Topic: Tag alias: cat_robot -> invalid_tag

Posted under Tag Alias and Implication Suggestions

listlesssky

Member

Searching "cat" and "robot" brings up a ton of posts that have both a cat and a robot, but no cat robot. Why remove this?

regsmutt

Member

listlesssky said:
Searching "cat" and "robot" brings up a ton of posts that have both a cat and a robot, but no cat robot. Why remove this?

Making a *_robot tag for every species would be a lot of tags.

listlesssky

Member

Nobody would be forced to make one for literally every species. There's clearly interest here though, and it's a somewhat common trope in furry art. Nuking tags simply for the sake of having less tags is silly.

Donovan DMC

Former Staff

listlesssky said:
Nobody would be forced to make one for literally every species. There's clearly interest here though, and it's a somewhat common trope in furry art. Nuking tags simply for the sake of having less tags is silly.

That's not the point though, the point is that validating one paves way to all combinations being valid, which explodes into thousands of theoretically valid tags

listlesssky

Member

donovan_dmc said:
That's not the point though, the point is that validating one paves way to all combinations being valid, which explodes into thousands of theoretically valid tags

There are already infinite theoretically valid tags. Every single character already gets their own tag, and tons of them are random OCs with exactly one entry. If someone loves [X]_robots and wants to go tag them all, what's the harm?

And if it has to be aliased, why not simply alias to robot_anthro instead? It makes no sense to alias it to 'robot' when a more specific catchall tag already exists.

Donovan DMC

Former Staff

listlesssky said:
There are already infinite theoretically valid tags. Every single character already gets their own tag, and tons of them are random OCs with exactly one entry. If someone loves [X]_robots and wants to go tag them all, what's the harm?

Those infinitely theoretically valid tags do not have implications legitimizing them, there is a stark difference between a tag just existing in a void and a tag existing within an implication tree

listlesssky

Member

munchmallow-frosty said:
post #5466433

Tbf this one isn't even tagged cat_robot either

donovan_dmc said:
Those infinitely theoretically valid tags do not have implications legitimizing them, there is a stark difference between a tag just existing in a void and a tag existing within an implication tree

Why are we prioritizing worrying about a hypothetical snowball effect *_robot tag singularity instead of prioritizing the ability of users to find what they're looking for? There's 5.5k "cat robot" posts, most of which are just a robot and a cat in the same image. Digging through those to find the 51 currently tagged cat_robots is ridiculous.

Ruppari

Privileged

listlesssky said:
Tbf this one isn't even tagged cat_robot either

It's not, it's an visual example of why you cannot alias cat robot to robot anthro. Cat robots are not inherently anthros.

Why are we prioritizing worrying about a hypothetical snowball effect *_robot tag singularity instead of prioritizing the ability of users to find what they're looking for? There's 5.5k "cat robot" posts, most of which are just a robot and a cat in the same image. Digging through those to find the 51 currently tagged cat_robots is ridiculous.

This is literally not how this work. If we rule that cat robot is allowed tag
1. it means that we start a tagging project applying it to every single cat robot image
2. it means that we now also need to start working on establishing and populating *species*_robot tags for all other species, because it would be weird to have one singular exception for no reason at all.

The snowball effect is not hypothetical. It's inherently baked into how tagging is handled here. You cannot make up random exceptions just to suit your own personal wants.

Emionix

Member

Just alias it to robot. Cat_robot isn't needed. Felid robot and cat robot both return plenty of catlike robot characters. If they need to meet a certain individual's standard of roboticism or catness then that's what sets are supposed to be used for.

Updated

listlesssky

Member

ruppari said:
The snowball effect is not hypothetical.

If this was true then 95% of tags on this site need to get nuked. There are plenty of them with thousands of entries that are still only used on a fraction the images they're applicable to, hell the aforementioned robot_anthro is one of them. It's inevitable in a dataset of literally 5 million images. Completeness is not a requirement for a tag to exist, the only ones that even get close to it are gender tags.

I also keep seeing the line of logic "Instead of X_Y, just use X and Y" to justify aliasing or invalidation of many different tags without consideration that "X and Y" pulls up thousands of images with both X and Y but no X_Y. Nobody is asking anyone to personally participate in a campaign for something they don't care about, just to not actively tear down other tagging efforts. It's very demotivating.

Lafcadio

Janitor

listlesssky said:
If this was true then 95% of tags on this site need to get nuked. There are plenty of them with thousands of entries that are still only used on a fraction the images they're applicable to, hell the aforementioned robot_anthro is one of them. It's inevitable in a dataset of literally 5 million images. Completeness is not a requirement for a tag to exist, the only ones that even get close to it are gender tags.

I also keep seeing the line of logic "Instead of X_Y, just use X and Y" to justify aliasing or invalidation of many different tags without consideration that "X and Y" pulls up thousands of images with both X and Y but no X_Y. Nobody is asking anyone to personally participate in a campaign for something they don't care about, just to not actively tear down other tagging efforts. It's very demotivating.

The inability to search for a character that satisfies X and Y, instead of a post that satisfies X and Y, is a problem with the way tags are stored.

If somebody wanted to search for Bowser with yellow_hair, we would not encourage tagging yellow_hair_bowser, but a tagging schema that allows a post to have a separate tag list for each character would eventually allow this to be searchable even if there wasn't a specific tag for it.

If 20,000 separate people requested some X+Y combination I would still vote negatively each and every time because the actual problem is not being addressed, they're just requesting 20,000 different band-aid fixes.

listlesssky

Member

lafcadio said:
20,000 different band-aid fixes.

As lovely as a schema change would be, it doesn't seem possible. There are far more minor changes like revalidations that have overwhelming support often even from admins, but still never get accepted because effort is limited and there are higher priorities. Migrating to an entirely new tagging schema is orders of magnitude more involved than that, it'd require going back through every post on the site to properly split tags based on characters. Throwing out useful tags in favor of hoping for a massive change that will likely never come is silly.

And there are useful X_Y tags, some have hundreds of thousands of applications that nobody would argue for invaldating. Hell there are even X_Y_Z tags such as anthro_penetrating_[anthro | gender] for which every subcombination exists, some of which have 150k+ uses. Whether a combo tag lives or dies right now feels totally arbitrary.

YummyTummySketties

Member

listlesssky said:
And there are useful X_Y tags, some have hundreds of thousands of applications that nobody would argue for invaldating. Hell there are even X_Y_Z tags such as anthro_penetrating_[anthro | gender] for which every subcombination exists, some of which have 150k+ uses. Whether a combo tag lives or dies right now feels totally arbitrary.

[form]_penetrating_[gender] should be gone honestly.

Lafcadio

Janitor

listlesssky said:
As lovely as a schema change would be, it doesn't seem possible. There are far more minor changes like revalidations that have overwhelming support often even from admins, but still never get accepted because effort is limited and there are higher priorities. Migrating to an entirely new tagging schema is orders of magnitude more involved than that, it'd require going back through every post on the site to properly split tags based on characters. Throwing out useful tags in favor of hoping for a massive change that will likely never come is silly.

And there are useful X_Y tags, some have hundreds of thousands of applications that nobody would argue for invaldating. Hell there are even X_Y_Z tags such as anthro_penetrating_[anthro | gender] for which every subcombination exists, some of which have 150k+ uses. Whether a combo tag lives or dies right now feels totally arbitrary.

Tagging Renamon's double tomoe correctly is also a massive change that a lot of the site's users will not contribute to (if you know who Renamon is, but not what a tomoe is, that includes you.), but we wouldn't care to not establish the tags just because it's a lot of effort.

You are grossly underestimating how people actually use these X+Y tags. Different people will give the same female_(character) posts cutesy nicknames, devine_slut [sic] was the combination of submissive/slutty and legendary_pokemon, things like that.

Giving these X+Y tags the maintenance they actually deserve would still require an absolutely gargantuan amount of aliases and implications. More labor is still being done but it's by admins who are being distracted from other kinds of A/I/BURs, not by a developer who wants to work on an interesting project. It would be infinitely less effort to reinvent how tag lists are stored if it prevents arbitrary inventions like giantess_dragon_submissive_to_human_knight, and it would still facilitate searches where <macro female dragon submissive> and <human knight> exist in the same post, but as two separate character tag lists.

Donovan DMC

Former Staff

listlesssky said:
Hell there are even X_Y_Z tags such as anthro_penetrating_[anthro | gender] for which every subcombination exists, some of which have 150k+ uses. Whether a combo tag lives or dies right now feels totally arbitrary.

The mixed form & gender tags should absolutely be tossed out
It isn't really arbitrary, there have been discussions but no one has actually made a BUR to start the process of going either way with tags, admins have many better things to do than micromanage tags so if people are getting processes started, nothing will be done with/about a tag

listlesssky

Member

lafcadio said:
Tagging Renamon's double tomoe correctly is also a massive change that a lot of the site's users will not contribute to (if you know who Renamon is, but not what a tomoe is, that includes you.), but we wouldn't care to not establish the tags just because it's a lot of effort.

Isn't the entire premise of aliasing it to robot "we don't want to include these species tags because it's too much effort"? It certainly seemed to be ruppari's case:

ruppari said:
If we rule that cat robot is allowed tag

1. it means that we start a tagging project applying it to every single cat robot image
2. it means that we now also need to start working on establishing and populating *species*_robot tags for all other species, because it would be weird to have one singular exception for no reason at all.

The snowball effect is not hypothetical.

A hypothetical distant future schema change isn't a solid enough reason to invalidate these tags in the present. And in the present, searching "cat" "robot" returns thousands of non-cat_robots. But searching double_tamoe brings up only seven non-renamons. Surely if the former is redundant the second is too?

It just feels like people are far too nonchalant about nuking tags, especially ones with huge numbers of applications. It's extremely demotivating to see peoples effort get erased and the site actively become less usable in the name of potential future changes that aren't even remotely guaranteed to happen.

YummyTummySketties

Member

listlesssky said:
It just feels like people are far too nonchalant about nuking tags, especially ones with huge numbers of applications. It's extremely demotivating to see peoples effort get erased and the site actively become less usable in the name of potential future changes that aren't even remotely guaranteed to happen.

If you're interested enough, you can use a set to put posts with a particular tag before it gets aliased away. That way, I guess efforts won't be completely erased.

listlesssky

Member

yummytummysketties said:
If you're interested enough, you can use a set to put posts with a particular tag before it gets aliased away. That way, I guess efforts won't be completely erased.

Sets are unhelpful here. I don't know a way to add posts to a set in bulk, and manually adding thousands or hundreds of thousands of posts to one is not practical. They also do almost nothing for anyone except the set creator, because there's basically no good way to find a set if you don't already know its name or author. Many invalidated/aliased tags did not have a set made anyway, and now it's too late.

Honestly, if any invalidation/aliasing/bulk updating action automatically created a set of the modified posts as a backup that would remove many pain points.

YummyTummySketties

Member

Obviously sets are not as useful as actual tags. But that's the best I can think of right now.

listlesssky said:
I don't know a way to add posts to a set in bulk, and manually adding thousands or hundreds of thousands of posts to one is not practical.

You can do it by changing "Mode"
Choose "Add to Set" and you can add posts by clicking/tapping post thumbnails.
But you're right. If there are like 1000 posts or more, it's tiring.
I do know one way to mass-add posts semi-automatically, but only 1% of the site can do it so it might not be that helpful.

listlesssky said:
They also do almost nothing for anyone except the set creator, because there's basically no good way to find a set if you don't already know its name or author. Many invalidated/aliased tags did not have a set made anyway, and now it's too late.

We can start creating sets in the future. Still better than never. If you want a set for tags that are going to be aliased away, you can make it and you can also invite people as "maintainers" so they can add posts too. As for searchability, uhhh I'm not sure. If the creator can mention it in the related AIBUR forums like this, that seems to make it slightly easier to find.

listlesssky

Member

yummytummysketties said:
If you want a set for tags that are going to be aliased away, you can make it and you can also invite people as "maintainers" so they can add posts too. As for searchability, uhhh I'm not sure.

It's not practical to manually add thousands to hundreds of thousands of posts to a set, especially when essentially nobody else is going to be able to find that set. For it to be practical we either need access to bulk tools, or for the process to simply be automatic. We also need easier ways to find sets, such as showing the sets an image is in on the post page itself somewhere.

YummyTummySketties

Member

listlesssky said:
It's not practical to manually add thousands to hundreds of thousands of posts to a set, especially when essentially nobody else is going to be able to find that set. For it to be practical we either need access to bulk tools, or for the process to simply be automatic. We also need easier ways to find sets, such as showing the sets an image is in on the post page itself somewhere.

Yeah sure. I agree.
You should post a feature request if you have some good ideas to improve sets.

Ruppari

Privileged

listlesssky said:
such as showing the sets an image is in on the post page itself somewhere.

This is already a thing. It's the "Sets with this post" button.

Donovan DMC

Former Staff

ruppari said:
This is already a thing. It's the "Sets with this post" button.

And this is almost certainly the best you're gonna get, there won't be a list on posts always visible like pools because sets aren't pools, they explicitly serve different purposes

Lafcadio

Janitor

listlesssky said:
Isn't the entire premise of aliasing it to robot "we don't want to include these species tags because it's too much effort"? It certainly seemed to be ruppari's case:

A hypothetical distant future schema change isn't a solid enough reason to invalidate these tags in the present. And in the present, searching "cat" "robot" returns thousands of non-cat_robots. But searching double_tamoe brings up only seven non-renamons. Surely if the former is redundant the second is too?

It just feels like people are far too nonchalant about nuking tags, especially ones with huge numbers of applications. It's extremely demotivating to see peoples effort get erased and the site actively become less usable in the name of potential future changes that aren't even remotely guaranteed to happen.

cat robot will get you cat robots sometimes. Reinventing tag storage so that cat and robot can be on the same character will solve every possible X_Y tag for the rest of time without more and more implications and aliases. There is no world where we do any weird shit to double_tomoe because "oh, almost every double_tomoe is renamon!" Okay, and what about renamon -double_tomoe?

Please engage with the rest of my post instead of just acting like you've made up your mind and everybody else is wrong.

lafcadio said:
You are grossly underestimating how people actually use these X+Y tags. Different people will give the same female_(character) posts cutesy nicknames, devine_slut [sic] was the combination of submissive/slutty and legendary_pokemon, things like that.

Giving these X+Y tags the maintenance they actually deserve would still require an absolutely gargantuan amount of aliases and implications. More labor is still being done but it's by admins who are being distracted from other kinds of A/I/BURs, not by a developer who wants to work on an interesting project. It would be infinitely less effort to reinvent how tag lists are stored if it prevents arbitrary inventions like giantess_dragon_submissive_to_human_knight, and it would still facilitate searches where <macro female dragon submissive> and <human knight> exist in the same post, but as two separate character tag lists.

listlesssky

Member

lafcadio said:
cat robot will get you cat robots sometimes. Reinventing tag storage so that cat and robot can be on the same character will solve every possible X_Y tag for the rest of time without more and more implications and aliases. There is no world where we do any weird shit to double_tomoe because "oh, almost every double_tomoe is renamon!" Okay, and what about renamon -double_tomoe?

Please engage with the rest of my post

My reply wandered a bit, but my main disagreement with the meat of your post was with curating tags in a way that harms short-term usability in expectation of a massive long-term change that as far as I can tell isn't even guaranteed to happen. And if it does happen, it will take a huge amount of time and effort.

People occasionally misusing X+Y character tags is unfortunate, but in the current system there are things only they can do. "X and Y brings up X+Y sometimes" is cold comfort when the number of X+Y posts is dwarfed by the number of "X and Y" posts by two or more orders of magnitude. A/I/BURing tags in such a way that makes those combos de facto unsearchable just because they hypothetically will be searchable in a more sane way in a future schema feels like putting the wagon before the horse, at least until it's truly confirmed that the site is going to do that and work actually begins.

Also if that future schema does get implemented, couldn't combo tags simplify some of the migration work? If a work has an X+Y tag, and it's a solo work, then that character can be automatically given X and Y. If it's not a solo work but only one of the characters has X then that character must also have Y (and vice versa). A script or tool implementing this logic could take a solid bite out of the manual work required. Also, X+Y character tags could be filtered for in order to easily identify many posts that have not yet been updated to the new system (since they would be truly nonexistent after the migration).

Kyiiel

Member

[robot_anthro + cat]
[robot_feral + cat]

These seem to give pretty darn good results. Haven't checked other body types.