag? $\endgroup$
    – Sycorax Mod
    Commented Jun 17, 2022 at 0:23
  • $\begingroup$ Wait, you can't read between the lines and know what I was thinking??? Hahaha. It's 1000. $\endgroup$
    – Catija Staff
    Commented Jun 17, 2022 at 0:36
17
$\begingroup$

I think us going towards a higher threshold is reasonable. In a way, CV.SE is a mature site/community anyway so the "obvious" tags are by and large covered. Bringing our threshold in line with Math.SE that already uses 1000 is most likely adequate. The even higher threshold using SO might be motivated by other parts too (e.g. much larger user base).

$\endgroup$
3
  • 7
    $\begingroup$ +1, I'd also add that we have a lot of tags that should already cover the vast majority of the topics that are relevant for the Q&A page, so I won't hesitate in having the high threshold. $\endgroup$
    – Tim Mod
    Commented Mar 11, 2022 at 9:33
  • 4
    $\begingroup$ I just nuked a newly introduced tag stat by a sub-400 user $\endgroup$
    – kjetil b halvorsen Mod
    Commented Apr 30, 2022 at 14:12
  • $\begingroup$ @kjetilbhalvorsen How often have users tried to create that particular tag? If it's happened repeatedly, it may be worth requesting that it be blacklisted as an intrinsic tag. $\endgroup$
    – V2Blast Staff
    Commented Jun 14, 2022 at 16:11

You must log in to answer this question.

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