So, I've just learned this tag exists. Seems to me like that relies entirely too much on 100% subjective personal judgement, which goes against site tagging policies. I'd like to open a discussion on what we should do to not only get rid of it, but ensure it never gets tagged again.

Aliasing it to invalid_tag would effectively get rid of it. Tags like cute are aliased away for similar reasons.

watsit said:
Aliasing it to invalid_tag would effectively get rid of it. Tags like cute are aliased away for similar reasons.

Thing is, I've been informed that the powers that be are trying to steer away from aliasing things to invalid_tag, these days, and want to come up with alternative solutions. That's why I didn't just go ahead and submit an alias request like the one you're suggesting.

Way back when the relaunch was new and getting shat upon from all corners, one of the improvements we were promised was the creation of a new tag category, "invalid", to replace the many invalid_tag aliases and solve the age old problem of not having a clue which one of the tags was causing the problem.

You might have noticed that this has still not happened.

In the general sense that might be true, but I think there can still be cases for it. The alternative would be to disambiguate it, with a wiki page suggesting what the tagger may have meant. However, for something like "nasty", there isn't really a good set of alternatives... it's just something someone really doesn't find appealing. Cub is nasty to someone, scat is nasty to someone, veiny_penis is nasty to someone, size_difference is nasty to someone. In my opinion, I don't see any good set of suggestions for what it could mean, so just getting rid of it would be the cleanest option.

wat8548 said:
Way back when the relaunch was new and getting shat upon from all corners, one of the improvements we were promised was the creation of a new tag category, "invalid", to replace the many invalid_tag aliases and solve the age old problem of not having a clue which one of the tags was causing the problem.

You might have noticed that this has still not happened.

Yes it has. But recategorizing to the Invalid category has to be admin-approved, which is why there aren't many there yet (they have to take the time to consider it and make the change, on top of all the other running-the-site stuff they do).

watsit said:
In the general sense that might be true, but I think there can still be cases for it. The alternative would be to disambiguate it, with a wiki page suggesting what the tagger may have meant. However, for something like "nasty", there isn't really a good set of alternatives... it's just something someone really doesn't find appealing. Cub is nasty to someone, scat is nasty to someone, veiny_penis is nasty to someone, size_difference is nasty to someone. In my opinion, I don't see any good set of suggestions for what it could mean, so just getting rid of it would be the cleanest option.

Fair enough.

Also, I've just realized it's almost exclusively one user tagging it, too...

jacob said:
Thing is, I've been informed that the powers that be are trying to steer away from aliasing things to invalid_tag, these days, and want to come up with alternative solutions. That's why I didn't just go ahead and submit an alias request like the one you're suggesting.

These days they should be moved to the 'invalid' category (with BUR), instead of aliased to invalid_tag. But that's basically still invalidation.

watsit said:
Yes it has. But recategorizing to the Invalid category has to be admin-approved, which is why there aren't many there yet (they have to take the time to consider it and make the change, on top of all the other running-the-site stuff they do).

...So, it hasn't happened, then.

Admin time isn't an excuse, given it's a 100% automatable process. The list of current invalid_tag aliases hasn't gone anywhere.

wat8548 said:
...So, it hasn't happened, then.

Admin time isn't an excuse, given it's a 100% automatable process. The list of current invalid_tag aliases hasn't gone anywhere.

It has happened, the category exists with some tags in it already and some more recategorization requests pending. Just because the requests haven't been accepted as fast as you want doesn't mean the category isn't implemented and in use.

It has to be admin-approved so it doesn't get abused by random users throwing random tags into the Invalid category. Recategorization requests to Invalid have to be reviewed, so it can't be automated. Not sure what your point is about the invalid_tag aliases; like I said before, not all tags are worth bothering with, and are better off aliased to invalid_tag so it can be cleaned up expediently (with an automated bot). Ones that are worth de-aliasing and recategorizing, for disambiguation purposes or whatever, need to be suggested like any other tag change that needs admin approval.

watsit said:
It has to be admin-approved so it doesn't get abused by random users throwing random tags into the Invalid category. Recategorization requests to Invalid have to be reviewed, so it can't be automated. Not sure what your point is about the invalid_tag aliases; like I said before, not all tags are worth bothering with, and are better off aliased to invalid_tag so it can be cleaned up expediently (with an automated bot). Ones that are worth de-aliasing and recategorizing, for disambiguation purposes or whatever, need to be suggested like any other tag change that needs admin approval.

Why shouldn't every currently aliased invalid tag have its alias replaced with a category change? If you want to review the list of invalid tags or whatever, surely that's a job for some other time. Nothing would be lost by simply porting over the existing database from the old system to the new, strictly superior, one.

wat8548 said:
Why shouldn't every currently aliased invalid tag have its alias replaced with a category change?

Because various tags don't need to be bothered with. Is it really necessary to dealias 0, 1, 2, 3, 4, etc, so they each have to be cleaned up manually? Similar for tags like and, or misspellings of unnecessary tags, like anatomically_incorect? I would say no. Certainly there's some worth dealiasing, but not all of them should be, in my opinion.

wat8548 said:
Why shouldn't every currently aliased invalid tag have its alias replaced with a category change?

Because they need wiki pages first. And most of them lack those, or are outdated.
When someone tags 'invalid' category tags, they get directed to the wiki. Therefore, the wiki entries -- which should include the list of tags that should be used instead -- must be filled out before we can move tags there.

genjar said:
Because they need wiki pages first. And most of them lack those, or are outdated.
When someone tags 'invalid' category tags, they get directed to the wiki. Therefore, the wiki entries -- which should include the list of tags that should be used instead -- must be filled out before we can move tags there.

But... all of this is already a problem with the existing system. Except even worse, because instead of a blank wiki page, you don't even know which wiki page to look for in the first place. Is there any compelling reason not to upgrade to the new system just because it won't solve every existing problem at once, given that it won't make any current problems worse, will make the most major current problem (not knowing which tag was invalid) better, and is less effort than exhaustively going through the entire list and picking out select tags to upgrade, a job which can always be deferred to a more convenient time?

watsit said:
Because various tags don't need to be bothered with. Is it really necessary to dealias 0, 1, 2, 3, 4, etc, so they each have to be cleaned up manually? Similar for tags like and, or misspellings of unnecessary tags, like anatomically_incorect? I would say no. Certainly there's some worth dealiasing, but not all of them should be, in my opinion.

What do you mean by "cleaned up manually"? Is there no way to search for posts which contain any tags from a certain category? If not, then assuming that the "invalid" category is planned to be implemented on a large scale eventually (otherwise why was it added in the first place), are you implicitly accepting here that there will come a day when extensive, regular "manual cleanup" does indeed become necessary? That sounds like a problem with the site's code, not with the policy.

wat8548 said:
What do you mean by "cleaned up manually"? Is there no way to search for posts which contain any tags from a certain category?

There is (I think), but the point of the invalid category is that a tag can be looked at to discern what the intent was, and fix it. E.g. if a post has the sexual tag in the Invalid category, someone should figure out what was meant by it (that the post is explicit? there's sex going on? foreplay? a character's being extra seductive?) and add any appropriate tag in place of the invalid one. The invalid tag shouldn't be removed until that's done. Otherwise, if you're just clearing out invalid tags en masse without checking them and adding what it likely meant, there was no real point to letting the tag stay to begin with. The invalid_tag itself can be left as that dumping ground, where there's nothing to gain from checking such tags over (e.g. a tells you nothing useful about what the intended tag is that may be missing to fix), and an automated bot can remove that specifically from posts relatively quickly.

watsit said:
There is (I think), but the point of the invalid category is that a tag can be looked at to discern what the intent was, and fix it. E.g. if a post has the sexual tag in the Invalid category, someone should figure out what was meant by it (that the post is explicit? there's sex going on? foreplay? a character's being extra seductive?) and add any appropriate tag in place of the invalid one. The invalid tag shouldn't be removed until that's done. Otherwise, if you're just clearing out invalid tags en masse without checking them and adding what it likely meant, there was no real point to letting the tag stay to begin with. The invalid_tag itself can be left as that dumping ground, where there's nothing to gain from checking such tags over (e.g. a tells you nothing useful about what the intended tag is that may be missing to fix), and an automated bot can remove that specifically from posts relatively quickly.

As of this post, there are 21 active posts tagged invalid_tag - not an insurmountable obstacle by any means. Even assuming that sometimes a post will end up with tags like "the" from forgetting underscores and the like, I think you're assuming more bad faith than is necessary on the part of the site's posters. Invalid tags should be there to help people tag correctly first and foremost, not for the convenience of other people coming after them to clean up their mess (and if users rely on them for that purpose they'll probably end up getting banned, just as is currently the case for large-scale nonsensical tagging). The first step in that process is telling users which tag was invalid, instead of leaving them to scan the list and try to work out which one's missing from memory. Even if the tag wouldn't be of any help to a third party, seeing a word like sexual in the list should be enough for the person who entered the tags to spot that they mistyped e.g. sexual_competition. Or if it wasn't a mistake, they'll just know that they shouldn't use that tag again. This also applies to stuff like a, since whoever typed that in the first place must have known what they intended, or in the worst case they'll know that a cat walked across their keyboard while they weren't looking, and can promptly fix it without having to work out why they're getting the mysterious purple warning first.

Right now the only purpose invalid_tag serves is straight-up tag deletion with extra steps. This cannot, by definition, lead to an improvement in tagging quality.

wat8548 said:
As of this post, there are 21 active posts tagged invalid_tag - not an insurmountable obstacle by any means.

Sure, because it's regularly cleaned out with little issue. But like I said, other tags under the Invalid category are intended to be reviewed and fixed, not just deleted, so getting rid of those will be much slower. There are thousands of posts with some *_(disambiguation) tag, which are scheduled to go into the Invalid category; that will take a significant amount of time and effort to clean up.

wat8548 said:
Even assuming that sometimes a post will end up with tags like "the" from forgetting underscores and the like, I think you're assuming more bad faith than is necessary on the part of the site's posters. Invalid tags should be there to help people tag correctly first and foremost, not for the convenience of other people coming after them to clean up their mess (and if users rely on them for that purpose they'll probably end up getting banned, just as is currently the case for large-scale nonsensical tagging).

No bad faith is being assumed. If someone mistypes and tags a cat_is_fine_too, they'll get a red tag listed at the top of the list (either a or invalid_tag), and a tag with a low count (which if it has a count of 1, gets colored red; presuming cat_is_fine_too isn't aliased to a_cat_is_fine_too, correcting that part for them). Regardless of how a is handled, they'll know to fix it and can do so, though if it's left as a, that won't be a very useful indication of what was wrong; accidental space in a ss_up? shot a? or just an accidental key hit?). And if someone is in a hurry after posting something and doesn't notice the red tag or a low tag count, it gets left. a would end up getting left on such posts, which someone has to review before deleting, while invalid_tag can be quickly removed automatically.

wat8548 said:
Right now the only purpose invalid_tag serves is straight-up tag deletion with extra steps. This cannot, by definition, lead to an improvement in tagging quality.

That's why the current aliases need to be looked over, to find which ones will benefit from dealiasing and recategorizing, and leaving it alone for ones that would just create more work.

watsit said:
No bad faith is being assumed. If someone mistypes and tags a cat_is_fine_too, they'll get a red tag listed at the top of the list (either a or invalid_tag), and a tag with a low count (which if it has a count of 1, gets colored red; presuming cat_is_fine_too isn't aliased to a_cat_is_fine_too, correcting that part for them). Regardless of how a is handled, they'll know to fix it and can do so, though if it's left as a, that won't be a very useful indication of what was wrong; accidental space in a ss_up? shot a? or just an accidental key hit?). And if someone is in a hurry after posting something and doesn't notice the red tag or a low tag count, it gets left. a would end up getting left on such posts, which someone has to review before deleting, while invalid_tag can be quickly removed automatically.

And in this example, the red tag with a count of 1 is still getting left on such posts. Just because it isn't a common enough error to be aliased doesn't make it less of an invalid tag. What you appear to be arguing here is that there is an inherent greater value in a post being tagged with invalid_tag ss_up than there is in it being tagged a ss_up, despite both being equally invalid and the latter providing strictly more info. We do not and cannot have a system that catches every possible typo, and I'd argue that aliasing the common ones creates an illusion that such mistakes can be fixed without manual review. At the end of the day, someone's still gotta go in there to de-tag ss_up.

I think we're still talking past each other regarding the real purpose of invalid tags. I believe the onus is on people who add tags to make sure they're valid, and the site rules agree with me, since you can get banned for tagging badly enough. With that in mind, we should not create systems that make it harder (even slightly) to tell what you did wrong. Even if it was a false alarm and you accidentally hit the "1" key before submitting or whatever, seeing the invalid_tag warning means you've got to first remember everything you tried to tag and then check to make sure it's all still present, a time-consuming process. Whereas you seem to believe invalid_tag should be repurposed as some kind of automated cleanup aid, a task at which it can never fully succeed.

The only long-term solution to bad tagging is teaching users to tag better, and the concept of invalid_tag is directly at odds with that.

wat8548 said:
And in this example, the red tag with a count of 1 is still getting left on such posts. Just because it isn't a common enough error to be aliased doesn't make it less of an invalid tag. What you appear to be arguing here is that there is an inherent greater value in a post being tagged with invalid_tag ss_up than there is in it being tagged a ss_up, despite both being equally invalid and the latter providing strictly more info. We do not and cannot have a system that catches every possible typo, and I'd argue that aliasing the common ones creates an illusion that such mistakes can be fixed without manual review. At the end of the day, someone's still gotta go in there to de-tag ss_up.

That's kind of it, though. Having separate tags, a red a tag under Invalid and a low-count ss_up under General, it's only the second part that's meaningful. Someone can see the ss_up and logically assume they meant ass_up (especially if the picture contains a character with their ass up). The a itself doesn't help. And in other situations, where it's just an accidental key press for example, someone has to take time to work through what it could've meant, when it doesn't mean anything.

wat8548 said:
I think we're still talking past each other regarding the real purpose of invalid tags. I believe the onus is on people who add tags to make sure they're valid, and the site rules agree with me, since you can get banned for tagging badly enough. With that in mind, we should not create systems that make it harder (even slightly) to tell what you did wrong.

Sure, and I'm not disagreeing that invalid tags should be there to help tell you what you did wrong. I'm not saying that there aren't many tags currently aliased to invalid_tag that should be dealiased and recategorized. This side discussion came up because you claimed the Invalid category isn't implemented, despite the dozens of tags in the Invalid category already, because they didn't automatically dealias all tags from invalid_tag and recategorize them. And in that, I disagree that automatically doing that for all tags is a good idea, as it would create more unnecessary work in the long run.

watsit said:
That's kind of it, though. Having separate tags, a red a tag under Invalid and a low-count ss_up under General, it's only the second part that's meaningful. Someone can see the ss_up and logically assume they meant ass_up (especially if the picture contains a character with their ass up). The a itself doesn't help. And in other situations, where it's just an accidental key press for example, someone has to take time to work through what it could've meant, when it doesn't mean anything.

Maybe if you could come up with a better example your point would be more apparent, but I'm struggling to think of a case where I might see a picture tagged a and sincerely wonder what on earth the poster could have meant, absent any helpful clues like ss_up which would make it clear what had happened. It seems more likely that such tags would be treated (by those who did not add them) in much the same way as invalid_tag is now. An automated system, or a lazy human janitor, would remove either an invalid a tag or invalid_tag and leave ss_up present just the same. And it's not as if my proposed scenario never comes up under the existing system, given that we cannot possibly pre-emptively invalidate every possible typo.

Really, do we want to argue against giving tag cleaners the chance, however slim, of correctly implementing the original tagger's intent, on the basis that they would now have the choice of doing more work for better results? Not that I think it would change things for cleaners all that much either way, it would just make life easier for those who get tags wrong in the first place.

watsit said:
Sure, and I'm not disagreeing that invalid tags should be there to help tell you what you did wrong. I'm not saying that there aren't many tags currently aliased to invalid_tag that should be dealiased and recategorized. This side discussion came up because you claimed the Invalid category isn't implemented, despite the dozens of tags in the Invalid category already, because they didn't automatically dealias all tags from invalid_tag and recategorize them. And in that, I disagree that automatically doing that for all tags is a good idea, as it would create more unnecessary work in the long run.

There are dozens of tags! Dozens!

(Almost three whole dozens at the time of writing, of which nearly one dozen are just variants on *_background.)

The crux of my argument is that the practical impact on tag cleaners would be minimal in all cases, and the impact on uploaders would be minimal at worst (in the case of tags that nobody would add on purpose) and hugely positive in the case of tags which you do support de-aliasing. Given that, and given that selective de-aliasing and wiki page writing is itself a huge amount of work, an automated mass de-aliasing seems like a solution with fewer downsides than a manual selective one, especially since it doesn't stop the wiki pages from being written later and a blank wiki page with the name of your tag on it is still more helpful than the invalid_tag wiki page with the name of hundreds of tags in it.