Should we be tagging questions with the names of specific books? [duplicate]

Question

I can't help but notice that already, we have a lot of tags. I've gone through them and seen quite a few tags with a specific name of a book, with maybe 1 or 2 posts on it.

I'm thinking that okay, there are a few pieces of literature that a lot of questions have been asked about, but if we continue like this, because there are so many different books, we are eventually going to have a hell-load of tags.

So maybe we shouldn't tag questions with specific names of books, maybe to stop having too many tags to sort through, we should tag questions with

• The name of the series the book is in
• The name of the author

And not just the name of the book.

And for books not part of a series, maybe just the author? Or it may be better to have the book name in the scenario.

If we continue like this it is going to become a near impossible task to give all the tags tag-wikis and excerpts.

Should we be tagging questions with such specific tags? Or should we have slightly more broad ones?

Answer 1

I find that tags per book, story, or poem are needless upkeep with little value.

Per-Book Tags Are Too Fine-Grained To Be Useful

While some particular works are going to receive lots of attention, many many works are not. Which is excellent -- we want a site that can deal, not only with the most popular and oft-discussed works, but also with pieces that aren't enjoying any particular spotlight.

But with per-book tags, that means we're going to see a huge number of tags with one or two questions in them.

What is the use of such a tag? You can favorite it or follow it or search for it, but it's perfectly likely that there will never be another question about that piece again.

For practically any purpose, a simple text search for the title will give you precisely the same information - possibly more easily, because you're searching natural text, the title as-is, rather than needing to find the correct tag in the tag index.

If You're Following Single-Use Tags, You're Following A LOT Of Tags

The only exception I see is following tags - if you use some subscription method (RSS, email) to easily find questions that interest you. You can't do that with a search query.

But -- unless I'm misunderstanding something here, in which case, please correct me -- I don't see a reasonable, typical use-case for being able to follow just one or two rare, single-use tags. I can see wanting to follow an author, or a series, or a theme. I can see wanting to follow lots and lots of individual works, say every single-work tag I come across that I've actually read -- but I don't see any way to actually do that. What I don't see is the person who reads the latest short story at Tor.com, writes and tags a question as the-virtual-swallows-of-hog-island, and then follows that tag and hardly any others.

So: single-use tags are useless, unless you're assuming readers are following lots of them, at which point they're still unusable.

I'm not saying there doesn't exist some notability criteria, at which point a tag that can be followed starts to make sense. I'm saying that setting that criteria at "somebody asked one question about it" is way, way too low.

Per-Book Tags Cause Clashes And Ambiguity

The scheme of "tag per book title" breaks down really easily and really often.

• Some book titles work very poorly as tags. Extremely short names will make for confusing tags - consider it (Stephen King) or love (Toni Morrison). Long names can be even worse, with ambiguous or confusing abbreviations - I've just stumbled across sld-of-adm-richard-e-byrd , which makes sense, but would be very difficult to find if you don't know what abbreviation was used.
• Titles are not unique. Lots of different books can have the same title. It gets worse once you add in short stories and poem titles.
• Books can be released under more than one title. Do I tag this book treason or a-planet-called-treason ? What language do I use? British spelling or American?
• Stories can be anthologized. Books can be compiled into omnibuses. Is this question little-lost-robot, because that's the story, or i-robot for the original anthology, or robot-visions because that's where I read it? Is this a-shadow-in-summer, or shadow-and-betrayal, which is the edition that's actually in print?

All of this ambiguity is much more of a problem in tags than it is in plain text. Because tags need to make sense on their own, define their scope and boundaries, have a well-defined independent existence -- whereas a title search, perhaps with an author tag, should have little trouble finding what you're looking for.

---

TL;DR: If you're looking for questions about Ian McEwan's "Atonement," I think you can search for [ian-mcewan] atonement just as easily as with tags. Whereas demanding tags for infrequently-discussed works is a requirement for upkeep and arbitrary categorization that is going to have lots of hiccups and speedbumps, and just doesn't provide any discernible value, to the best of my understanding.

Answer 2

Do these tags do harm?

Unlike very broad tags like short-stories, individual book tags will hardly be abused. In fact, they will considerably improve the process of searching, especially for prolific writers.

Think about it: Stephen King has written 54 novels. Surely, some people are expert in an author, not particular works, but how many of those "experts" do we have here?

Sure, we can use the manual search to write a search string like [stephen-king] is:q "dark tower", but that would require a person (a possible new user) to know how to use the search system, and it is not exactly intuitive to everyone (we're not SO here). Clicking just one dark-tower tag under a question will yield the same result, and with considerably less typing.

Besides, a search string will not be able to differentiate between a question that is about Dark Tower, and one that only mentions it.

This is a question of utility. Tags have to improve searchability; "following" is, in my opinion, a secondary task. I strongly object tagging a question with only the name of the author and whatever bizarre generic tag the asker comes up with, be it symbolism - symbolism of what? Is it actually possible to have experts in "symbolism"?

Personally, my take is this:

• If a book is a part of series, where each work is closely related to previous, have only series tags. This way, we will only have harry-potter, and not 7 individual tags. Note that [fantastic-beasts] will have its own tag due to being a prequel.

• Have tags for each individual work, unless conflicting with previous bullet.

• We may also have tags for universes, e.g. noon-universe by [strugatsky-brothers], or discworld (pardon my SF&F parlance).

Tags for individual works won't be blurry or arbitrary such as classics or science-fiction. Instead, people will be able to search for and subscribe to any work, author, or universe, they choose. We won't even have to create tag wikis, because it's all obvious from the title.

Answer 3

Let the taxonomy evolve!

I know I repeat myself a lot here on meta, but I stand by the approach of letting the taxonomy evolve. Let's not make problems where there are none. So let's look first if there is actually a problem. Perceived problems from question and other answers.

I'm thinking that okay, there are a few pieces of literature that a lot of questions have been asked about, but if we continue like this, because there are so many different books, we are eventually going to have a hell-load of tags. (OQ)

So we'll have a lot of tags. That in itself is not a problem as long as those tags are sensible.

If we continue like this it is going to become a near impossible task to give all the tags tag-wikis and excerpts. (OQ)

That is even less of a problem. We don't need tag wiki excerpts for every tag and certainly not full tag wikis. It is generally discouraged to write wikis/excerpts for well established concepts on Stack Exchange. The same way the technical sites don't need to define an email tag, we don't need to write wiki entries about books or authors. That's already been done—by Wikipedia.

We only need tag wikis to discern different authors/books by the same name or things like that. The tag the-lord-of-the-rings wouldn't need a tag wiki. It doesn't hurt, but it's not necessary.

But with per-book tags, that means we're going to see a huge number of tags with one or two questions in them.

What is the use of such a tag? You can favorite it or follow it or search for it, but it's perfectly likely that there will never be another question about that piece again. (Standback)

So we wasted a few rows in the database. That's not a problem. By definition a properly defined taxonomy will always have niche topics with niche tags. Depriving these topics of their fitting tag will make the taxonomy less coherent.

Let's see how many 1-question tags we have, when we have a thousand or two thousand questions.

All of this ambiguity is much more of a problem in tags than it is in plain text. Because tags need to make sense on their own, define their scope and boundaries, have a well-defined independent existence -- whereas a title search, perhaps with an author tag, should have little trouble finding what you're looking for. (Standback on title ambiguity, especially for short titles)

Finally, we have a problem. By definition tags have to be unique (duh!). We don't have disambiguation pages like Wikipedia to differentiate. The question is if simply skipping title tags makes that situation better?

What we'd win if we skip title tags now

• Less tags. Yep, that's it.
• We circumvented (not solved!) our one problem.

What we'd lose if we skip title tags now

• Tagging works with a lot of questions. Other answers admit that title tags are useful for important or popular works. I'll throw my standard example into the fray. Andy Weir's The Martian. It's a perfect example why that is a horrible way to discern if a tag is worthwhile or not. Unless we have clairvoyants among us, we'll never find out with the first question if a book (and it's corresponding tag) will be popular or not. However useful tagging has to involve all questions that should be tagged. It cannot be reasonably established after the fact.¹

• Searchability for short, ambiguous titles. While other answers point out that short titles create clashes the solution is hardly skipping those titles. Searching for them without tags is even worse. The mentioned examples like Love or It won't result in any useful results since those words will be in a lot of questions and answers. Thus, we should try to find a reasonable system to tag Stephen King's It properly, rather than forgoing it completely. Tagging is the only way to find short titles easily.

• Four of our top 16 tags. As mentioned above, who decides which works are worthy of a tag and which are not?

TL:DR;

Let's keep the title tags for now. We have a handful of questions a day. We can handle tagging those. There is much more to lose in abandoning the current practice than what there is to gain by doing so.

---

_{1: with reasonable effort}

Answer 4

As with the issue of author tags, let's look at this in terms of attracting experts. Tags are supposed to act as a "bat-signal" to allow people who have expertise in a topic to easily find questions about it.

• If a book is part of a series, tag it with the name of the series and only optionally with the name of the individual book.

  If you're asking about the Harry Potter series, for instance, a harry-potter tag makes much more sense than, say, a hp-and-the-philosophers-stone tag. For one thing, the latter would only be applicable to questions which focus on a specific book, which I expect most questions wouldn't. More importantly, nobody is going to be an expert on just a single book in the series; all 7+ books form a unified whole, and any real HP expert is going to know all of them well.

• If a book isn't part of a series, give it its own tag as well as the author tag.

  As I mentioned in my other answer linked above, author tags are worth having because many people will be experts on all the works by a particular author. But conversely, if there's nothing to link a collection of works except that they share an author, then there's no particular reason why an expert on one of them should have read all of them. I've read 1984 but not Animal Farm, for example. These works deserve their own tags, for the sake of people who are experts on a particular book but not on all of that author's works.

Answer 5

Following the discussion on this question, on this question, and in chat, I've been forming a bit of an insight on the tagging problem -- including why I'm finding it so vexing.

How Questions Are Tagged Affects How People Use The Site

Tags are the closest thing a Stack Exchange site has to structure.

They're how you find content that interests you; where you look first on the question feed; how you survey what content already exists on a particular topic; how you meander from one question to other related ones.

So, the question of what questions should be grouped together is inextricably linked to what are typical user-stories for using the site. Some very simple user stories could be:

• "I'm interested in the work of Flannery O'Connor."
• "I'm interested in the Three Investigators series."
• "I love talking about poetry; I'll follow the poetry feed, and see what grabs my attention there."
• "I'm an expert in 18th-century religious literature from Portugal."

We are never going to be all things to all people, and it would be impossible to maintain tags that slice the site for all conceivable, or even desirable, user stories.

That being said:

Focus On Author/Book Tagging Will Optimize For Low-Hanging Fruit

Author and book-title tags make sense, because they're an obvious topic for any given question.

But as the fundamental organization method of the site, I think they're extremely problematic. Because of all the conceivable user stories that are possible, author/book tags primarily support the stories of "I am interested particularly in Author X"; "I am interested particularly in Book Y". And they provide no support for any other conceivable use-case.

What this means, to me, is that author-book tags (and particularly tags for individual books, stories, poems) optimize for low-hanging fruit.

What is easiest to find? Questions about individual books. Why would you be looking for a question about an individual book? Because you're already aware of that book, intrigued by it, want to learn more about it. If you're not already interested in a very particular piece, author-book tags will not help you find anything. Some trivial user-stories that go entirely unaddressed:

• An interesting question on a book you'd previously considered uninteresting.
• An interesting question on a book you enjoyed, but had forgotten about.
• A question interesting for its subject matter.
• A question interesting for its genre.
• A question about a work that's easily available online, that you might consider reading just for the purpose of discussion.

What I'm describing here is a site that's hugely fragmented. It would have a small number of major clusters, around the low-hanging fruit -- the stand-out books and series that enjoy immense popularity. And -- this is the kicker -- it discourages practically any other kind of use.

Discouraging Niche Questions

If we build the site around an author/title structure, it will be difficult to use it in any other way.

• It is daunting to write a question knowing you are opening the category. That implies a lot about the level of response or interest you're likely to get.
• New users can't even create new tags. That's a pretty strong discouragement from coming to the site and posting a new question on a relatively-undiscussed book or story -- a common and crucial use-case if there ever was one.
• A user who is less interested in popular fiction -- certainly not our primary user profile, but one of the types of user profile we're interested in, if we want a scope that isn't popular fiction exclusively -- is going to have a terrible time finding anything that interests him, even if it's only as basic as "show me a hundred questions that aren't likely to be popular fiction."
• Even browsing for tags of interest is difficult -- the list of tags, naturally, will see popular authors and books rising to the top. If that's not what you're looking for, your search is going to be a discouraging one.

And so on. Now, "niche topics are niche" is going to be a problem on any Stack Exchange site. But the problem here, as I see it, is that probably a good 95% of written literature is pretty darn niche. That's its nature. The balance is different here. And if we don't figure out a structure that can work with that, then that balance is going to go askew.

This is already happening. Scroll down the main question page, and ask yourself a simple question: How many questions are about books or stories or works that you, personally, have never heard of? How many concern works that you don't consider to enjoy immense popularity, or are pretty darn close to being universally known?

I'm not criticizing -- there are many excellent reasons for this to be the case, at this early stage, at the tail-end of the private beta. But if this remains the situation over time, I don't think that speaks to good health for the site. It implies that the user base is focusing on a small number of works that are extremely well known -- while niche topics,which is where a lot of Stack Exchange's power is, are poorly represented and are underserved.

This Is Not An Answer

The conclusion of this whole lengthy scroll is not "author tags are bad" or "book-title tags are bad." It is that we don't know yet how we expect, or want, the site to be used.

My objection isn't to title tags. It's to the lack of other, more meaningful kinds of tags -- which, to me, means we really don't have much of a sense yet of what we cover or how to make the site usable. It's the smoothing-over of that lack, by saying that a question about a minor author and a minor story aren't particularly expected to get any attention after its day on the front page. I object to adopting author-title as our default tagging scheme, because I think it will serve us poorly -- but I have very little idea of what a better scheme might be.

I believe we need to devote effort to building user stories. That's the information we're missing at the moment -- a good sense of how we want the site to be used. The beta is a good start for that, but if our seed population is different enough from our target audience, then we're naturally going to steer the site off-target.

All I can say right now is the fragmenting the site by author/title, and nothing else, is a structure that I find unappealing -- but whether it's the structure that the community wants, whether it's the structure that the community is aiming for, and whether any alternative structures are an option, is more than I know how to answer right now :)

Answer 6

With a few exceptions, no, we shouldn't use book title as tags. These are far too specific. Stack Overflow has tags like [c#] and [html], not tags for every single C# method and HTML tag attribute. Science Fiction & Fantasy started off with tags for individual authors, but not for individual books except for a very few extremely popular ones (they've since then drifted towards tags for individual works, for reasons that favor theoretical purity over practical use).

Tags have two main purposes:

• To allow people to subscribe to them if they're interested in a specific topic (or conversely to filter posts out if they're uninterested). A specific work is far too fine-grained for that. Almost all such filtering can be performed on author names.
• To allow searching by topic, especially where the word or phrase is also one that is commonly used to designate something else. Author names are mostly enough for that.

The only book titles that should be tags are popular works that are very well-known, often independently of their author (if they even have an identifiable author), and have been an important source of inspiration. illiad, arabian-nights are some example of good tags dedicated to specific books. Most other title tags should be scrapped.

Answer 7

The guiding principle here is whether a tag is actually useful for grouping things. I'd say it is for books which prompt a lot of questions, but for books that are asked about once and then never mentioned again, I'm not sure I can see what we gain there (except consistency, I suppose).

A good rule, I think, would be that any book which seems more likely than not to attract future questions should have its own tag . For example, 1984 would have its own tag (since there's evidence that it continues to attract questions), but V for Vendetta (as things stand) wouldn't (because it only has one question associated with it, and this is insufficient evidence of future questions related to it).

This way the number of tags could be kept relatively low, but questions could still be usefully grouped where required.

Answer 8

I would say not, simply because it will create too many tags, and may make the tagging system too cumbersome to use. Even having a tag for each author is going to create a huge number of tags. If the tagging system was better than it currently is, it might be feasible to tag individual works.