Madoka Magica group tag question

AngryZapdos said:
Additionally, there are a mere 45 posts also tagged with Magia Record. That's just shy of F/GO's 1% pull rate for 5-stars. Calling this "fairly common" is disingenuous at best.

As in: a decent percentage (>20%, surely a generous gacha percentage) of posts outside that are MagiReco, showing that it contributes pretty meaningfully to any perceived issue in the search (especially as if you're arguing for a two-tag search it must be ordered by recently posted, and such posts are increasingly frequent which is at least in part due to MagiReco)

As far as a need for the tag itself, I don't think one necessarily needs to exist: "hol(l)y quintet" is just something said in Rebellion once after a sequence of those five characters' transformations (to my knowledge, it's not used as an official term ever again?) and it's mostly just a fan term for "the primary characters of the Madoka Magica original anime + whoever I want to include from a sequel or spinoff that works with those five": any construction that's more vague than "exactly these 5 characters" would certainly lead to overtagging.

CoreMack said:

To be more precise, in every 14 results there is one false positive.

As a general rule, for danbooru, that's not "perfectly acceptable," that's actually quite terrible. Imagine if you searched for hetero sex and every 14th post was yaoi.

Apples and oranges - it wouldn't be acceptable for even 0.1% of the hetero sex search to just be yaoi. It's not helpful to conflate what is essentially eye sandpaper for the searcher to innocuous posts from the very franchise they're already looking through.

zetsubousensei said:

I don't know if that's a far comparison but also mahou_shoujo_madoka_magica_(anime) 5girls is (theoretically) this tag with no false positives.

Sorry, but unfortunately that's an even worse replacement search.

Post count for mahou_shoujo_madoka_magica_(anime) 5girls: 2690
Post count for mahou_shoujo_madoka_magica_(anime) 5girls (-kaname_madoka or -akemi_homura or -miki_sayaka or -tomoe_mami or -sakura_kyoko): 217
= False positive every 12.4 posts

Beyond that, I'm seeing more than 50 false negatives, in the form of animalization or genderswap_(ftm) posts featuring all five characters which wouldn't come up on either of the 5girls searches.

I think PMMM anime as a tag is a bit questionably overtagged; if mahou_shoujo_madoka_magica:_hangyaku_no_monogatari isn't going to imply it, the presence of the thread-discussed important characters shouldn't really be sufficient. Somewhat off topic, but I would guess that a restrained version of that tagging would be very high percentage (but also exclude many posts that do focus on all 5 of those characters)

AngryZapdos said:

Apples and oranges - it wouldn't be acceptable for even 0.1% of the hetero sex search to just be yaoi. It's not helpful to conflate what is essentially eye sandpaper for the searcher to innocuous posts from the very franchise they're already looking through.

I'm obviously exaggerating to make a point. The point is that a 1 in 14 false positive result (with a decent handful of false negatives) is a pretty bad result, not remotely acceptable. Maybe a desire not to have tag bloat weighs heavily enough against it, but it's objectively a quite bad result that we shouldn't be happy with.

If this tag is bloat, then we should also nuke: kessoku_band (replaceable with gotoh_hitori 4girls), holomyth (replaceable with gawr_gura 5girls), etc...

I don't know why searches with 5girls are suddenly b eing proposed as replacements for group tags, while a key feature usually is that you can find genderswaps, animalizations, and other indirect and non-orthodox representations that involve the entire group.

CoreMack said:

I'm obviously exaggerating to make a point. The point is that a 1 in 14 false positive result (with a decent handful of false negatives) is a pretty bad result, not remotely acceptable. Maybe a desire not to have tag bloat weighs heavily enough against it, but it's objectively a quite bad result that we shouldn't be happy with.

I really don't understand all this doom and gloom over what's essentially just ≤2 posts per page of false positives. Nobody is going to be complaining when the 18 other posts on every page are all exactly what they're looking for. As was mentioned earlier, it would be different if those ≤2 posts were all something most users explicitly avoid, like yaoi or bestiality or futanari, but they're not.

If this tag is bloat, then we should also nuke: kessoku_band (replaceable with gotoh_hitori 4girls), holomyth (replaceable with gawr_gura 5girls), etc...

If those searches easily replace those group tags, then yes, those group tags should be nuked.

AngryZapdos said:

I really don't understand all this doom and gloom over what's essentially just ≤2 posts per page of false positives. Nobody is going to be complaining when the 18 other posts on every page are all exactly what they're looking for. As was mentioned earlier, it would be different if those ≤2 posts were all something most users explicitly avoid, like yaoi or bestiality or futanari, but they're not.

If those searches easily replace those group tags, then yes, those group tags should be nuked.

I don't understand all the doom and gloom about tag bloat when the tag proposed in this thread does no harm and is exactly as useful as all of the hundreds of other group tags we have on the site. There's well established precedent for exactly this type of tag. We either allow this tag, or nuke all of them.

They're useful since there is no replacement search for them that doesn't have tons of false negatives and false positives. There's side characters that can throw off the 5girls search. There's crossovers that throw off the 5girls search. There's genderswaps, there's animalizations, there's abstract representations of the group, all of which get missed.

Group tags are very clearly useful for fans of the show and they literally do no harm. You can't just cry "tag bloat" to get rid of a useful, informative tag that makes something searchable that isn't otherwise searchable. Go nuke something that actually serves no purpose like closed mouth instead.

Freshblink said:

But it gives you a lot of false positives if you're looking for a line up of them in their magical girl outifts.

Are you still insistent on only using holy quintet for their magical girl depictions even though multiple people have told you that HQ just refers to Madoka/Mami/Sayaka/Kyoko/Homura as a group in general?

AngryZapdos said:

I really don't understand all this doom and gloom over what's essentially just ≤2 posts per page of false positives. Nobody is going to be complaining when the 18 other posts on every page are all exactly what they're looking for. As was mentioned earlier, it would be different if those ≤2 posts were all something most users explicitly avoid, like yaoi or bestiality or futanari, but they're not.

If those searches easily replace those group tags, then yes, those group tags should be nuked.

You can't just say "easily" as if the definition of that isn't part of the discussion here. Posts like post #7747221 and post #7934182 are easily found with group tags (guess how I found them), but wouldn't be caught with 5girls and 4girls or chartags:4.

Even if a 1 in 15 false positive rate sounds fine to you it doesn't to me), having to wade through dozens of false positives when trying to find something specific is stupid.

CoreMack said:

Group tags are very clearly useful for fans of the show and they literally do no harm. You can't just cry "tag bloat" to get rid of a useful, informative tag that makes something searchable that isn't otherwise searchable. Go nuke something that actually serves no purpose like closed mouth instead.

I absolutely can cry "tag bloat," because the logical conclusion of green-lighting these redundant padding tags is that every series that has any named groups will have them all tagged regardless of how little purpose they serve. Take Code Geass, for instance - we could potentially end up with black_knights_(code_geass), ashford_student_council, the_four_holy_swords, peace_mark_(code_geass), knights_of_the_round_(code_geass), geass_order, glaston_knights, valkyrie_squadron, six_houses_of_kyoto_(code_geass), chinese_federation_(code_geass), japan_liberation_front_(code_geass), britannian_imperial_family_(code_geass), glinda_knights, and more that I've probably forgotten. Go populate something that would actually serve a purpose like hand_under_another's_clothes instead.

If people are actually searching and using tags like holy quintet or kessoku band, then I don't see why they need to be nuked.

AngryZapdos said:

I absolutely can cry "tag bloat," because the logical conclusion of green-lighting these redundant padding tags is that every series that has any named groups will have them all tagged regardless of how little purpose they serve.

You listed examples of group tags that potentially nobody will use, but I don't understand why you're using this argument to nuke most group tags that people do actually use. And as others have pointed out in this thread, these group tags can't easily be replaced with char_tag xgirls, because you have cases where characters are depicted as animals, and Gotoh Hitory being as some kind of slime or something. This seems to be the crux of your argument that you haven't addressed.

AngryZapdos said:

I absolutely can cry "tag bloat," because the logical conclusion of green-lighting these redundant padding tags is that every series that has any named groups will have them all tagged regardless of how little purpose they serve. Take Code Geass, for instance - we could potentially end up with black_knights_(code_geass), ashford_student_council, the_four_holy_swords, peace_mark_(code_geass), knights_of_the_round_(code_geass), geass_order, glaston_knights, valkyrie_squadron, six_houses_of_kyoto_(code_geass), chinese_federation_(code_geass), japan_liberation_front_(code_geass), britannian_imperial_family_(code_geass), glinda_knights, and more that I've probably forgotten. Go populate something that would actually serve a purpose like hand_under_another's_clothes instead.

"For years, users have been creating tags for specific groups of characters who are closely associated with each other, e.g. bands and vtuber generations. They have been responsible about this, and have only created a small handful of tags per copyright (often only one) for named groups that are central to the copyright. Fans of the copyrights like these tags, since there's no other 2-tag way to search for the members of that group all together. However, we cannot allow this practice to continue, because what if they suddenly became extremely irresponsible and created thousands of useless tags?"

CoreMack said:

"For years, users have been creating tags for specific groups of characters who are closely associated with each other, e.g. bands and vtuber generations. They have been responsible about this, and have only created a small handful of tags per copyright (often only one) for named groups that are central to the copyright. Fans of the copyrights like these tags, since there's no other 2-tag way to search for the members of that group all together. However, we cannot allow this practice to continue, because what if they suddenly became extremely irresponsible and created thousands of useless tags?"

Are you purposely forgetting the time we had to nuke a bunch of "group" tags from Blue Archive because they were for literally just two characters? I'm reminded of how we said "yes" to some x_school_uniform tags, and as a result we now have a shitload of useless ones that are analogous to a "copyright school_uniform" search because the copyright in question only has one school uniform. I'm not against all group tags; the remaining Blue Archive ones, for instance, would need you to search for all characters of the group without them because Blue Archive has like a hundred characters. I'm only against useless bloat tags like this one, where the vast majority of the posts are already searchable because the extremely popular anime's cast only has like 5 side characters with barely any posts to pollute searches.

HyphenSam said:

...as others have pointed out in this thread, these group tags can't easily be replaced with char_tag xgirls, because you have cases where characters are depicted as animals, and Gotoh Hitory being as some kind of slime or something. This seems to be the crux of your argument that you haven't addressed.

I beliveve I said "if" those searches easily replace those group tags, then yes, those group tags should be nuked. When you have a series like Bocchi where the character(s) are frequently depicted as animals or other non-humanoid things, then yes, group tags have a reason to exist because gender counts are not reliable and you could miss several hundred posts. However, when the argument for having a group tag that's thousands of posts strong becomes "how else will I find these 7 posts where they're depicted as animals", it's difficult to see it as anything other than bloat.

AngryZapdos said:

Are you purposely forgetting the time we had to nuke a bunch of "group" tags from Blue Archive because they were for literally just two characters? I'm reminded of how we said "yes" to some x_school_uniform tags, and as a result we now have a shitload of useless ones that are analogous to a "copyright school_uniform" search because the copyright in question only has one school uniform. I'm not against all group tags; the remaining Blue Archive ones, for instance, would need you to search for all characters of the group without them because Blue Archive has like a hundred characters. I'm only against useless bloat tags like this one, where the vast majority of the posts are already searchable because the extremely popular anime's cast only has like 5 side characters with barely any posts to pollute searches.

2-person group tags are quite different from the tags being discussed here as "is it just the same as a 2-tag search" is usually the first litmus test people use to see if a tag is worth keeping. 2-person group tags don't even pass that, the tags in question here do.

I beliveve I said "if" those searches easily replace those group tags, then yes, those group tags should be nuked. When you have a series like Bocchi where the character(s) are frequently depicted as animals or other non-humanoid things, then yes, group tags have a reason to exist because gender counts are not reliable and you could miss several hundred posts. However, when the argument for having a group tag that's thousands of posts strong becomes "how else will I find these 7 posts where they're depicted as animals", it's difficult to see it as anything other than bloat.

Bocchi is just an easy example. All it takes is a single artist who likes animalization and whoops too fucking bad now you can't easily find these posts.
So now we have an argument that "group tags can go away if":

• The copyright has no animalization art
• The copyright has no genderswap art
• There's not too many side characters (How many are too many? Who knows)

And we can probably continue finding "exceptions that make it fine". What's the point at this point.

ANON_TOKYO said:

2-person group tags are quite different from the tags being discussed here as "is it just the same as a 2-tag search" is usually the first litmus test people use to see if a tag is worth keeping. 2-person group tags don't even pass that, the tags in question here do.

This was in response to the idea of "group tags are only ever maintained with the utmost care and precision". I'm well aware that group tags for three or more characters can be useful, which I explicitely stated just one sentence later.

Bocchi is just an easy example. All it takes is a single artist who likes animalization and whoops too fucking bad now you can't easily find these posts.
So now we have an argument that "group tags can go away if":

• The copyright has no animalization art
• The copyright has no genderswap art
• There's not too many side characters (How many are too many? Who knows)

Group tags are bloat if you can find the majority of their results with a two-tag search. That is and will always be my only point.

I don't think people actually search for these, though I'm certain if I'm wrong someone will correct me. I understand that they do have some theoretical utility, but that utility is, frankly, minimal in comparison to the massive bloat they invite and actively have. People don't maintain these tags. When I said this would happen with the Blue Archive tags, people swore they'd be good with them. Well, some of them haven't been gardened in nearly a year, many are barely used on proper images, and several of the people who made them aren't even active anymore.

Despite what was said earlier, y'all aint usin' group tags responsibly, usefully, or consistently. People generally don't make these tags in the first place unless it's "their" copyright or something they just happen to get attached to. They're somewhere between bloat tags and pet tags.

Veraducks said:

...

Despite what was said earlier, y'all aint usin' group tags responsibly, usefully, or consistently. People generally don't make these tags in the first place unless it's "their" copyright or something they just happen to get attached to. They're somewhere between bloat tags and pet tags.

This is hardly a new thing or a surprise, people garden the tags they use. This is good because otherwise a bunch of people who have no clue as to the importance of specific considerations will end up making misinformed decisions. That's obviously not to say anything goes, which is why forum discussion is important, but this *generally* happens.

AngryZapdos said:

Group tags are bloat if you can find the majority of their results with a two-tag search. That is and will always be my only point.

But what is the cutoff? To me 1 in 15 seems pretty bad. I use a page size of 99 (9 rows of 11 on 1440p) , and if I find 7 obvious mistags when gardening my takeway is that someone was pretty careless.

Also, using Kessoku Band as an example since it's easy despite there being no good replacement. Even if there was, and we tell people to "just" use bocchi_the_rock! 4girls (or worse: bocchi_the_rock! chartags:4. If I'm a casual browser looking for some Kessoku Band pictures, type it in and get greeted with a cryptic search term to use instead (best case) or nothing at all (worst case), my browsing adventures would end pretty quickly and I'd be pretty annoyed.

We already somewhat restrict group tags to mean "everyone has to be present", while it's very common to use those group names as catch-all terms for "any character from that group" on i.e. Twitter and Pixiv. Completely ignoring that actual usage on the internet because of some perceived problem of tag bloating is ridiculous.

And then there's still the false positive/negative rate. The first 99 posts of bocchi_the_rock! 4girls contain 65 Kessoku Band posts and bocchi_the_rock! chartags:4 61, so we're talking 1 in 3 posts. These are also mostly not pedantic "not quite", but a lot of crossovers, parodies, etc., somethign every copyright is susceptible to. Unless you want to replace every one of such tags with a unique "simple" search, I highly doubt there is a good one-size-fits-all search replacement (and why would there be).