[bulk] poorly_drawn

Type-kun said:

I'm thinking about building admin tool for mass-undo instead. It's a long-needed update anyway, since it can also be applicable to manual mass-mistagging and tag vandalism. Mandatory creator ID, start and end version ID, and optional field for added and removed tags theoretically should make it fast enough while allowing to fix virtually any mistake of any scale, not to mention it will work as a delayed job like alias/implication/BUR approval.

Either way, something like that should be built-in. Fixing what's described in this thread will take 2 hours at this rate, and it's relatively small-scale.

For the optional add/remove tags, a check also would need to be run for existing implications:

Example:

hummer -> military_vehicle (a mistake since not all hummers are military vehicles)

However, in addition to military_vehicle, it also adds the military tag. So if you're mass removing the military_vehicle tag with Range/CreatorID combination, it would also need to check to see if military was added as well and remove it if it was.

BrokenEagle98 said:

For the optional add/remove tags, a check also would need to be run for existing implications:

Example:

hummer -> military_vehicle (a mistake since not all hummers are military vehicles)

However, in addition to military_vehicle, it also adds the military tag. So if you're mass removing the military_vehicle tag with Range/CreatorID combination, it would also need to check to see if military was added as well and remove it if it was.

Not really; if it will undo the post version as a whole, military would be reverted as well. The only caveat I see is the merged versions. E.g. you added some incorrect tags, then within an hour went and added other tags to the post. The edits are merged, so if the version is reverted, it reverts both edits.

"Optional added/removed tags" I mentioned would be search parameters rather than mass-update scripts. That is, "undo only versions where (tag) was added". That said, perhaps it would be nice to include mass-update capabilities as well.

Ah sorry, I forgot you were doing reversions instead of mass tag editing.

Can you check if there is a later version of a post once you find the version you're looking for? That would let you see what tags were added/removed after the BUR/Implication/Alias went through.

BrokenEagle98 said:

Can you check if there is a later version of a post once you find the version you're looking for? That would let you see what tags were added/removed after the BUR/Implication/Alias went through.

Nope. If the same user that approved the BUR also edits the post further within an hour, the versions will be merged.

Well, it's not that much of a problem now, since not long ago the merging algorithm was changed - older version is deleted and new version containing all changes is inserted. So ID boundaries should work well with that; updated versions will have newer IDs and would stand out above the bulk, to be manually fixer later.

Also there's generally no need to see what happened after the version, because "undo" adds/removes tags that were removed/added in the post version, regardless of current post state.

I'll think it through and create a github issue later. We're going offtopic here :3

Yeah, no worries... coding/logical issues like this engage the problem solver in me... :)

reiyasona said:

Go through "status:active poorly_drawn -bad_anatomy -bad_proportions -bad_perspective" and flag all posts with bad overall quality. (Because "bad overall quality" should imply the meta tag status:deleted.)

No objections here, if it's truly poorly drawn it deserves to be flagged, not merely tagged.

• Purge the poorly_drawn tag. Alternative: Keep it as a "catch all" tag.

+1 to purging. poorly_drawn is a highly subjective tag, it only exists as a holdover from the early days. Its only use is in blacklisting but it's too subjective to have much real value there.

create implication bad_proportions -> bad_anatomy

Link to request

Reason: forum #119504

EDIT: The bulk update request #883 (forum #119634) has been rejected by @Type-kun.

forum #119504:

• Go through "status:active poorly_drawn" and flag all posts with bad overall quality because "bad overall quality" should imply the meta tag status:deleted. [✓]
• Go through "status:any poorly_drawn -bad_anatomy -bad_proportions -bad_perspective" and search for images that deserve the bad_anatomy, bad_proportions or bad_perspective tag. [✓]

Done!

evazion said:

• Purge the poorly_drawn tag. Alternative: Keep it as a "catch all" tag.

+1 to purging. poorly_drawn is a highly subjective tag, it only exists as a holdover from the early days. Its only use is in blacklisting but it's too subjective to have much real value there.

@Danbooru: What shall we do with the poorly_drawn tag? Keep or delete? (I vote for delete.)

Well, I don't know how user #11321 feels about this ... ;)

As for myself, +1 for deleting it...

I'm up for deletion, but still waiting for feedback.

Purge the tag, abhor the tag, shun the tag.

If it's really poorly drawn, that's a flagging reason for review.

Purged the poorly_drawn tag.

So..one implication is still active of this topic regarding the bad proportions -> bad anatomy implication request.
I wonder...what is with bad proportions due to bad perspective (i.e. a both arms are super long for example)

remove alias bad_art -> poorly_drawn

Link to request

No need for this anymore.

The bulk update request #946 has been approved.

poorly_drawn was populated again, somehow.

For the last pending implication on this thread (bad_proportions -> bad_anatomy), are we sure that's always the case? I can't think of counterexample right away, but I still have a weird feeling that wrong proportions don't always constitute bad anatomy. It's just a hunch though, so I might well be wrong about that. It'd be nice if someone sacrificed a bit of their SAN points and did a selective (or full?) check of posts under bad_proportions -bad_anatomy to see if all of them should actually be tagged as bad_anatomy.

Every use of the poorly drawn tag seems to be by @Fenen

Added a notice to the poorly drawn wiki to not use the tag.

Regarding bad_proportions -> bad_anatomy: I supposed one could argue they are distinct, but is such a distinction actually useful? Does anyone search for bad_proportions -bad_anatomy? Is anyone going to blacklist bad_proportions -bad_anatomy? I find it highly doubtful.

Bump!

Current scenario:

• anatomical_nonsense implies bad_anatomy
• bad_proportions is separate

Alternative scenario 1:

• anatomical_nonsense implies bad_anatomy
• bad_proportions implies bad_anatomy

Alternative scenario 2:

• anatomical_nonsense implies bad_proportions implies bad_anatomy

evazion said:

Added a notice to the poorly drawn wiki to not use the tag.

Regarding bad_proportions -> bad_anatomy: I supposed one could argue they are distinct, but is such a distinction actually useful? Does anyone search for bad_proportions -bad_anatomy? Is anyone going to blacklist bad_proportions -bad_anatomy? I find it highly doubtful.

Type-kun said:

For the last pending implication on this thread (bad_proportions -> bad_anatomy), are we sure that's always the case? I can't think of counterexample right away, but I still have a weird feeling that wrong proportions don't always constitute bad anatomy. It's just a hunch though, so I might well be wrong about that. It'd be nice if someone sacrificed a bit of their SAN points and did a selective (or full?) check of posts under bad_proportions -bad_anatomy to see if all of them should actually be tagged as bad_anatomy.

@Type-kun + @evazion

In my opinion, bad_proportions should imply bad_anatomy because bad_proportions is a more extreme (visible) form of bad_anatomy.

I looked though the first two pages of bad_proportions -bad_anatomy and got the impression that some of the posts are either not anatomically incorrect enough (in terms of proportions) to be tagged with bad_proportions (see post #2582927) or they are tagged with bad_proportions but in reality the problem is caused by bad_perspective (see post #2542675).

reiyasona said:

... or they are tagged with bad_proportions but in reality the problem is caused by bad_perspective (see post #2542675).

OK, then proportions->anatomy implication is a no-go for now, since we'll get those bad perspective posts tagged with "bad anatomy". Rejecting to clear up the backlog :3