Classification woes again!

My latest story submission has once again been put into Incest/Taboo - and while it would be disingenuous for me to say incest isn't a strong element in it, I don't think I'm actually writing for that genre per se!

Do you guys think it belongs there? I had wanted it in "Fetish", because it has many different kinks in it, and I don't know that any one of them predominates.

It's "Daughter, Lover, Superheroine".

Thanks in advance to any of you who take the time out.

Laurel decides these things; we don't.

That said, if it contains any incest, it probably belongs in incest. That's because incest is about the most polarizing category; people who want it really want it; people who don't tend to hate it. I know I'd be pretty annoyed if I read a story that sprang incest on me; I might not comment or rate but if there was a way to block an author such that I wouldn't see their stuff again, I'd use it. Laurel's probably protecting you from angry mobs and bad ratings.

The same logic applies to Gay Male and non-con. Readers don't want those things sprung on them; for some they completely destroy the buzz.

Lit needs a filter system. Don't want to see incest? Turn that off. It would be much easier to figure out where things go if the choices for genres were actually genres. Incest is not a genre any more than gay male is, and there's nothing wrong with trying to write a Romance story that includes both of those.

Incest is considered to trump most if not all other sorts of content. If it's there and readily found (it is, I looked) then you can expect it to be posted in I/T.

HandsInTheDark said:

Laurel decides these things; we don't.

Click to expand...

Wait, the authors don't get to decide where their stuff is posted? Or is just that Laurel gets veto power?

I assume that Lit has used the category system they have in place for many years and there is no plan to change it. And it's not like greatly increasing the number of categories would resolve some of these issues. But wouldn't it make more sense to then have tags on the top of the first page of the story rather than at the end where they are only seen after the person has already read, except in cases where a reader uses the search feature for stories with certain tags.

More categories woukd not help.

What would help are checkboxes. Authors would check off the elements a story contains. Got GM? Check. Incest? Check? First time? Check. And on through a long, tedious but very useful list.

Readers would color-check off what they did and did not want to see. Super-open-minded? Check nothing. Gay sex make you ill? Red check on that. Etc. Yay first time? Blue-green it.

Now everyone sees the literotica they want to see. Heck, think what it would do to LW. You don't like cuck stories? Why did you read one?

HandsInTheDark said:

What would help are checkboxes. Authors would check off the elements a story contains. Got GM? Check. Incest? Check? First time? Check. And on through a long, tedious but very useful list.

Click to expand...

Agree this. We can put in the tags when the story is loaded, but there is a limit of ten (maybe more from the new dash - does anybody know?) - and it's always a matter of wondering what tags to use. A checklist would be handy, we can only hope!

In this case, though, it looks like the OP wrote their own destiny by their disclaimer - as per RubenR's observation.

HandsInTheDark said:

Laurel decides these things; we don't.

That said, if it contains any incest, it probably belongs in incest. That's because incest is about the most polarizing category; people who want it really want it; people who don't tend to hate it. I know I'd be pretty annoyed if I read a story that sprang incest on me; I might not comment or rate but if there was a way to block an author such that I wouldn't see their stuff again, I'd use it. Laurel's probably protecting you from angry mobs and bad ratings.

The same logic applies to Gay Male and non-con. Readers don't want those things sprung on them; for some they completely destroy the buzz.

Click to expand...

This! I will have nothing to do with incest. I won't write it, won't read it. If I was reading a story in fetish and incest came up in it, I would back out quickly and be leery of reading here.

HandsInTheDark said:

What would help are checkboxes.

Click to expand...

I'd love an efficient Amazon.com version . . . insofar as allowing users/readers to toggle features (or fetishes) up and down an itemized/hierarchical tree of interests.

LeandraNyx said:

Wait, the authors don't get to decide where their stuff is posted? Or is just that Laurel gets veto power?

Click to expand...

Veto power. I haven't had a story moved, but other people definitely have. This appears to be one of her benign dictatorship roles; she knows the crowd well enough to know when to move stories.

the_stoic said:

insofar as allowing users/readers to toggle features (or fetishes) up and down an itemized/hierarchical tree of interests.

Click to expand...

I can think of two reasons why it won't happen. First, the hierarchical feature tree is bound to be long, complex, and offensive. "Asshole! You only mentioned stockings! But only BLACK stockings matter and this story has RED stockings!" And the categorization is fractal; for every checkmark there's a finer subdivision. GM: gentle and affectionate, or rough? If rough, with or without tearing? Etc. There's no bottom to this stack of turtles.

Second, it requires a new database table, to handle the intersection of user preferences and author offerings - a somewhat complex one, and once you design it, Lit has HOW many stories that would have to be exhaustively fitted to the table? It's fine for new stories, but the old ones become nightmarish retrofit problems. You could persuade current authors to go back and fill in the checks, but that's, what, maybe 2% of Lit's catalog?

I'd want a checkmark for "grammar matters to me", though. And one to filter out any vampire that sparkles.

I found this How To essay, Love your readers: categories very useful regarding categories, and their hierarchy/trump status, which often has to do with the squick factors of various taboo topics.

I agree if it contains any incest it probably should be placed in that category. I'm just stunned one individual is responsible for final classification of all material. That sounds exhausting.

HandsInTheDark said:

More categories woukd not help.

What would help are checkboxes. Authors would check off the elements a story contains. Got GM? Check. Incest? Check? First time? Check. And on through a long, tedious but very useful list.

Readers would color-check off what they did and did not want to see. Super-open-minded? Check nothing. Gay sex make you ill? Red check on that. Etc. Yay first time? Blue-green it.

Click to expand...

If I understand correctly, this is part of the plan for the eventual site upgrades. I've seen other sites that did it this way and it worked pretty well.

The title says it all.

HandsInTheDark said:

Second, it requires a new database table, to handle the intersection of user preferences and author offerings - a somewhat complex one, and once you design it, Lit has HOW many stories that would have to be exhaustively fitted to the table? It's fine for new stories, but the old ones become nightmarish retrofit problems. You could persuade current authors to go back and fill in the checks, but that's, what, maybe 2% of Lit's catalog?

Click to expand...

Theoretically, if you trained a machine learning algorithm on new content, it would be relatively easy to go back and use it to categorize old material. It would just take a while to train.

HandsInTheDark said:

Laurel's probably protecting you from angry mobs and bad ratings.

Click to expand...

Oh... I hadn't thought of it that way before.

Well, for this particular series I really do want to mix a bunch of themes together. I guess I'll refrain from adding on to Laurel's workload and just accept this then.

LeandraNyx said:

Theoretically, if you trained a machine learning algorithm on new content, it would be relatively easy to go back and use it to categorize old material. It would just take a while to train.

Click to expand...

OR you just flag unspecified content as unspecified content with a warning to the reader. Make it show up red when it's listed among other, specified content.

LeandraNyx said:

Theoretically, if you trained a machine learning algorithm on new content, it would be relatively easy to go back and use it to categorize old material. It would just take a while to train.

Click to expand...

Who are you and what do you know about AI?

You're right in theory. In practice I don't know how well that would work on erotica. These aren't scientific papers, using tight terminology. Stories here are vague and fluffy in both structure and terminology. And the audience is not forgiving; put one incest story in front of someone who's been real-life abused by their father; or one RAAC story in front of a BTB flamer, and fireworks will ensue. Unlike today, it won't ensue at an author (the site doesn't care about that), but at the site. I think it's a vaguely risky move.

With a different crowd, I'd crowdsource the problem - ask readers to vote on the applicable checkmarks for old stories. It would be faster and better than any MLA and it would cover the most popular stories the quickest. But, well, I think you're new here, so I'll just say in passing that the rating system has already been hijacked by a small group of people from a particularly deplorable basket of deplorables, and they register fake ratings and favorites to either promote their own stuff or engage in vendettas against others. There has even been rumor of an organized clique doing it in the past. Letting them categorize other peoples' stories - no. I can just imagine where the LW crowd with go with that power.

I'm probably the least likely to think an MLA would get this right; my stories are poor fits for the existing categories (lesse see - noncon and romance in a sciFi setting with some BDSM themes... um... sciFi it is.) I have problems believing an algorithm would get my Becoming Marie or Rent Comes Due right. I think for simple stuff - recognizing incest or interracial, or GM if all the names were recognizably male - it might be ok. But for the stuff I'm interested in - the line between Reluctance and Non-Con, or D/s, or Romance - I see problems.

But you've got me curious enough to want to try writing some algorithms and try it.

If AI, algorithms, or any kind of technology could be used to categorize stories accurately then sure, why not. But that isn't today at Lit. Still say it would be simpler to just put tags on top of first page of story so a reader can see before they read. Nothing wrong with readers having their own checklist of what is acceptable/non-acceptable though many wouldn't want to go to the trouble. It would still depend on writer accurately tagging with questionable themes/content to work right, which is how it should be. Let writer deal with negatives if they don't tag a story that contains gay, incestuous BDSM or non-con, nonhuman sci-fi. And for sure don't leave it up to readers to vote on what categories story should be. Would be opening big can of worms.
r5

HandsInTheDark said:

You're right in theory. In practice I don't know how well that would work on erotica. These aren't scientific papers, using tight terminology. Stories here are vague and fluffy in both structure and terminology.

Click to expand...

Yeah, it's definitely a risky move. We're not really at the point yet where AI can be effectively trained to identify or create creative work. Turns out training AI on work with a fuckton of variables is super hard. Who knew? (For reference, check out this sci-fi short film made using a neural network: https://www.theguardian.com/technol...unspring-silicon-valley-thomas-middleditch-ai). But I think if you used the tags on a story, rather than the actual content of the story, and weighed the tags according to their order (i.e. I think most authors list tags in the order they come to mind and thus, from most prominently featured to least prominently featured), the results wouldn't be too shabby. I'd use a markov chain over a neural network, though. Neural networks are really sexy right now, but aren't particularly useful yet.

LeandraNyx said:

Yeah, it's definitely a risky move. We're not really at the point yet where AI can be effectively trained to identify or create creative work. Turns out training AI on work with a fuckton of variables is super hard. Who knew? (For reference, check out this sci-fi short film made using a neural network: https://www.theguardian.com/technol...unspring-silicon-valley-thomas-middleditch-ai). But I think if you used the tags on a story, rather than the actual content of the story, and weighed the tags according to their order (i.e. I think most authors list tags in the order they come to mind and thus, from most prominently featured to least prominently featured), the results wouldn't be too shabby. I'd use a markov chain over a neural network, though. Neural networks are really sexy right now, but aren't particularly useful yet.

Click to expand...

I'd follow this up with "Marry me!" but 1) It sounds like a response a neural network would come up with and 2) I can think of at least one person who would object rather strongly.

Wait. You're a human female, right? You're not a function of weighted markov chains yourself, are you?


(The question is amusing now, but in twenty years, people you meet on line might well not be people. Oh brave new world, with such synthetic imponderables...)

HandsInTheDark said:

I can think of two reasons why it won't happen. First, the hierarchical feature tree is bound to be long, complex, and offensive. "Asshole! You only mentioned stockings! But only BLACK stockings matter and this story has RED stockings!" And the categorization is fractal; for every checkmark there's a finer subdivision. GM: gentle and affectionate, or rough? If rough, with or without tearing? Etc. There's no bottom to this stack of turtles.

Second, it requires a new database table, to handle the intersection of user preferences and author offerings - a somewhat complex one, and once you design it, Lit has HOW many stories that would have to be exhaustively fitted to the table? It's fine for new stories, but the old ones become nightmarish retrofit problems. You could persuade current authors to go back and fill in the checks, but that's, what, maybe 2% of Lit's catalog?

I'd want a checkmark for "grammar matters to me", though. And one to filter out any vampire that sparkles.

Click to expand...

Agreed, a hierarchy would twist and turn and might never end . . . but I think if that's the case, then the real problem isn't the breadth of content, or of the paralyzing fear of the "universal choice"; if this is the case, then I think the real problem is that the categories aren't (or were never) properly defined (or adapted) to suit the system. I liken readers parsing minute story variables to a discerning grocery shopper: if managers provide them with a map, articulate the quantity of materials available, and smartly incorporate an analysis of how users consume those materials into the store's design (analyzing aisles and layout, traffic flow, purchase frequency/patterns), then worrying about the outliers will become far less of a liability or a concern than acquiring data to sharpen or recalibrate the structure everybody relies on.

LeandraNyx said:

Yeah, it's definitely a risky move. We're not really at the point yet where AI can be effectively trained to identify or create creative work. Turns out training AI on work with a fuckton of variables is super hard. Who knew? ............ I'd use a markov chain over a neural network, though. Neural networks are really sexy right now, but aren't particularly useful yet.

Click to expand...

Chloe's brain, such as it is, overloads and explodes. Just build a spaceship that can take me to Mars, guys, okay.