# [computability] behaves like a top-level tag

This is related to the question about top-level tags.

It seem to me that the tag acts differently than other tags. While it belongs to the top-level tag (is it?), a large number of question with just don't have any top-level tags. I guess some of the poster raises their questions in the field out of , and it is just hard to think of this top-level tag while posting. And maybe some of them just think that is the top-level tag.

It seems that there are no related categories in Arxiv, so what should we do? Should we leave them there? Should we tag them with ? Or should we make a fake Arxiv tag, like ?

Should we have a list of suggested top-level tags somewhere in the FAQ (the tagging guide seems like the obvious place), and encourage people to one of them?

• I had a similar text in the "How to write a good question?" post but removed it later since it didn't get even one up vote. Feb 24, 2011 at 5:06
• See related question here: meta.cstheory.stackexchange.com/q/1012/1800. Maybe after the opinions are stable, we can follow it accordingly. Feb 25, 2011 at 2:20

Yes, we have treated the [computability] tag as a top level tag, it is like the [quantum-computing] and [physics] tags, and I personally don't see any problem with them. I also don't think they need to have the [lo.logic] tag.

It might be better to have one corresponding to the AA.BBBB syntax like other top level tags, in which case I should say I don't think [cp.computability] is a good one, [ct.computability] (for computability theory) would be much better, but I think [computability] is also good enough.

Official interim procedure for doing retagging.

• Thanks for the link! I do not know that it is already a top-level tag. Can we emphasize that just like what @Peter has said in his answer? Feb 22, 2011 at 14:16
• I strongly dislike the idea of making up arxiv-like tags for things that are not related to arxiv categories. Feb 22, 2011 at 18:56
• I agree with Jukka and Tsuyoshi (in his answer) that we shouldn't use fake arxiv tags [xx.yyyyyyy]. Maybe we could use [.yyyyyyy] (like [.computability]) to indicate that these are top-level tags, although I'd rather see some kind of color coding. This may be beyond the capacity of the software, though. Feb 22, 2011 at 20:07
• MO uses computability-theory tag and most of its questions also have the [lo.logic] tag. This works well on MO because such questions are mainly classical recursion theory questions. On the other hand I don't like following a similar approach on cstheory, since these questions are not logic questions (as long as you are not going to call half of TCS logic), some of them are more [fl.formal-lang] questions, some of them are not. So I like the way things work now and to keep [computability] top level to other proposed options. Feb 23, 2011 at 6:41

Should we make top-level tags stand out by, say, putting them in a different color (is this feasible?) or maybe labeling them [.computability], or would that totally confuse everybody?

• I like this idea. In meta it seems to have such feature. (say the tag [discussion] has a darker color.) Feb 22, 2011 at 14:13
• @Hsien-Chih: The [discussion] tag in meta is shown in a darker color because it is one of the “required tags.” The meta is configured so that each question has to be tagged by at least one of the following four tags: [discussion], [bug], [support] and [feature-request]. These are called required tags. Currently we do not have required tags on the main site. I guess that moderators can change the configuration of the main site so that certain tags are required, but obviously that is more of a change in behavior than a change in appearance. Feb 22, 2011 at 16:08
• Having a set of required top-level tags on main sounds like a good idea. Feb 22, 2011 at 18:55
• What I wrote about required tags might be incorrect. Now I am not sure whether moderators can decide the set of required tags on the main site, given that I have only seen required tags on meta sites and none of Stack Overflow, Server Fault or Super User has any required tags on main sites. Feb 22, 2011 at 21:57
• I like the idea of a different color for supertags, maybe we should post a feature request. ps: I don't like [.computability]. Feb 23, 2011 at 6:33

I will again advocate against using arXiv style tags. Coarse tagging is probably useful if the set of tags is designed carefully, and I guess that is a good tag for that. We do not have to follow the naming convention in arXiv in tags. Adding meaningless letters “cp.” to the tag is a terrible way to indicate that it can be used as a coarse category. Let’s stop using arXiv style tags, and let’s stop creating more arXiv style tags than we unfortunately already have.

• I agree with the idea of no fake arxiv tag. I don't necessarily see a problem with the tags that are currently in use. Notice that we've already split ds.* to make it more meaningful Feb 23, 2011 at 21:54
• @Suresh: Don’t you think that it is ugly to call algorithms “ds.algorithms”? Feb 23, 2011 at 23:07
• maybe, but that decision was made a long time ago. and ugly is a very subjective term. Feb 24, 2011 at 0:01
• arxiv-style tags can sometimes cause minor confusion, but nothing that is to worry about. However, is there a gain with using this particular style? Are people more familiar with them? As with software, if it adds no additional functionality, it can only cause problems. A counter-argument from the software world would be that changing the tags or the tag-style could result in new problems and inconcistencies. Feb 24, 2011 at 2:40
• @Suresh: “Long time” is just six months or so. And while I do not think that a subjective term is bad, if you do not like “ugly,” I can restate it as “meaningless.” Feb 24, 2011 at 4:26
• I guess I don't care strongly one way or another, except for consistency. and since we can always rename tags, even that's not a major issue. Feb 24, 2011 at 5:05
• @Tsuyoshi, I think a counter point is that researchers are familiar with arXiv tags and the process of getting used to them is smoother and faster. Note that the users need to type the tags and I think most users won't go through the tags page to find the right tag for their question (they don't even read the FAQ). It is good to keep the cstheory familiar to what TCS researches are used to. In general increasing the amount of information that a new user needs to learn to use the system is not a good thing. Feb 24, 2011 at 5:11
• @Kaveh: I doubt that most TCS researchers are familiar with arXiv subject area codes. My point is that these ridiculous two-letter mnemonics are increasing the amount of information that a new user needs to learn to use the system. Moreover, it seems to achieve nothing. Feb 24, 2011 at 13:54
• @Tsuyoshi, I don't agree, it seems to me that ACM, arXiv, and AMS subject classifications are the most often used ones. I think inventing a new one is not a good idea unless there is a very strong argument for it. Feb 25, 2011 at 2:46
• @Kaveh: I do not know what you are referring to when you say “inventing a new one is not a good idea.” I am not talking about making a new set of subject areas. We are currently using mixtures of obscure two-letter codes and randomly chosen phrases such as [ds.algorithms] and [cc.complexity-theory]. Note that cs.DS in arXiv does not stand for “algorithms” and cs.CC does not stand for “complexity theory.” Someone invented them, knowingly or not knowingly. According to your logic, that was not a good idea. Feb 25, 2011 at 3:28
• @Tsuyoshi, no, my point is that IMHO there needs to be strong arguments for diverging from usual familiar classifications. On the other hand, I guess I agree with you, to some extent, we can drop the first two letters of arXiv classification if we find a better way of making them stand out from other tags. I also think this is a problem of the tagging system, the tags are used for subject classification and key-words. IMO it would have been better if we had separate fields for them, this way of using tags seems to be a result of adopting the way they are used on SO. Feb 25, 2011 at 3:49
• @Kaveh: I have never stated about changing classifications. All I am talking about is which tags we use. If the system allowed some tags to be shown in a different color, we could get rid of the two-letter prefixes completely. Without such a functionality, I understand the role of the two-letter prefixes as a marker that it is an area tag, but that is a very poor way in my opinion. Adding a fixed prefix (such as “area.”) instead of the two letters would have been much easier to understand. Feb 27, 2011 at 0:04
• @Tsuyoshi, I see. Another possibility is to use a symbol, something like [@computability] or [.computability] or [#computability] or some other symbol, but I personally find the [.computability] ugly and the idea a little bit confusing. Something like [area:computability] or [subject:computability] looks better. (The ideal solution in my opinion would be having two separate list of tags, one for (primary and secondary) subject classification and one for keywords, which would be closer to the way papers are tagged, but that need modification to software.) Feb 27, 2011 at 9:59
• If we agree on some specific structure like [subject:computability] or [area:computability] for area tags then we can write user scripts for script-monkey that would show tags in a way that we would like, e.g. would show two separate lists, those following the structure mentioned above as subject classification near each other with some color and those not following that structure as keywords with a different color.in Feb 27, 2011 at 10:03
• I posted a more concrete suggestion, which is largely inspired by the comments by Kaveh. Feb 27, 2011 at 18:18