Tag Suggestion: Reducing cum_on_* tags

jxh2154 said:
I don't think this argument is as strong when it comes to combination tags though. Food, yes. Cum, yes. But cum_on_food? People can just search cum and food and find them it themselves. There's nothing wrong with that. Combination tags should be used when a search of the two tags together brings far, far more images than can be weeded through manually (not the case here) and when the joint concept itself is well established as a single unit of thought.

I can understand this, though obviously combination tags do add information as they indicate the two tags interact in the composition of the image, whereas the simple presence of two tags only indicates their existence in the image, and not if the two tags interact.

In general, I don't like combination tags (because there are better, but currently infeasible ways to describe such situations), but recently we have been supporting and creating new such tags rather than deleting them. By your logic we should also kill white_school_swimsuit, striped_thighhighs, and short_twintails.

jxh2154 said:
Otherwise what's to say cum_on_food is much more significant than cum_on_painting or cum_on_car or cum_on_book? Or book_in_bowl or panties_on_statue or... you get the point. If cum_on_food were an exceptionally well established tag it'd seem more convincing, perhaps, but instead it just looks like someone figured "I like food and I like cum so let's put them together!" and only two dozen images ever used it and well... I'm not seeing it, I guess. It doesn't seem to pass the test to justify a combo tag.

On the other hand, it is in greater use than 15 *_thighhighs or *_swimsuit combination tags. The reason why book, bowl, or statue aren't important is simply because they aren't used, moreover those items are more likely to be background props rather than the focus of an image.

Again, I'm guessing these tags offend me less than others in this thread, but the principal reason I'm arguing this is because it seems to be a double standard compared to our other actions, and is potentially destructive in terms of semantic content on the site.

We are entitled to regulate the images we accept on this site, for example we banned semenonfigure, finally purging all the images from that tag. We never banned the tag itself though, which is good because it was descriptive and allowed those images to be identified as such.

In all honesty, I don't really care much what we do with these tags, but I don't quite understand the double-standard either and am hesitant to set precedents where we wholesale delete batches of tags.

Shinjidude said:
By your logic we should also kill white_school_swimsuit, striped_thighhighs, and short_twintails.

On the other hand, these tags aren't vulgar/undesired. It's not so problematic having 100 tags people don't mind seeing versus 50 tags people don't really like seeing. Changing 50 into 10 is worth what's lost in this case (it must be anyways since action is being taken)
Rest of post was tl;dr sorry lol.

I guess that's my problem. I don't see how being vulgar/undesired is reason enough to delete tags. If we don't want to see it, we should ban content, not the description of content.

In many cases we *should* tag what we don't want to see: jpeg_artifacts, lowres, guro, watermark, duplicate... It helps the moderation process. Deleting redundant tags is fine, but deleting a bunch of stuff simply because we don't like the terminology sounds unjustified.

Shinjidude said: By your logic we should also kill white_school_swimsuit, striped_thighhighs, and short_twintails.

Well, regarding those, I'm not a fan of many of the color/pattern_clothingtype tags and have argued against some of them before. white_school_swimsuit specifically, I believe.

However, I think they're much better than cum_on_food because particularly in the case of "white" and "short" we don't have standalone tags that really convey that idea. cum_on_food is a noun (cum) on a noun (food), both of which are perfectly taggable concepts on their own. "White" however is not, because it's simply an adjective. A "white" tag would be exceptionally useless. Same with short. We do have a "striped" tag but I've always felt that it didn't mean much on its own - but I guess it's more readily interpreted by people as being clothing-related than white or short would be, because they're more likely to refer to many other things.

About all you could do is "white_clothing" "school_swimsuit" and I think that's just as bad. Although I'm always hesitant to mention tags in jest, because people will actually want them to be used... in fact, that's exactly how short_twintails was created. =P Although, again, much more reasonable tag than cum_on_food.

Lastly, those three have very high post numbers on their side.

In all honesty, I don't really care much what we do with these tags, but I don't quite understand the double-standard either and am hesitant to set precedents where we wholesale delete batches of tags.

Tags are rarely deleted, and this thread won's be used as an excuse to do so in bulk in the future either. It's a specific response to a specific and frequently cited problem, i.e. the ridiculous explosion of cum tags. I can't think of another comparable problem on the site today that would necessitate similar action.

Lokispawn said: I still don't get what cum on upper body & cum on lower body is supposed to gain us.

I was perfectly ready to make them all cum_on_body, for the record, and agree they're a bit... awkward. But it was brought up as a way of making the change a *bit* less drastic, so I just kinda rolled with it. If there were a sudden massive movement to have them all just go to cum_on_body you would receive no resistance from me. Although I'd have to delete and redo all those aliases. =P

Godel- I'm gratified you don't find my argument annoying. Yet. However, I gather I've been misunderstood once more; I'm not lamenting any missing tag.

The rationale behind the adoption of an upper/lower body distinction (even in spite of the confusion over whether it was intended to cover clothing or body parts or both) was very much an exercise in specificity. Or rather, to substitute in a general sense for what until recently had been specified according to the various enumerated fetishes. As a practical matter, generalizing the location of the cum resulted in a functionally useless tag. It's simultaneously too vague to serve in it's original capacity, that of indicating what's being soiled, and too redundant with the concise umbrella term cum on body to justify it's independent existence.

So what's wrong with all three being on a post is only slightly more apparent than what's wrong of having either alone + body. Essentially, there's three iterations of tags saying "there's cum somewhere on this girl."

To reiterate, cum on upper/lower body is useless and can just as well be nuked. Of course, I'd rather not know how much work is involved so that I may make such a recommendation with a clear conscience.

Shinjidude said:
In all honesty, I don't really care much what we do with these tags, but I don't quite understand the double-standard either and am hesitant to set precedents where we wholesale delete batches of tags.

This is exactly my opinion as well. My stance on porn is well-known, but if we allow porn at all, and allow images that can meaningfully use cum_on_left_pinky or whatever, then deleting tags just to delete them is a huge departure from our entire history of policing so far. I mentioned in another thread that while porn always loses to other things, we aren't actively interfering with browsing porn when it doesn't conflict with anything. This is clearly no longer true.

It's not like all the cum* tags get sprinkled on random images and jump at you without warning. If you browse images with cum (gah, I still hate the word) in them, you should expect the relevant tags as well. If you don't want to see them, blacklisting is easy. Safebooru is even easier.

I haven't been around when the thread started, so I couldn't really voice my opinion then, but I must say I find this development troubling, on the scale of the ill-conceived "let's delete all non-animu posts lol" purge. This is not how I would like it dealt with, and I can't say I understand or agree with the motives behind.

jxh2154 said:
Lastly, those three have very high post numbers on their side.

Which could be very well taken as a *downside* of them. The more uncommon the tag, the more information it conveys. If having a high post count was really important, the best tag would be one that was applicable to every single post. And that's clearly nonsense.

Okay. Gonna give my humble opinion here. I understand the need to reduce the number of cum_on_* tags.
I actually like their use, 'cause I like cum on specific places.

On the three tags that seem hard to decide, all I can say is that I am one that actually searches for both "cum_in_eye" and "cum_on_food".
Nothing wrong with searchin cum+food, but it will return pictures with cum and food, but no cum ON food.

葉月 said: Which could be very well taken as a *downside* of them. The more uncommon the tag, the more information it conveys. If having a high post count was really important, the best tag would be one that was applicable to every single post. And that's clearly nonsense.

The high post count in and of itself does not mean much. "fanart" would be an example of "high post count = good" gone wrong, not something as specific as white_school_swimsuit . The fact that they are rather specific and *still* have high post counts (i.e. lots of images feature them) is the reason why they're more justifiable. I still didn't say I was a fan of any of them (I said the opposite in fact), but their popularity does count in their favor.

And Hazuki, we've been talking about tightening up the cum tags for ages, we just never got around to actually taking the time to d it.

I simply do not agree that if a tag is technically applicable it is inherently justified in existing. Your argument is dangerous because it has no end. cum_on_third_knuckle_of_left_hand conveys even more specific "information" than the tags that got aliased, and absolutely nothing in your argument thus far precludes going to that level. And you cannot say "but of course we wouldn't use that, that's ridiculous, it's too specific", because then you're making exactly the same kind of value judgment you're criticizing here. You'd just be applying it to a different level of specificity.

"But the tags already existed" is not all that convincing either, given that we have no approval process for tags. They were never validated in the first place, never agreed upon, never decided as conveying useful information. Any tag can be created at any time, and can go for a very long time before anyone notices it if it's small. So the strict definition of information loss as "removing any information that existed" is one I don't think is useful.

Simply "conveying information" is not automatically the sign of a worthy tag. Yet that's where the argument is leading.

I already tried to make a case for salience. cum_on_third_knuckle_of_left_hand is no more salient than the cum_on_thighs I mentioned above and which has already been aliased. Being rare doesn't necessarily mean something is good, nor that it mean that it conveys important information. cum_on_{various_unimportant_body_part} is typically subsumed by cum_on_body and not salient enough to stand on its own.

Where a tag is rare, salient, and not subsumed by any current tag, it conveys strong usable information.

jxh2154 said:
And Hazuki, we've been talking about tightening up the cum tags for ages, we just never got around to actually taking the time to d it.

First, we've been talking about how much we hated the tags, which is not exactly the same, and I'm aware of that. Second, we have been talking about what exactly qualifies as danbooru content for ages too, and then when the purge came, it didn't make the implementation any less hasty, sly and reminescent of pushing a personal agenda. "I have been talking about how I'd kill him some day for ages, so why are you all suprised that I shot him?" is not a valid defense.

I simply do not agree that if a tag is technically applicable it is inherently justified in existing. Your argument is dangerous because it has no end. cum_on_third_knuckle_of_left_hand conveys even more specific "information" than the tags that got aliased, and absolutely nothing in your argument thus far precludes going to that level.

Yes, there is. Namely the fact that no-one uses or searches for cum_on_third_knuckle_of_left_hand. Like it or not, semen and other bodily fluids are very sexualised, and thus there's a high concentration of various, often specific preferences revolving around them. It doesn't mean it's superfluous, just like many words for snow aren't when you are dealing with snow all day.

"But the tags already existed" is not all that convincing either, given that we have no approval process for tags. They were never validated in the first place, never agreed upon, never decided as conveying useful information. Any tag can be created at any time, and can go for a very long time before anyone notices it if it's small. So the strict definition of information loss as "removing any information that existed" is one I don't think is useful.

But it still departs from our previous policy of not interfering actively with porn browsing. That policy hasn't been explicitly abandoned, and that's what bugs me. And it was executed under the pretence of preserving information via the aliases to cum_on_{upper/lower}_body, which are in reality patently useless.

Yes, we had precedents of sorts, for example with that pee_on_* thread someone started, and we rejected it summarily. It's true that the main difference between that and cum_on_* is mainly that no-one has announced the latter up-front. But if the goal was truly to reduce the undesirable specificity while preserving as much as possible, a much better way would've been to select a couple of strongest numerous tags (such as cum_on_breasts, facial, what have you), keep them intact, and fold the tiny ones. This way the loss of specificity mostly causes tiny tags to be merged into slightly less tiny ones, which is relatively easy to recover from. But with the whole upper/lower body business, *everyone* loses.

葉月 said: But if the goal was truly to reduce the undesirable specificity while preserving as much as possible, a much better way would've been to select a couple of strongest numerous tags (such as cum_on_breasts, facial, what have you), keep them intact, and fold the tiny ones.

Uh, that's exactly what happened.

jxh2154 said: Which I hear what john1980 is saying about _in_eye, it probably doesn't get enough posts for facial to be insufficient, honestly.

Facial is insufficient for the posts that have _in_eye. _in_eye conveys information that facial does not. Is that information important/non-trivial? I would say yes. Eyes are a big deal. We see with them O_O

----------------------

In general, the cum aliasing has been overkill. I'm not sure what is so bad about on_belly, on_arm, on_hands, on_foot etc. Those are not overly specific body parts, but still specific enough to be useful. upper/lower_body are too generic to be of any use.

I think cum_on_tongue->tongue implication (or the manual retagging of them) and alias cum_on_tongue->cum_in_mouth would make more sense as well as getting rid of a tag.

This will make cum_on_tongue still searchable by searching tongue cum or tongue cum_in_mouth (with only a few extra results comming up)