Topic: Policy on redundant color tags

Posted under General

I've had a look around in the wiki and forum but can't see anything that specifically indicates whether over-specific tags are warranted or undesirable in general. For example, tags like "green_hair" "green_arm_hair" "green_happy_trail" "green_chest_hair" "green_pubes" seem too specific to make those kind of tags truly useful, but they are insisted upon for orthogonal reasons. Like the artist insists on a personal set of tags, or the poster is interested in inflating the number of tags, or a contributor is just obsessed with green furred dragons or something.

I have in the past tried to edit tags into something less overly specific (eg: chest_hair and green_fur), but have seen issues where invalid tags have been re-added. Getting into a tagging fight on a post is the last thing I want (esp with a mod or janitor), if I am just trying to "clean up" a tag. Also, if someone re-adds an invalid tag without then going and fixing it, what is the best way to resolve this without either getting into an argument or giving up and ignoring it?

Is this something that posters should even be concerned about? Apologies if I've posted in the wrong place.

cimatrie said:
I've had a look around in the wiki and forum but can't see anything that specifically indicates whether over-specific tags are warranted or undesirable in general. For example, tags like "green_hair" "green_arm_hair" "green_happy_trail" "green_chest_hair" "green_pubes" seem too specific to make those kind of tags truly useful, but they are insisted upon for orthogonal reasons. Like the artist insists on a personal set of tags, or the poster is interested in inflating the number of tags, or a contributor is just obsessed with green furred dragons or something.

I have in the past tried to edit tags into something less overly specific (eg: chest_hair and green_fur), but have seen issues where invalid tags have been re-added. Getting into a tagging fight on a post is the last thing I want (esp with a mod or janitor), if I am just trying to "clean up" a tag. Also, if someone re-adds an invalid tag without then going and fixing it, what is the best way to resolve this without either getting into an argument or giving up and ignoring it?

Is this something that posters should even be concerned about? Apologies if I've posted in the wrong place.

Coming fro me , this isn't unexpected, but I want the tags as they are. Even if they are very specific, it probably helps searching.

I hate that cyan_* and panther tags are now unsearchable because some people can't be bothered to tag correctly.

In that specific case, it wouldn't affect me because I will never search for green_arm_hair, but I don't agree with you one bit.

Updated

cimatrie said:
I've had a look around in the wiki and forum but can't see anything that specifically indicates whether over-specific tags are warranted or undesirable in general. For example, tags like "green_hair" "green_arm_hair" "green_happy_trail" "green_chest_hair" "green_pubes" seem too specific to make those kind of tags truly useful, but they are insisted upon for orthogonal reasons. Like the artist insists on a personal set of tags, or the poster is interested in inflating the number of tags, or a contributor is just obsessed with green furred dragons or something.

I have in the past tried to edit tags into something less overly specific (eg: chest_hair and green_fur), but have seen issues where invalid tags have been re-added. Getting into a tagging fight on a post is the last thing I want (esp with a mod or janitor), if I am just trying to "clean up" a tag. Also, if someone re-adds an invalid tag without then going and fixing it, what is the best way to resolve this without either getting into an argument or giving up and ignoring it?

Is this something that posters should even be concerned about? Apologies if I've posted in the wrong place.

Don't "cut down redundant color tags" many of the color_* tags have implications. They are valid. Don't remove them...

Updated

When unsure, don't remove any tags without prior discussion with the community.

You can look through the post's tagging history or the post changes search tool to find out who has been using these tags.
If the tag has a lot of posts and has been used by multiple people, then it is likely that it is valid.
If the tag has only a few posts and is only used by one or two people, then the validity of the tag is dubious.

To clarify, my question was in regards to tags that are unique and/or invalid. Sorry that wasn't obvious above.

So, to rephrase, "skin" is an invalid tag due to being too generic. Someone creates a unique tag on their post, such as "mauve_skin" that has a post count of 1.
Is that something the admins to encourage or is it undesirable? If undesirable, how should we go about fixing it? If the site marks it as invalid, does that change anything?

I simply wanted to check if there was some official guidance already in place. If however the consensus is to just leave unique or invalid tags as they are, or get site staff involved, then I am satisfied with staying out of the tag editor.

cimatrie said:
To clarify, my question was in regards to tags that are unique and/or invalid. Sorry that wasn't obvious above.

So, to rephrase, "skin" is an invalid tag due to being too generic. Someone creates a unique tag on their post, such as "mauve_skin" that has a post count of 1.
Is that something the admins to encourage or is it undesirable? If undesirable, how should we go about fixing it? If the site marks it as invalid, does that change anything?

I simply wanted to check if there was some official guidance already in place. If however the consensus is to just leave unique or invalid tags as they are, or get site staff involved, then I am satisfied with staying out of the tag editor.

Skin would be like searching for -fur, so it is invalid. Further, I think fur is too generic of a tag to be useful similarly to skin, but that aside blue_fur is the generic tag + an adjective which makes it not generic. The same is true for blue_skin. You should replace mauve_skin with purple_skin or magenta_skin whichever is more appropriate, the color mauve is way too specific to be useful.

wolfmanfur said:
Skin would be like searching for -fur, so it is invalid. Further, I think fur is too generic of a tag to be useful similarly to skin, but that aside blue_fur is the generic tag + an adjective which makes it not generic. The same is true for blue_skin. You should replace mauve_skin with purple_skin or magenta_skin whichever is more appropriate, the color mauve is way too specific to be useful.

Thanks for following up. I mentioned skin only in that it is an invalid tag, but yes checking the wiki it does seem to encourage using a limited list of preferred colors as modifiers for that particular tag.
https://e621.net/wiki_pages/2694
To be honest I am not a fan of editing the tag color as it may not be obvious to all or even be authoritative even if there is a list. Did the list get updated when other things changed? Is it worth it to me to change, what is after all, a tag used on only one post, when I might pick the wrong choice or break things? Will it always be obvious that burlywood_skin is okay and mauve_skin isn't? What about for underwear or headwear?

I'd rather not make any change at all unless there's clearer guidance, and that's fine. However the site does tend to want to highlight those unique 1-post tags, and if I am fixing something obvious then I am already in the tag editor.

Edit to add: Found this Wiki entry. https://e621.net/wiki_pages/1980
However I feel that the entry isn't really appropriate for the editing of existing tags.

Updated

cimatrie said:
To be honest I am not a fan of editing the tag color as it may not be obvious to all or even be authoritative even if there is a list. Did the list get updated when other things changed? Is it worth it to me to change, what is after all, a tag used on only one post, when I might pick the wrong choice or break things? Will it always be obvious that burlywood_skin is okay and mauve_skin isn't? What about for underwear or headwear?

There is only a set number of basic colours that we actually use for most tags, i.e., black, blue, brown, green, grey, orange, pink, purple, red, tan, teal, white, and yellow.
Using other terms like burlywood_* or mauve_* is discouraged, if any exist they would have mostly been aliased to the closest colour tag listed above.
There are unique exceptions such as blonde_hair, but I won't be going into that. Bottom line is that most tags follow the set number of colours mentioned.

And yes, <color>_underwear and <color>_headwear do exist, as well as many other objects. You can find tags like green_underwear or yellow_headwear and they are accepted.

I'd rather not make any change at all unless there's clearer guidance, and that's fine. However the site does tend to want to highlight those unique 1-post tags, and if I am fixing something obvious then I am already in the tag editor.

Unique 1-post tags get highlighted to you because they are essentially unused tags. It just wants to point out to the tagger that this is a tag nobody has used before, in case it was a mistag.

There isn't a set rule of what tags get the colour treatment. Some may seem redundant while others may seem over-specific.
You don't really get to discuss their validity unless you bring it up to the forums first.

A good rule of thumb is for the colour tags to be focused on the characters themselves (e.g., blue_hair, tan_breasts), the clothing they wear (e.g., white_shirt, black_pants), and in a limited sense, the objects around them (e.g., pink_dildo, white_bedding).
Just don't go creating new tags like yellow_taxi, green_grass, or blue_bottle that people won't normally be looking at.

There is no rule against a tag being overly specific. If it's visible in the image, it's eligible for a tag. You should only remove tags that disobey that rule.

Overly specific tags will likely be severely undertagged and not very useful as a result, but that's a problem for the people who add them, not you.

thegreatwolfgang said:
There is only a set number of basic colours that we actually use for most tags, i.e., black, blue, brown, green, grey, orange, pink, purple, red, tan, teal, white, and yellow.
Using other terms like burlywood_* or mauve_* is discouraged, if any exist they would have mostly been aliased to the closest colour tag listed above.

Burlywood was specifically mentioned in the Wiki as acceptable under the invalid skin tag, I included it to highlight that lists get out of date, and that an "obviously correct" color name, isn't necessarily obvious.
I'll take the advice for new tags, but it doesn't change the argument for not editing tags.

thegreatwolfgang said:
Unique 1-post tags get highlighted to you because they are essentially unused tags. It just wants to point out to the tagger that this is a tag nobody has used before, in case it was a mistag.

wat8548 said:
Overly specific tags will likely be severely undertagged and not very useful as a result, but that's a problem for the people who add them, not you.

If I shouldn't be editing a tag because I am not a poster, then why do I have the ability? I don't want it if the OP is required to fix tags. If a tag is "undertagged" (used on only 1 post) or a "mistag" according to TWYS, then that makes it okay? I find it a weak incentive to "correct" bad tagging, especially if I have to litigate them individually on here.

This site seems to get a lot of its content from a proportion of users which use bulk uploaders via the API. You see posts where a whole tag cloud was largely just copy and pasted from a different site, perhaps with no intention of ever fixing the tags. It is not a problem per-se, contributors should be thanked for what they can provide. But e621 is then relying on others like myself to, perhaps come along later and fix up the worst and most obvious issues.

I asked for advice and the consensus seems to be, aside from raising each on the forum case-by-case, is that it is simply better to leave the little red "1"'s as-is.
[Edit: wording]

Updated

cimatrie said:
Burlywood was specifically mentioned in the Wiki as acceptable under the invalid skin tag, I included it to highlight that lists get out of date, and that an "obviously correct" color name, isn't necessarily obvious.
I'll take the advice for new tags, but it doesn't change the argument for not editing existing ones.

Seasoned taggers would know which basic colours to use all the time, but I understand your concern on outdated wikis being read by new taggers.
You are welcome to edit the numerous outdated wikis if you feel so inclined. Just make sure you do not remove any valid tags that are still in-use.

If I shouldn't be editing a tag because I am not a poster, then why do I have the ability? I don't want it if the OP is required to fix tags. If a tag is "undertagged" (used on only 1 post) or a "mistag" according to TWYS, then that makes it okay? I find it a weak argument for changing anything I don't have a stake in, considering replies so far.

This site seems to get a lot of it's content from a proportion of users which use bulk uploaders via the API. You see posts where a whole tag cloud was largely just copy and pasted from a different site, perhaps with no intention of ever fixing the tags. It is not a problem per-se, contributors should be thanked for what they can provide. But e621 is then relying on others like myself to, perhaps com along later and fix up the worst and most obvious issues.

I asked for advice and the consensus seems to be, aside from raising each on the forum case-by-case, is that it is simply better to leave the little red "1"'s as-is.

You are not obligated to remove any 1-post tags that you come across. You can always help by adding new (more established) tags to make the post more searchable.
If you are not confident in your ability to determine what tag is valid or invalid, leave it for more experienced taggers to go through them.

In my own experience, if I come across a tag that seems to be unused, I take a moment to consider whether it is in-line with the other valid tags that we are using.
For example:

  • Under the facial_hair umbrella, there are the beard and mustache subtags. Both are valid tags.
    • Beard's wiki mentioned various colour tags associated and implied to them, such as black_beard or white_beard. Both are valid tags.
    • Mustache's wiki does not mention any colour tags but numerous colour tags do already exist, such as black_mustache and white_mustache. Both seem dubious, due to their unused/underused nature.
      • Since mustache's counterpart, beard, has numerous established colour tags. I can deduce that mustache may also have their own set of colour tags.

Now a question for you, do you think purple_mustache (a currently 1-post tag) is an invalid tag and should be removed?

Updated

thegreatwolfgang said:
Now a question for you, do you think purple_mustache (a currently 1-post tag) is an invalid tag and should be removed?

Not going to comment on "confidence" or "seasoned taggers" because it doesn't remove any of the subjectivity from the decision making process, for me or anyone else. Is there a post score that lets you into the club? I am not interested in bothering power users on the forum for hot tips unless they want to post here.

That tag above meets TWYS of course, so obviously it is okay until someone finds a better argument. The gauntlets_only tag is more contentious, but I'm not arguing I should make the decision to remove it. I'm asking if there are any good policy-driven clear-cut cases where I shouldn't ignore it, and so far I've seen none. Again, I am okay with that.

cimatrie said:
If I shouldn't be editing a tag because I am not a poster, then why do I have the ability? I don't want it if the OP is required to fix tags. If a tag is "undertagged" (used on only 1 post) or a "mistag" according to TWYS, then that makes it okay? I find it a weak incentive to "correct" bad tagging, especially if I have to litigate them individually on here.

cimatrie said:
Not going to comment on "confidence" or "seasoned taggers" because it doesn't remove any of the subjectivity from the decision making process, for me or anyone else. Is there a post score that lets you into the club? I am not interested in bothering power users on the forum for hot tips unless they want to post here.

I've no idea what you're on about, my answer was quite simple and easy to follow.

Since you brought up the infamous "burlywood" tags, I'll point out that what happened to them was not, in fact wholesale removal as you are proposing. Instead, all the examples that existed were aliased to their "tan" equivalents. You are allowed to replace a 1-post tag with a more heavily-used tag that has the same meaning, if you can find one. That's not at all the same thing as removing it entirely, which causes information loss.

A good example here is character tags. I've been spending a lot of time recently going through the many, many overpopulated name_(disambiguation) tags and trying to find which tags should have been used instead. At least 50% of the time I end up having to create a new tag, and about 50% of those times I never use that tag on more than one post. But sometimes I get lucky. I'm always thankful when I type in the form of words I was planning to use and the autocomplete shows one existing post that already had that tag. It means somebody else went on the same journey as me, and chose to leave a helpful answer behind instead of giving up and assuming a lesser-posted character wasn't worthy of acknowledgement. It may have taken years, but eventually their foresight was indeed rewarded with a second post of the same character. I can only hope that one day one of my many single-post character tags will provide a similar signal to a future tagger. Disambiguation tags already have a serious problem with people removing them instead of replacing them, so when somebody takes the trouble to tag a character properly, even if it seems pointless, it increases the amount of information catalogued on the site and makes it a better place.

cimatrie said:
Not going to comment on "confidence" or "seasoned taggers" because it doesn't remove any of the subjectivity from the decision making process, for me or anyone else. Is there a post score that lets you into the club? I am not interested in bothering power users on the forum for hot tips unless they want to post here.

That tag above meets TWYS of course, so obviously it is okay until someone finds a better argument. The gauntlets_only tag is more contentious, but I'm not arguing I should make the decision to remove it. I'm asking if there are any good policy-driven clear-cut cases where I shouldn't ignore it, and so far I've seen none. Again, I am okay with that.

Mauve is synonymous to Violet, Lavender and Purple. Mauve is more whitish than Purple, but less whitish than Lavender. If the staff aliased Cyan_* to Blue_* for being too similar, despite being a series of tags more worthy to be used in order to distinguish teal characters from blue characters, then what makes you think Mauve has any chance to survive? Most likely scenario is it gets aliased to purple.

Mauve is exactly like Burlywood in the sense that both tags are way too extremely specific. Burlywood even had a color code in its wiki page which is the only color tag with this distinction.

cimatrie said:
That tag above meets TWYS of course, so obviously it is okay until someone finds a better argument. The gauntlets_only tag is more contentious, but I'm not arguing I should make the decision to remove it. I'm asking if there are any good policy-driven clear-cut cases where I shouldn't ignore it, and so far I've seen none. Again, I am okay with that.

I can easily see gauntlets_only as being a potential subtag for gloves_only, since we also have elbow_gloves_only & fingerless_gloves_only implied to it.

I can give you one policy-driven case where you shouldn't ignore it. That is if it is clearly spam. A group of jumbled up words or even harassing speech being added to posts.
In that case, you should immediately report the user in question and revert any damage that have been done.

That being said, there is no fixed rule, policy, or process when it comes to the introduction of new tags. Everything you learn about tagging comes with experience and word-of-mouth from other users.
If it helps with searching or blacklisting, then it would probably be kept.
If another tag already exists that can be used to replace the new tag, then it would be replaced or aliased away.
If it is too common or specific, then they would be invalidated (e.g., alias white_sclera --> invalid_tag). However, invalidated tags aren't always set it stone (e.g., topic #33035).

The validity of tags are routinely brought into question here in the forums when it is mentioned in threads or alias/implication suggestions, where the community would begin debating on whether it should be kept or invalidated.

Updated

If I see a single-digit-count tag and can't find a popular tag that replaces it, then I leave it because in future there may be more images with that tag, or people looking at older images may add it to them. "Deep tagging" theory I'd guess you call this?

I used light_* and blue_* today because light_blue_* is too generic, much like the hybrid tags that got aliased away. The color was the blue version of pink, as in it having a very high level of white in the color balance. Since pink seems popular enough, no reason to do this for 'light_red_*'.

I frown at people deleting applicable tags just because low count. Especially if it's a new character or something and of COURSE there's mostly the artist or an uploader doing the (new) tags. I also hate seeing redundant new tags that aren't aliased to same-meaning popular tags. Can't win every time! XD

Normally, the forums or image comments are where you ask about rather a tag makes sense... For example, massively renaming a large number of tags, or asking for a definition not in the Wiki and not obvious from examples... Or asking for advice on adding a Wiki entry based on the examples. There you'd want to be on the forums. If a it's a single image and kind of edge case/gray area, posting it on the image would get the attention of future taggers who may be more sure or able to see it better. If it's a tag that you're not sure if it exists, well, there's a stickied topic for that. ;) We have votes for alias/implication/BUR posts for a reason, also. Your reputation is incentive to not do crazy votes? Heh...

TheGreatWolfgang already covered a lot of this in last post.

  • 1