Text on pictures add context and we often tag based on what is said or written, which is all well and good. But what about a story in the description? Should you tag stuff that's in the story but not the picture per se? It's still part of the post, after all.

No, because:
1. It'd effect searching and blacklisting in unintuitive ways.
2. Tagging literature should be an entirely different system than images in almost any context, our Tag What You See procedure in particular can't be directly translated to literature in a practical manner.
See topic #39539

only the general tags related to contents of text itself or to a post's general theme should take dialogue or text into account. so, stuff like impregnation_request can only be tagged based on dialogue, and stuff like forced can be tagged on textual or visual content.

however, most general tags are appearance-based and should only be tagged based on what visibly appears in the image part of a post. there's several reasons for this, most notably having this consistent across all posts and not needing to think about context or edge cases _{(a visually ambiguous male character being called "girl" as like a humiliation thing, for example)} makes both tagging and searching more simple.

also, just don't add general tags based on the context provided in a description. it's impossible to tell what tags would relate to the content of the image and description, so it'd just be a mess.

impawstor said:
Text on pictures add context and we often tag based on what is said or written

No you shouldn't. You can tag the text itself (e.g. profanity if it contains the word "fuck", or good_boy if it contains the literal words "good boy", irrespective of meaning or context), but not what the text means (e.g. a narrator saying a character is male should not be tagged male if the character doesn't look male). Dialog and narration is considered external information for tagging purposes.

impawstor said:
But what about a story in the description?

Should be considered even less. The tags describe what's visible in the image, and people will search or blacklist what they want to see or avoid. Someone who blacklists rape, for example, probably wouldn't appreciate completely innocent SFW images being blacklisted because a novel in the description contains a short rape sequence. Or someone looking for watersports wouldn't like vanilla sex getting into their search results because it's mentioned in a short description under the image. This would also cause a mess with ratings, as certain tags demand certain ratings (e.g. sex is always Explicit).

anicebee said:
2. Tagging literature should be an entirely different system than images in almost any context, our Tag What You See procedure in particular can't be directly translated to literature in a practical manner.
See topic #39539

I still don't think that .txt posts would necessarily require a separate tagging syetem (in theory), pretty much everything we tag right now could be applicable to a text story.

the main problem is more of a practical thing; monitoring tagging for consistency and ensuring quality would be monumentally more difficult and time consuming, so you'd probably need to limit the posting of .txt to only certain trusted users both the control the absolute flow as well as hopefully have some quality pre-filtering.

maryland_p_sevenson said:
I still don't think that .txt posts would necessarily require a separate tagging syetem (in theory), pretty much everything we tag right now could be applicable to a text story.

Yes, but

vulpes_artifex said:
[...] I don't think the level of detail we use for images is appropriate for stories. We can't tag a story table and book because it contains the sentence "He put his book on the table." [...]

watsit said:
No you shouldn't. You can tag the text itself (e.g. profanity if it contains the word "fuck", or good_boy if it contains the literal words "good boy", irrespective of meaning or context), but not what the text means (e.g. a narrator saying a character is male should not be tagged male if the character doesn't look male). Dialog and narration is considered external information for tagging purposes.

In practice, outside of tagging gender or the text blatantly contradicting the image or referring to something unseen, text does kinda matter. A lot of kinks are text-based: orgasm_denial, small_penis_humiliation, orientation_play, ...

crocogator said:
A lot of kinks are text-based: orgasm_denial, small_penis_humiliation, orientation_play, ...

They still need some visual element to work. Text is too open to subjective interpretation, particularly when dealing with translated text. Text can say whatever it wants, it doesn't need to adhere to what's meant to be depicted, so there still needs to be some visual indication of the thing being tagged that it's not purely based on the text.