Browsing the site today, I came across this question, which was unique in Diablo 3 in that it is the only question with a secondary tag for the player class. Taking guidance from this meta question, it seems that these types of tags shouldn't be allowed. Additionally, while I've seen many class specific questions for Mass Effect 3 (the top two Vanguard related questions), I've never seen them tagged with something like mass-effect-3-vanguard or me3-vanguard.
What does the community think?
These tags add useful information. For example, I am very interested in Diablo 3, but not at all interested in things having to do with the Barbarian. If I mostly play as Witch Doctor and Monk, I can watch those tags specifically and add my expertise to those classes while not being bothered with questions about Wizards, who I don't know about.
This is very similar to races in StarCraft.
The difficulty here is where do we draw the line? In League of Legends, there are 90-something heroes each with unique skillsets and strategies. Clearly we shouldn't have separate tags for each hero. I think Diablo 3 is going to have enough quantity of questions and few enough class options to warrant distinct tags, but I don't think a blanket policy is going to be able to be created. We will have to analyze games on a case-by-case basis to come to a reasonable decision for that game.
I largely agree with bwarner and StrixVaria (particularly StrixVaria's point about judging this on a case-by-case basis).
Update
The below concerns have now been addressed per my suggestion at the end, see:
Change the title when the two most popular tags share a common prefix / substring
The biggest concern I have is the longstanding issue with our tagging system: every question gets the two most popular tags added to its <title> element for the purposes of SEO, provided the tags don't also appear in the title, and the 25-character tag limit.
So, let's say we've decided questions about Demon Hunters in Diablo III are frequent enough to warrant their own tag. And let's take the question at hand:
If the Demon Hunter tag is demon-hunter:
<title> element would be:
diablo 3 demon hunter How long do Caltrops last?
which would allow us to rank better on "demon hunter" and have no repeated words, decreasing the chances Google gets annoyed for having spammy-looking content.
Less chance of hitting the 25-character tag limit for longer game-specific tags
Less discoverability when tagging for people who immediately start typing "diablo 3" (although can be mitigated by a tag synonym)
Messes with people who are only looking for diablo-3 questions: past experience here and on Stack Overflow shows some people forget or intentionally omit the more broad tag and you can't set up a wildcard tag (e.g., diablo-3*) to catch all the game-specific tags (could this be mitigated by tag synonyms?).
What happens when another game releases something with a demon hunter? Even if that game doesn't attract enough questions to warrant game-specific tags, it'll lead to confusion (or possibly resentment: why does Diablo 3 get to own the "demon hunter" namespace and not Indie Game RPG 4: Eternal Obscurity of the Night?)
If it's diablo-3-demon-hunter:
Tag is properly namespaced to the game, preventing confusion and allowing different senses of the word for other games
Less having to deal with the magic of tag synonyms: tag wildcard filters work out of the box as does tag auto-completion.
The <title> element is:
diablo 3 diablo 3 demon hunter How long do Caltrops last?
This looks spammy and can have an adverse affect on our rankings on Google SERPs. Even if Google doesn't penalize it, word repetition in titles, to me, reeks of the over-SEO content farms do and I'm concerned it leads to less people clicking through.
More chance of running into the 25-character tag limit for longer game-specific tags, as several characters (10 in the Diablo 3 case) would be dead space.
This is sort being on the horns of a dilemma: the cons for both suck in equal measure to me. It's much easier when we can create tags that are unambiguously about a specific game, like protoss.
But I think the second option, diablo-3-demon-hunter, could suck a lot less if SE would willing to make any of a number of slight chances to the tag system:
If a tag has a stem that's equivalent to another tag, remove the stem from the tag in the <title> element. So, take diablo-3 and diablo-3-demon-hunter: since diablo-3 appears in diablo-3-demon-hunter, it should be removed so the <title> element simply begins with "diablo 3 demon hunter".
Alternatively, if two tags share the same stem, only show the most popular one. Would be less SEO juice, but would also be less magic in how tags are parsed.
Increase the 25-character limit on tags (chances of this happening are slim to none).
Since I am the culprit here, let me share my reasoning regarding this specific case. If you want my more general opinion, see below.
At the beginning I was wary of creating a tag per each class this way, I thought it's overly spammy.
But then I realized:
We already have precedents, with world-of-warcraft (look at the related tags to the right), starcraft-2, team-fortress-2 and more. TF2 even has a similar [tf2-class] format.
It's not an open-ended list like in LoL, where there are gazillion champions and every version adds more; there are just 5, and only an expansion pack will increase that number.
It makes a lot of sense in the context of tags; one can be an expert in Demon Hunters while knowing nothing about other classes, one can wish to filter to just questions about Demon Hunters, or even filter out questions about all the classes one doesn't care about, etc.
Every class in D3 is very distinct, with its own set of skills, strategies and quirks. A lot of questions can be easily categorized according to these tags. Furthermore, questions that involve two (or more) of these tags will be unlikely. This is even more pronounced since the Diablo games, although multiplayer-enabled, are very friendly towards the concept of single player. So it's very possible one will spend months playing Diablo and caring only about a single character type. That is very different from the case with SC2, TF2 etc., where a lot of games involve other classes/races.
So I decided this makes a lot of sense and went with it.
Regarding [classname] vs [diablo-3-classname] - going with [diablo-3-classname] was a pretty obvious choice because while Demon Hunter and Witch Doctor can be claimed to be rare in other games, Monk, Barbarian and Wizard (the other classes) certainly aren't!
However, Mark raises an excellent point regarding how these would appear in the title (the other problem he mentions, tag length, is not as bad in my opinion because once you have the game prefix, abbreviations can be used without fear of ambiguity). However, this also applies to any other [gamename-suffix]-style tag, of which we already have a few.
Luckily, both of Mark's suggestions on how to deal with that (we assume the 3rd won't happen) are excellent. I've raised the issue as a feature-request, also adding Mark's suggestions.
I'm with StrixVaria here, that we should judge it on a case-by-case basis. When judging we should consider:
Based on Jeff's answer in Have we not been clear enough that meta-tags are a Bad Thing™? and the fact that Diablo 3 will certainly be over the threshold for number of questions in a short period of time, I think this tag makes sense. It is properly specific so that it will not be used for other games, and is broad enough that it should apply to a number of questions in the diablo-3 tag.
The problem with class specific tags isn't a contention of whether or not they are useful - the problem is one of consistency: how the tags are used are just as important as whether they're used at all.
We've got a fair number of questions on how to respec your character. All three of these questions have ambiguous titles that could apply to any number of games. Yet each of these questions refers to a specific game. Users cannot judge their ability to answer the question based on the title alone, so they use the question blurb and the tags to decide whether or not to open the question.
This is part of why we don't require game names in the title - it's in the tags, and that's good enough when it comes to identifying the requirements of a particular question.
A while ago, we had a dwarf-fortress-fun tag. In Dwarf Fortress, "Fun" is related to losing when things go catastrophically wrong - but that's only because it's a bit of an in-joke within the Dwarf Fortress community. But you'll notice something: The dwarf-fortress-fun tag has been removed from Gaming. It added zero information (all dwarf-fortress-fun questions are dwarf-fortress questions), and on top of that, was used inconsistently - either you knew about the joke and added it when things were going catastrophically wrong, or you didn't because you didn't.
The [Dwarf-Fortress-fun] tag was not fulfilling its role as a tag, so there was no reason to keep it around as a tag.
Which brings us to the crux of the argument:
There's two separate issues here that make consistent use of the tag hard. First, there is a sort of "Double-Standard Ignorance" going on: most games have only one tag, so the less savvy users will stop at one tag, especially if they have a clear idea of what they're asking and don't need (or bother to use) the tag auto-complete.
How many minecraft questions have we retagged [minecraft-redstone] because users were unaware the latter existed? (Perhaps Agent86 could whip up a script to tell us!)
Then there is the separate issue of "Content Threshold". How much redstone needs to be in a question for it to be minecraft-redstone? Is it a minecraft-redstone question if they want to open a door with a pressure plate? You could use Redstone, but you could just as easily put the pressure plate next to the door, avoiding the issue entirely.
Soldiers are the only class capable of using The Black Box - should questions asking about it have the tf2-soldier? What about a question asking how much damage fire does? Sure, tf2-pyros are the primary way to set someone aflame, but tf2-soldiers, tf2-snipers, tf2-scouts, and [tf2-engineer]s (I think?) can all use fire - do you tag with the most pertinent classes? As many as you can? None of them?
This isn't an issue specific to Mass Effect 3 or Diablo 3 - we've already got issues where game-specific tags are being used mostly randomly. And if the tags are being used randomly, they're not doing a proper job of filtering things.
Team Fortress 2 has a bunch of questions that fall into this problem.
None of the TF2 game specific tags have tag wikis, either, which certainly doesn't help!
But that's Team Fortress. What about the poster-child of game-specific tags, Starcraft 2?
Well, first some numbers (all of these as per the date of this post).
Total Starcraft 2 Questions: 945
Even if you add all these together (and ignore the fact that some questions are tagged with both), you only get 204 questions, which, considering that every game of Starcraft 2 involves at least one of these races, seems rather low. Do we really have 500+ questions that are race-agnostic?
And again, we have Threshold and Ignorance to consider.
Why do questions like these not have race tags?
And, more relevantly, if a question mentions a witch doctor-exclusive skill, does it also need the diablo-3-witchdoctor tag? After all, there's no potential confusion about which class you're talking about. If the tag doesn't reduce confusion, it is nothing more than needless clutter (much like adding steam to every Skyrim question with pc). And finally, if it's just clutter, why have the tag at all?
It is a good theoretical idea, but Gaming hasn't been able to satisfactorily implement such things in the past, and there is nothing different about the situation this time that leads me to think anything differently will happen.
On the whole, I think Game-Specific tags as specific as this do more harm than good, and should not be used for individual character classes.
These tags add little or no value to the questions they're attached to. The relevant information will, by necessity, be in the body of the question, and should also be in the title. One important role that secondary tags play is in search, such as multiplayer for games that are primarily single player, but include a multiplayer component. I do not see class specific tags fulfilling this role, as I foresee them being a sparse tag set already, even if judiciously applied.
They should not be allowed.
