Sorting out the tagging mess

Question

A lot of users, myself included, believe the current tagging situation is a bit messy. This has been discussed before:

• A personal appeal to Jeff Atwood: Recent Changes to Gaming
• Let's talk about [Mods]
• Easter eggs are evil
• Spoiler alert: [spoiler] dies
• Proper usage of spoiler tag?
• Why does Gaming.SE use tags differently from other SE sites?
• Should we allow tags for a specific class in RPG games?
• Game-specific tags
• Tagging the Skyrim questions
• Should questions always be tagged with either a console/platform or with "multi-platform"?

So, perhaps the time has come to sort the tagging mess once and for all (or at least, until the next dispute).

My suggestion, after considering the linked discussions, is:

Our guideline as to whether a tag should be used is whether it's likely users will use that tag as a filter (either filter as interesting, or filter as ignored). If a tag's utility as a filter is strongly contested, err on the side of allowing it.

So, according to that logic, and with a bunch of examples after each entry:

1. Game tags are IN.
   • skyrim
   • minecraft
   • starcraft-2
2. Platform tags are IN.
   • pc
   • steam
   • ios
3. Tags that mean different things in different games are OUT.
   • stealth
   • spells
   • paladin
4. Tags which are specific to one game are IN, and should be prefixed with the game they are specific to, unless they are unique.
   • minecraft-redstone
   • zerg
   • diablo-3-demon-hunter
5. Tags that describe the question but not its content (meta-tags) are OUT.
   • spoiler ← already burninated
   • troubleshooting
   • game-mechanics
6. (and here comes the hard part) - tags which mean roughly the same thing in different games, but are otherwise very unlikely to be used as filters by themselves, are OUT.
   • achievements
   • quests
   • save-games
   • character-build
   • controls
   • dlc
   • npcs
   • boss-fights
   • cheats
   • crafting
   • leveling
   • weapons
   • money

Now, for me, items (1-5) are no-brainers. It’s (6) which is the problematic one, and removing these tags will affect hundreds, if not thousands, of questions on this site. So, what are the pros and cons of allowing these tags?

Pros

The primary pro I can think about is that they can help when searching. Tags are originally intended for filtering, not for searching, and are not really required because one can search for any word in the question or answer title or body - as opposed to filtering, which is only possible for tags.

However, these tags can help by introducing a common set of keywords people will expect to use. For instance, if I want to search for an achievement-related question, I know I can use achievement in my query (as well as any other tag or keyword I want) and still find the question, even if the content only actually contains “trophy” or “medal” or whatever that name/platform uses for the concept of achievements.

Likewise, if I search for questions related to the Templar follower in Diablo III, I can just search for diablo-3 followers and find the question, whether the actual question content contains “Templar” (the follower’s class), “Kormac” (the follower’s name) or even just “follower” or “companion”.

Cons

1. The tagging is a bit vague and inconsistent. If we do leave those tags around, there are a lot of questions which should really be utilizing them, but don’t. This kind of weakens the search pro above. If we do go these routes, it will probably mean that a lot of new questions will need to be edited for the “appropriate” tags... and I don’t believe that’s a healthy situation.
2. It’s hard to draw the line between items under category (3) and those under (6). paladin is obviously (3). achievements is pretty strongly (6). But what about money or crafting? To which do they belong? I’m sure some of the readers already disagreed with me listing some of the tags above under (3) or (6) - this shows there is some disagreement here, as opposed to the other categories.
3. These tags are useless when they are by themselves - in other words, they are (almost) useless for filtering. It’s unlikely that many will star (mark as favorite) crafting or go to its “unanswered” tab, simply because without the game name, the scope is too wide.

Conclusion

I'm actually not 100% confident in my opinion above and will love to hear opposing arguments - of which I'm sure there are many, given that my suggestion is in contradiction with some of the linked discussions at the start.

If this guideline is accepted, I propose for all tags on this site to be evaluated based on it and be mercilessly burninated as necessary. I propose the burnination to take place as tag deletions, not retags, to avoid polluting the front page.

Overall this will lead to the site having a lower average number of tags per question; but that's not necessarily a bad thing, in fact some believe it to be the preferred situation.

Answer 1

Given the positive reception to my post on Platform Tags, I'd like to propose that when/if we do any sort of tagging overhaul, that Fixing Platform Tags be part of the change.

If nothing comes of this Meta discussion in a few weeks, I'll dredge up the platform tag issue and see about acting on it on it's own.

Answer 2

Regarding 6, I like to consider useful tags as those that someone can become an expert in. Nobody is going to become an expert in quests, because that means playing World of Warcraft, Skyrim, Flyff, et. al.

Yes, our tagging system is a bit "boring" then, but that's the nature of games; people don't "cross over" between games to play only particular bits (usually you can't because of their linear nature). When someone has a game, they are generally invested in it as a whole (category 1 game-name tags) or in part (category 4) for PvP/PvE, or this particular class/race.

For example, image-processing on SO is a great tag that should encompass a body of knowledge that is in and of itself rich, consistent, and nearly orthogonal to the programming-language-of-choice tags which generally underpin all questions on SO. There are very few examples like that here (none come to mind that aren't pseudo-platform (Steam) or technical (graphics-card)).

Answer 3

I actually do see a use for the final category of tags

Often I won't know the exact wording that is used when I'm looking to see if a question already exists, so I'll search by both the game tag, and the sub-category tag and then just a single keyword.

For example, I did a search for the Blood on Ice quest in Skyrim before asking for help with it, but didn't find the potential duplicate because the related question was titled How do I find the murdered woman in Windhelm? and made no mention of the actual quest name. If I had known better, I could just just filtered for something like "[skyrim] [quests] windhelm" or "[skyrim] [quests] house"

Of course, not many people think to search that way (I didn't at first), however with the number of questions being asked for some games (ie. diablo-3), I'm starting to find adding a second tag quite useful.

As a side note, if you do decide to keep the tags I wouldn't bother re-tagging stuff. If the tag exists, it helps the question get found in related searches, but if it doesn't, oh well no big deal.

Answer 4

It's probably a big ask from an engineering point of view, but I actually think there should be primary tags (i.e. game tags and possibly platform tags) and secondary tags (e.g. paladin, or achievements) which require a primary tag for context (e.g. you can't filter on a secondary unless you're already filtering on a primary). This approach would also remove odd cases like posts tagged with skyrim-quests but not skyrim.

Answer 5

Meta Posted

Considering the positive response this has received, I'll begin work on making a meta thread to perform a tag clean-up of the site similar to the one over at Meta Stack Overflow.