This is a continuation of a discussion which took place in comments on this post.
It is recommended to tag every question with at least one area tag (formerly called “top-level tag”). Currently area tags are named in the following way:
- The area tags which originate from CoRR subject areas are named as
xx.description
wherexx
is the coresponding CoRR subject area codes anddescription
is the descriptive name of the area. Examples are cr.crypto-security, cg.comp-geom, ds.algorithms and ds.data-structures. Note that some area tags deviate from the categorization in CoRR. - The other area tags are labeled like usual tags. Examples are computability and quantum-computing.
There are two problems with the current area tags.
- Names depend on whether they originate from CoRR subject areas or not, which is not an important distinction. Therefore tags look simply inconsistent.
- There is no advantage of having the corresponding CoRR subject area codes in the tags. As I understand it, the only purpose of these codes is to indicate that they are area tags. However, this is incomplete because not every area tag has a two-letter code.
I think that a more consistent naming is better.
If the Stack Exchange server had a functionality of showing area tags in a different color, we would not have to distinguish them by names. Because it does not have such a functionality, it is better to distinguish area tags by names.
Therefore, I propose the following:
Proposal. Remove the xx.
part from the area tags of the form xx.description
and instead add a fixed prefix to all the area tags. As the fixed prefix, I suggest to use either a.
(for “area”) or just .
. [Edit: I removed the earlier suggestions a:
and :
because it turned out that a colon cannot be used in tags.]
a.
instead ofarea.
because I thought thatarea.
might be too long. $\endgroup$a.
is very cryptic. Can tags be made case-sensitive ? that would be an option then. $\endgroup$a.
is very cryptic while you do not complain the same thing to the current subject area codes, which are equally cryptic (I really do not buy the argument that TCS researchers are familiar with CoRR codes). The point is that we lose nothing if we replace the two-letter codes by a fixed prefix, and a fixed prefix is more consistent. $\endgroup$area.
is too long. Maybe it is not. $\endgroup$