As it currently stands, I agree that askers should add both tags and subtags if that doesn't blow the 5 tag limit.
However, at least one of the following must be accepted:
all re tagging edits must be rejected, either manually or computationally.
all super or sub tagging edits must be rejected
users who make super or subtag edits must be prepared to give up any rep they get by doing so, since they will be bitten by the vote fraud script.
a minor edit mode that gives no rep should be created, and all retag edits should be minor edits.
or else it would be too easy to get the 2000 edit rep by adding super tags.
For example, there are more
visual-studio-2010 questions than there are
visual-studio ones, and I got 200 in one hour by adding
visual-studio tags to
visual-studio-2010 questions. Even a bot could do it.
It seems that the vote fraud script does not yet cover that case, but I have a gut feeling it will soon do so.
How reviewers, the vote fraud script or the askers can decide if something is a valid super or sub tag or not consistently, I don't know. For example:
- is software versioning subtagging? (
- is library of a language subtagging? (
- is product of an enterprise subtagging? (
- etc., etc.
It is clear that there will always be many hard-to-decide on edge cases.
For the vote fraud script to work, there would have to be a computationally maintained human voted tag hierarchy, but this has been proposed and generally rejected and even explicitly vetoed by Jeff in his comment. Therefore, if the vote fraud script starts covering this case, be prepared to give up any rep you get by retagging, even non super / sub retags, since the vote fraud script will get things wrong sometimes.
For editors to reject consistently, consensus could be achieved via discussions on meta. However if this is done systematically, it would be one step away from a computationally maintained tag hierarchy. The following cases have already been discussed:
the version case has a related discussion here that considers only when to add version specific tags, but not if both tags should be added.
the language vs library case has a discussion on double tagging here
If there was a no rep minor edit mode, the vote fraud script would be unnecessary for this case. This has already been suggested but not implemented, possibly because it is hard to implement. But on this specific case, I think it is easy to implement since the check can be done automatically. In any case editors would still have the hard task of deciding what is valid super or sub tagging.