When are "sub-tags" appropriate?

Question

Stackoverflow uses a lot of "sub-tags" as I call them, for example c#-4.0 that help add value to the question by letting you search for specific information that wouldn't really work as a combination of tags (does anyone really want a 4.0 tag?!?!). These are basically tags that are a more specific version of a general tag.

Things are similar on this site. For example, dominion-openings probably shouldn't exist as there is already a dominion and openings tag, and this would merely be the intersection of the two, as would be c#-polymorphism on Stackoverflow. (sorry for the Stackoverflow example for those who don't frequent there as well, but it helps to illustrate the situation).

But what about tags that apply to specific aspects of specific games? I just retagged several questions to mtg-manabase as it would likely be difficult to find these questions without a tag and a manabase tag would apply only to magic anyway. We have already chosen to use "sub-tags" for expansions, like dominion-seaside. What sub-tags should we encourage/discourage? Would chess-sacrifices, stratego-placement, go-defence, or bridge-bidding be allowed, or should these be organized in some other way (I tried to find some edge-cases)?

Answer 1

I don't really see the point of a tag like bridge-bidding, and can see some disadvantages. As you point out, you can just search for the intersection, by typing either informally

bridge bidding

or formally (to force the search by tags)

[bridge] [bidding]

into the search box.

There are a few other issues:

• There's a distinction between top-level tags, and the rest. A tag like magic-the-gathering is "top-level". It's the kind of tag people will want to follow. In fact, if you're a hardcore Magic player, you can follow that tag, use that as your homepage, and just never see the rest of the site at all! A tag like bidding isn't going to be used like that. In your SO example, I think that's a key difference between c#-4.0 and the non-existent c#-polymorphism - the former is top-level, the latter is not.
• Tags should be as simple as possible. Tag conventions are great, but most users don't know or care about what they "should" be using. Most people would just tag bridge bidding and move on. Any policy that doesn't do what people expect needs to be considered carefully.
• Special conventions need maintenance. The corollary to the previous point is that higher rep users will have to do more monitoring and retagging when non-obvious conventions are in use.
• Determining game-specific tags is non-trivial. How do you make the distinction between a tag that only applies to a single game, and one which is more general? It's impossible to guarantee that no other game has the concept of "manabase". Even worse, they might have a similar mechanic, but call it something a bit different. Choosing where to draw that line with specialised tag naming is going to be endless.

One place where this might be a little different is the case of game expansions, and how they should be tagged. I'll bring that up in a separate question.

The best summary we have of how tagging is intended to work is Why do we tag questions?. It doesn't address your question directly, but is still somewhat helpful.

I've done a bit of further reading. Here are two takes from gaming, both for and against "sub-tags".

• In favour (full disclosure - I was involved in this one): Game-specific tags
• Against (Jeff thinks a single tag per game is almost always enough): A personal appeal to Jeff Atwood: Recent changes to Gaming