12

Krazer's post from July 2014 about identification requests asked the following question:

Do [identification-request questions] attract so-called "help vampires" that plague much of the Stack Exchange network of sites (especially Stack Overflow)?

That is, do identification-request questions attract users to the site that consume the time of answerers, while not contributing back to the site?

It sure looks like they do!

Here are some data about participation of users whose first post (meaning "question or answer", i.e. comments ignored) on the site was...

All of these exclude users who created their accounts less than 4 weeks before the latest data dump was imported to SEDE, as well as users who created their accounts during the site's private beta.


Consider the following summary statistics:

╔══════╦════════════════════════╦═════════════════════╦═════════════════════════════╗
║      ║ Asked other questions? ║ Posted any answers? ║ Commented on non-own posts? ║
╠══════╬════════════════════════╬═════════════════════╬═════════════════════════════╣
║ cat1 ║  42 ->  8%             ║  45 ->  8%          ║ 161 -> 30%                  ║
║ cat2 ║  53 -> 39%             ║  33 -> 24%          ║  69 -> 51%                  ║
║ cat3 ║ 184 -> 19%             ║ 153 -> 16%          ║ 367 -> 37%                  ║
╚══════╩════════════════════════╩═════════════════════╩═════════════════════════════╝

It comes as no surprise that category 1 users (people whose first post was an id-request question) contribute less to the site in aggregate than category 2 users (people whose first post was a big-3 or FMA question) or category 3 users (people whose first post was a question of any kind).

Data becomes very sparse as you look for more fine-grained statistics (since we're still a small site), but a glance at the histograms for the various columns in the SEDE queries universally has the right tail being meatier for categories 2 and 3 than for category 1, indicating, again, that category 1 users are (on average) more vampiric.


I will now take a moment to reiterate all the reasons I'm aware of that we should ban identification-request questions:

Other things that peeve me about identification requests:

  • 1
    I don't understand why you are comparing us to sites like SO with very different scope and needs. If you compare us to Scifi or Movies & TV you'll see that our rate of identification questions is pretty normal (actually much lower than Movies & TV's), and those sites seem like much more reasonable goals to aspire to than just blindly trying to follow SO logic. But your examples are examples of another issue which I intend to bring up in the next day or two with many questions and why the current guidelines are flawed. – Logan M Dec 30 '14 at 23:36
  • @LoganM There are only two references to SO in there - (1) Krazer's question from the previous post; and (2) a throwaway remark about COBOL. I would think that regarding (1), help vampires are universally undesirable; and regarding (2), the point could just as well be made about any site - a lack of diverse content is boring. I didn't think to look at SF&F; I'll update this with some statistics from over there for comparison. – senshin Dec 31 '14 at 0:17
  • The entire concept of "help vampire" is SO-specific jargon that means nothing unless you view SE in that lens. – Logan M Dec 31 '14 at 0:20
  • @LoganM Sure, it's a phenomenon that is obviously most prevalent on SO, but the notion predates Stack Overflow and can be a feature of any internet community that provides "help". If you'd prefer, this could just as well read "Do identification-request questions attract users who don't contribute to the site?". – senshin Dec 31 '14 at 0:26
  • I guess we're debating semantics at this point. I'm fine with saying that identification requests will attract users who don't contribute as much as other users on average, but I don't have any problem with such users coming and going as they choose. The term "help vampire" to me implies a negative connotation, as well as the idea that the person consistently produces low-quality questions which aren't useful to anyone else, and that such users are not desirable. That isn't what I see from these users, or even how I view our role here. – Logan M Dec 31 '14 at 0:31
  • @LoganM Sure, that's a fair point of view (though I disagree with it). I've reworded the title to explicitly describe the kind of users we're talking about here rather than calling them "help vampires". – senshin Dec 31 '14 at 0:37
  • 1
    If prurient id requests get the most views, we should just ban non-prurient id requests. – Torisuda Jan 1 '15 at 0:03
  • 2
    I, coming from a site that is also flooded with this shit, have another reason in addition to the very good ones you already have: They are not improvable by the community. Whenever a new user, who might not be the best asker yet, posts a question, it can always be improved to a better state by anyone in the community, as long as the general idea of the question is clear. But with ID questions this is just not possible, except for grammar and spelling stuff. You can't ever improve the content of the question without the original asker, and let's face it, they'll usually not do it themselves. – Christian Rau Jan 20 '15 at 19:33
17

I've noticed a trend among regular Stack Overflow users which, as someone who doesn't use SO, is frankly bizarre. SO users tend to think of SO (and by extension all SE sites) as a battleground between "good questions" and "help vampire questions". They think the people asking for help are lowering the quality of the site, and will ultimately lead to the site being unusable, by good questions not getting the spotlight they deserve.

That may be the case on SO, but there is no battleground here. Questions are not starved for attention. We only get around 10 questions per day, and many of these are seen by the regular users here. Even questions with low view counts still typically get to around 100, and nearly 1/6 of all our questions have passed 1000 views. I have no trouble reading all of them except a few series I don't follow. Of all the sites in the network we could become, SO is number 1 on my list of SE sites I hope we don't become like. This is a recreational site about several media of entertainment. If we take ourselves as seriously as SO, it would simply not be fun, and there's no point to this site if it isn't any fun. The problems they face there, with not enough users to answer all the questions (even the well-written ones) are problems which I would just give up on if they were here, and I'm someone who is rather dedicated to this site. I have a hard time believing we'll ever be much like SO, or that it would be a good thing for us to try to be. So to me, the idea that people who aren't "contributing" and "only" ask one question is very strange: that question is already a contribution!

More to the point though, that's really what we're here for: to help people with questions about anime. That's in fact the only reason we're here. There are no "help vampires". There are people who want answers to questions about anime. Some of them ask questions well, and some poorly, but we ideally want all of them to get their answers here, and in the exceptionally poor cases we'll send it back to the OP saying that the question needs work and hopefully offering constructive criticism. If the user sticks around, that's great, but they are definitely not under any obligation or anything of the sort, and we shouldn't count them as bad if they don't feel like sticking around.

Your analysis shows that users whose first questions are questions post fewer posts on average than the site. I don't see why that's a problem. We're a question and answer site. There can't be answers without questions. Each question is fundamentally keeping this site just a little bit more alive and active. If a user only feels like asking one question and gets their answer here, that's still a net positive for us. Sure, I'd love it if everyone stuck around, but not everyone particularly wants to. If the only way we can help them is by identifying an anime for them, that's still something.

And ID requests are helping people, whether or not you like it. For one, when we answer them correctly (which is not all that much less often than the site average), we help the OP. In addition, I (and several other regular users) have found them useful for purposes other than the OP's original one, e.g. to discover new series I was not aware of. This is extremely rare outside , but not uncommon while browsing it. You may say that "this isn't what the question was for", but you can't deny that it's still a helpful source.

It's not unheard of for identification requests to actually get a significant number of hits, presumably by people looking for the same series. Look at Anime with a little red haired character who pilots a racing pod or Where is "See you Space Cowboy" from? for two examples. Yes, this is rarer than for other tags, if by "other tags" we mean mostly things like and . But if you look at questions on tags with only a few uses (e.g. obscure manga series from decades ago) I think you'd be hard-pressed to claim the situation a whole lot worse there.

The really big sin of is that they're grouped in a way that makes it easy to think there's something wrong and target specifically. It's a big tag without the following and compiled knowledge-base of a big-name series. If you look at the tag alone, it looks huge. It has more questions than . And it has a lower answer rate than any of , , , and . Wow, it must be terrible, right? No, in fact, the 's 87.1% answer rate is almost the same as the site-wide rate of 87.5%. The oddballs here are the mega-popular series which have a combined 98.7% answer rate.

But identification questions are generally about pretty obscure stuff, so it only makes sense that the answer rate would be similar to the more obscure stuff which is more representative of our site as a whole than those few extraordinary tags. We aren't going to get 20% of our incoming identification requests looking for Naruto, Bleach, One Piece, or FMA no matter what we do, so the less popular series are over-represented in the tag. Relative to our site-wide stats identification requests are pretty much average in terms of answer ratio, but it's hard to identify or target the other "problematic" questions because they're not yet statistically significant or worth the effort. Likewise, the views on most identification requests aren't stellar, but they're only particularly bad when you compare them to other big tags; if you look at the site in full they're not terrible, though admittedly below average (but still above what a question would get on SO). Much the same story holds for the scores in the tag. looks like a bundle of really bad questions when compared to the likes of , but it's more of a bundle of fairly average questions compared to the site as a whole, and removing it doesn't really help our average stats appreciably.

Moreover, deciding whether or not a topic is a good fit based on this kind of statistical analysis seems profoundly backwards to me. Statistics like traffic are only useful for understanding trends in the absence of any major changes to the site. If we keep the policy and enforcement the same, an improvement in answer rate is a good thing. But it's a good thing because it means we're helping more people. Changing the policy simply because this stat will improve isn't helping more people, it's trying to game the stats. We should be thinking about in what way we can help the most people, and to me, none of the arguments I've heard presented here or anywhere else are convincing that the answer isn't to "allow good questions, both on identification and on other topics".


Now, with all that said, I do agree that there's a limited (but perhaps growing) problem, not with identification requests as a whole, but specifically with unanswerable identification requests, and a seeming inability to separate these from the acceptable ones via voting and/or closing. This is a real problem and in my mind requires some action. Some questions tagged are just a dump of a couple random disjoint plot points that aren't even memorable, and the OP expects us to read their mind. Even if users here know the series which is being asked about (which we probably do in many cases), we wouldn't be able to answer it. That is a problem, and frankly, our existing criteria don't cut it when it comes to doing this. Quite often a question will get by on a technicality when I actually check these. Other times, a perfectly answerable question (e.g. one containing a complete plot summary of a unique series) nonetheless doesn't technically fit.

I've thought a lot about this over the past month or so and have a fairly concrete proposal for a modification to the guidelines which I think will alleviate a lot of this. I'm intending to post this proposal in the very near future (to spoil it a bit, it's a modification of my earlier proposal at What kind of policy should we adopt for current and future "identification-request" questions with no audiovisual artefacts? from before we had any concrete guidelines, adapted and simplified to make it easy to decide). It also fixes much of the issue of repetitive/non-descriptive titles. I've been testing this for some time now and I found that in about 90% of cases I could immediately determine in seconds if a question was good or if it needed more detail. My intent is to post this suggestion in the next day or two if there are no major delays.

  • 5
    I really enjoy reading you walls of texts. You may have changed my mind, if even for a bit, regarding these types of questions. I'm looking forward to reading your proposed revision to the id request criteria. – Madara Uchiha Dec 31 '14 at 13:51
  • So, the reason I pointed out those various statistics is not because I think they are, in and of themselves, reasons to do away with ID requests. Rather, they are proxies for factors like user opinions (votes - "do I like this content?") and external usefulness (views - "do other people have a use for this content?"). What I'm getting at is that it's those factors that suggest we may ought to change our treatment of ID requests. Like you, I don't see gaming site statistics as a useful goal. – senshin Dec 31 '14 at 18:12
  • Aside: in the particular case of the "See you space cowboy" question, the reason that has a lot of views is almost certainly that it was in HNQs for a while (i.e. the source probably isn't search engines, but rather users on other SE sites that have seen or are aware of Cowboy Bebop). – senshin Dec 31 '14 at 18:13
  • I agree that we'll never have SO-scale problems, and I agree it's inconceivable that bad questions (of any kind, ID request or not) could become so prevalent as to completely stonewall the answering of good questions (as has come close to happening on SO at times). Nonetheless, answerer time and energy are limited resources. This site is certainly not done growing, and there could come a point where bad questions do become an obstacle to the answering of good questions. This is a problem that I believe is best nipped in the bud. (But +1; I look forward to reading your proposal.) – senshin Dec 31 '14 at 18:16
10

I feel conflicted about ID requests. At the time Krazer made the referenced post, I was still new to the site and thought it was a little harsh to talk about getting rid of them altogether. But hours of chasing wild geese and trying to squeeze blood from stones for the benefit of hit-and-run questioners who won't even accept an answer afterwards has changed my mind a little.

If I have a definite idea of what series the OP is thinking of, I don't mind answering an ID request, but there are many, many times when the question is impossible to answer, because, as senshin says, it's "tantamount to 'can you read my mind?', or worse, 'can you read the mind of my friend, on behalf of whom I'm asking this question for some bizarre and utterly incomprehensible reason?'". If, like me, you're not very bright, you might go spend time doing the OP's research for them and come up with nothing, because you have no idea what you're actually looking for. When I did my own research to find the incredibly obscure series Chiisana Obake Acchi, Kocchi, Socchi that I'd seen when I was five and only remembered fragments of, I had a pretty good idea when I'd found it, because what I found triggered more memories and fit with my recollections. But I didn't have that when I was searching for a series on someone else's behalf; this essentially made it impossible for me to find the series. Questions whose answers can't be found don't contribute anything.

I'm also pretty fed up with what senshin calls "a steady stream of people posting things that are just vaguely anime-style that we end up having to trace back to some rando's original art on pixiv", or even worse, things like this question. That doll doesn't look even vaguely anime-related to me, and when I asked why the OP thought it did, no response. Answering these questions doesn't require anime knowledge; it requires Google gymnastics and lots of spare time. And the answer is not likely to be worth much to anyone later.

On the other hand, I think Logan M. makes a lot of good points about why ID requests aren't a problem. I'd add that questions looking for a series, when they're well answered and aren't left dangling (as they so often are), have a somewhat subversive side benefit—they can introduce readers to new series. I've tried a few new series because an ID request looking for it caught my eye. So in that way, I suppose these users are contributing the site, albeit very very indirectly. And certainly, as Logan M. says, "identification questions are generally about pretty obscure stuff, so it only makes sense that the answer rate would be similar to the more obscure stuff which is more representative of our site as a whole", and I'm the last person who would want to limit questions about obscure series on the site just because there's a good chance the questions won't be answered. After all, I am the one who asked questions about Hanayamata, Reset!, Bungaku Shoujo, and Zaregoto, even though I was pretty sure no one on the site would be able to answer them. And on SO, while I don't read COBOL questions, I do rely on the community of Clojure experts who post answers about that relatively obscure language. So in another sense, these users are contributing by broadening the focus of our community beyond just Naruto, Bleach, and One Piece. Still, I have to admit that these hit-and-run questioners who show up, ask one bad question, and disappear, bother me.

One final note, about the prurient ID requests: this is likely something our community will have to address in the future in a larger way. While not all anime appeals to prurient interests, it does seem that anime deals with sex and romance more often and more pruriently than is typical in the works that SFF.SE and Movies.SE cover. The anime fandom is the fandom that made up terms like "trap", "shimapan", and "lolicon", along with entire specialized vocabularies for describing fictional homosexual relationships designed for the titillation of heterosexual readers of the opposite sex from the characters (yaoi, yuri, uke, seme, etc.). We will likely have to deal with this in a more proactive way than SFF or Movies if we want to keep things fairly clean, and ID requests, since they're so popular with new users, could be ground zero for this issue.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .