I find it misleading with two tags for the same application QGIS. I think the qgis-2.0 should be dropped. We still have lots of people posting question in qgis tag when running qgis 2+.
I don't see any advantage of having the qgis-2.0 tag. There are three QGIS releases per year and that means three new tags per year:
I suggest we simplify and only keep the qgis tag.
While it could be construed as a double tag, technically it isn't. One is a software tag, one is a version tag.
qgis should be applied to all questions related to the QGIS software, regardless of version. The version tag, such as qgis-2.0 is used only for questions where a) that is the version asker is using or b) the version has specific relevance to the question (ie something works a certain way only in 1.8 and not any other version). A or B depends on your philosophy as part of a larger discussion as mentioned below.
The question(s) I linked to in my comment to underdark's answer are on a similar subject have have some links to relevant faqs/blog posts/discussion/etc. on the use and formatting of version tags throughout the SE network. Since comments can be temporal, those questions are:
The short summary is no two SEs seem to enforce a consistent standard and it's left up to the individual sites to determine how version tags are best used. Here on GIS.SE, our current 'policy' varies by the software - as evidenced by this and the linked questions. However there does seem to be general agreement that version tags are useful - just in different ways to different people, and the debate over what they (should) mean.
My personal philosophy, as I've stated elsewhere but applied to this specific question, is that the QGIS version tags are useful and should be kept - but only applied when version is relevant to a question. For example "2.4 broke my x" or "I have to use 1.8" or the issue is a specific bug present in a specific version.
There have been at least 20 tags used on the 16,539 QGIS Q&As so far, which you can see by typing
[qgis*] into the search bar, where it expands automatically to
[qgis] or [qgis-plugins] or [qgis-2.0] or [qgis-processing] or [qgis-openlayers-plugin] or [qgis-server] or [qgis-2.8] or [qgis-modeler] or [qgis-2.12] or [qgis-2.14] or [qgis-2.4] or [qgis-2.6] or [qgis-web-client] or [qgis-2.10] or [qgis2web] or [qgis-2.16] or [qgis-2.2] or [qgis-2.18] or [qgis2leaf] or [qgis-cloud].
Consequently, the current QGIS tagging situation is more complex than indicated in the question which seems to be concerned only with its flagship desktop application and two of those tags.
To me the most valuable use of tags is to assist potential answerers to find questions matching their skillset and software, so that they can answer them, and their secondary use is to facilitate self-assembling FAQs to help them locate duplicates easier.
I suspect that the best guidance that can be offered on how to tag QGIS questions is likely to come from its Top Answerers because they will have the keenest interest in maintaining throughput of the Q&As related to their tags of de facto responsibility.