SystemError said:
But that makes me wonder why we have a "demon_girl" tag when we have tags like "horns", "pointy_ears" (and thus no "elf" anymore), "slit_pupils", "demon_wings", "demon_tail" and even "blue_skin" or "grey_skin" if necessary. By the "official" tagging policy here, wouldn't be "demon_girl" (and arguably "monster_girl" too, since that's more categorization than tagging) on the same level of redundancy as the composite tags "schoolgirl", "catgirl" and "elf"...?
Perhaps, but this is an inconsistency from Danbooru's side. As I've said, our tagging policy is very linked with theirs. If they adopt a large-scale tag like this, we're more or less forced to comply, as 70% of all posts on the site come in from Danbooru as opposed to our own users.
As for why Danbooru did it, it's probably because of technical problems. While catgirl is easy to alias to cat_ears and let the remaining missing tags filter in later, and most posts that would end up tagging something "schoolgirl" would be doing it because of a present school_uniform instead of other features, that's also a reasonable alias; however, there is no such similar alias for demon_girl. This means that breaking the composite tag would take a lot of manual work in one way or another. And, of course, if this is a problem for Danbooru, it's a worse problem for us, since our alias system can be taken as a last resort for tags that are way too frequently misused to correct with mass tagging, and because we have a
much smaller number of people actively participating in tagging.
Indeed, but it's still a question how to determine if the tag "kitsune" applies. I would go for the presence of the hitodama and/or nine tails and/or any signs of Inari, like rice and/or a white fox. These could be objective criteria, any more would be subjective. We need to tell mythological foxes in human form and plain foxpeople apart somehow. (Btw, I'm not a furry, but I adore kemonomimi.)
Yes, this is the correct way to use the tag.
Thanks, I'll try to keep that in mind. Why I said hoshi no tama instead of this...?
"star ball" sounds odd indeed.
I thought that Gelbooru's tagging and wiki system is more independent from Danbooru's. The last time I visited Danbooru in-depth was pretty much years ago when they restricted loli content, but as far as I know, they still have the catgirl tag, no? Or at least, there are some fundamental tagging differences between them and us.
Nope, not at all. Quite the opposite, in fact. We are
very dependent on their tagging policies, since as I said, 70% of our posts come in from Danbooru already tagged with whatever tags they end up with over there. Danbooru has a decent number of users who tag images pretty well and fairly quickly, so they usually end up coming in with more tags per image than a lot of our uploads do. This pretty much forces us into using those tags as well.
As for fundamental tagging differences, there aren't really. Maybe near the beginning, before it became apparent exactly
how dependent we are on Danbooru's tagging policy, but ultimately it's a lot easier to just go with what Danbooru does, since it affects us so much. There were some aftereffects of this early stage that hung around until about a year ago, but those only managed to stay because we couldn't really do anything about them. The number of tag changes that would be required to make a dent was on the order of ten, maybe hundreds of thousands, and back then there was no way to tag things other than manually. This more or less forced us into using old policies that we couldn't change. The worst part was that we were also, at the same time, forced to use Danbooru's policies, which meant that there were a lot of redundant tags that we just couldn't do anything about.
But now we have the mass tagger. This enables us to clear tags with 10k+ posts in the span of a couple hours. If you look at many of the mods' profiles, you'll see really big numbers for their tag edit counts. For most of us, the bulk of that came within the first couple of weeks after the tagger was introduced, fixing up things that we knew needed fixing but weren't able to before.
...speaking of foxes, why do you mods hate the "animal_tail" tag? I know, I know, it's might be redundant, since it equals to "tail -demon_tail -some_other_misc_tail" (given that taily images are properly tagged), but it describes a visible thing such as "black_legwear", which is btw implies ONLY:
"black_pantyhose", "black_stockings", "black_thighhighs", "black_kneehighs", "black_socks" and PROBABLY "single_kneehigh", "single_thighhigh" (5+2 tags),
whereas the tag "animal_tail" implies AT LEAST the following:
"cat_tail", "dog_tail", "bunny_tail" (why not "rabbit_tail"?), "fox_tail", "wolf_tail", "tiger_tail", "lion_tail", "cow_tail", "sheep_tail", "mouse_tail" (10 tags, all have examples on Gelbooru).
The difference here is functionality. Back before legwear color tags, there was a very large number of legwear color + legwear type composite tags. At the same time, there was the legwear type tags all on their own. This means that the legwear color + legwear type composite tags were not only composite tags, but also directly redundant. Another side-effect was that there were a
lot of them. There are a total of 12 color tags (and a whole bunch more patterns and styles), and four base styles of legwear, which led to a grand total of 48 legwear color + legwear style tags. So, the legwear color + legwear type tags were all merged into just legwear color, and legwear type was left separately tagged like it had already been. This means there are currently a total of only 12 color tags + 4 style tags, for a total of only 16 tags, with the exact same functionality as was present before, plus extra added functionality in being able to search across the legwear color tags all at once, which you couldn't before.
In short, the color legwear tags were made to
solve a pretty big tagging problem that already existed.
However, the animal_tag you're proposing doesn't solve any tagging problems. It doesn't condense a bunch of tags into a much smaller number, it doesn't break existing composite tags down into smaller pieces; all it does is give you a shortcut to finding things quickly. That isn't enough of a reason to create a redundant tag. I'll remind you that redundant tags in general are against tagging policy.
So the issue is consistency versus redundancy. In my opinion, if there are generic, objective catch-all tags are in place for colors of legwear (without catering to any fetish, as it seems) which could be searched quite comprehensibly with as few as 5 tags compared to 10+, then there is no problem with a generic, objective catch-all tag for all tails that are animal in nature. (Note that even when we have angel_wings, demon_wings and mechanical_wings, we have wings too, while, technically they would be searchable with a 3-tag combination.)
Your scale is a bit off, as explained before. Instead of 10+ tags being condensed into 5, there were 48 tags. This still reduces into 16 tags, and only going by color. As I said, there were many patterns (stripes, dots, etc) and other style features not part of the main style (frills, lace, torn, etc) that were also condensed. that was a gigantic clusterfuck of hundreds of tags when there only needed to be a handful. Merging all those tags created a very significant improvement in actual tagging, not just usability. Your proposed change will slightly improve usability, by creating a shortcut to something that could already be found but with a longer search string, but it can only cause problems in tagging.
As for wings, I can see where you're going, but that's wrong too. Base tags like "tail", "ears", and "wings" are required before any of those other subtypes even come to the table.
It also seems you're misunderstanding the search. Searching for all three of those types of wings should only return images of multiple characters or of characters with multiple sets of wings, since searches are subtractive. There is no tag combination using the other existing wing tags to just get "wing", as there is no functioning or operator in searches. There is an existing tag combination for animal tails, though, in the tail -all_tail_types_that_are_not_animal.
Some may argue that generally "tail" means "animal_tail", but that's not always true. Another argument could be that nobody wants to search for stuff based on animal tails without respect to animal ears, since usually the tails are missing, not the ears, thus ears matter (and one could just sift through them to look for tails or their absence), but this isn't always true either.
This isn't what we're saying at all, and how often people would use a tag is a very small factor in deciding whether to implement it.
If the "animal_tail" tag is bullshit, then what could one search for if one wants to make absolutely sure to find animal tailed characters without writing in all the applicable animal_species_tail tags and without relying on the more general and often not tagged (I know, I should tag that already!) "tail" tag with tag negation to account for the demonic stuff cropping up? So, consistency or redundancy is the more important key on Gelbooru?
Sry for being a provocateur against teh system now. :3
Once again, you seem to be misunderstanding the tagging system, as you
wouldn't be writing in all of the animal tails, as that would only return images of multiple characters or of characters with many, many tails. Your search should be for tail -demon_tail -dragon_tail -ghost_tail -pikachu_tail, which isn't even that big of a search string.
As for the negated "tail" string being undertagged, as I've said before, that has absolutely no bearing on our tagging policy. If you don't like that those tags are undertagged, don't try to propose additional tags; this will just worsen the problem as the tag you're proposing will
also be undertagged, and probably even worse so. The correct thing to do is to start going through and tagging them properly.
However, just to throw you a bone, I'm going to ensure that all of the posts tagged with any specific type of tail are also tagged with the base tail tag. That should help, no?