Seemingly widespread ratings manipulation

[EmilyMiller]

It all makes me think that the all-time top list is obsolete because the site can't protect the stories on it. The site might be able to protect future highly-rated stories by taking the list down.

I think that this is the right approach. As is, the toplists serve no function as either a curated list of good stories for readers or a target for keen writers. As is, they have been reduced to gray goo.

 

[Areala-chan]

I gave up and just disabled voting on my stories until this mess gets sorted out.

 

[EmilyMiller]

I gave up and just disabled voting on my stories until this mess gets sorted out.

What a crappy outcome

 

[Trionyx]

The sad result of this is the inability to distinguish truly great works from the rest (like mine.)

 

[JGittes]

The sad result of this is the inability to distinguish truly great works from the rest (like mine.)

A few years ago, I read many of the top-list stories in the seven or eight categories I was thinking of writing in. What I observed was that while the writing was technically sound and clear, the storylines were very vanilla. I wouldn't call them great, but they're well written and popular in a lowest-common-denominator way.

In some categories, romance, for example, I didn't want surprises, and the top list was a good guide to begin an exploration of the category. In most of the other categories I looked into (I avoid LW), many of the stories weren't even especially erotic.

 

[Stanfield2025]

A few years ago, I read many of the top-list stories in the seven or eight categories I was thinking of writing in. What I observed was that while the writing was technically sound and clear, the storylines were very vanilla. I wouldn't call them great, but they're well written and popular in a lowest-common-denominator way.

In some categories, romance, for example, I didn't want surprises, and the top list was a good guide to begin an exploration of the category. In most of the other categories I looked into (I avoid LW), many of the stories weren't even especially erotic.

That's curious to say the least.

 

[TheLobster]

I wouldn't call them great, but they're well written and popular in a lowest-common-denominator way.

You don’t get on the bestseller list — as in, actual market-based list, rather than what NYT thinks should be a bestseller — through feats of literary experimentation. You get it by dishing out decently crafted stories that appeal to the masses and getting lucky.

In other words, less James Joyce and more Dan Brown.

 

[JGittes]

You don’t get on the bestseller list — as in, actual market-based list, rather than what NYT thinks should be a bestseller — through feats of literary experimentation. You get it by dishing out decently crafted stories that appeal to the masses and getting lucky.

In other words, less James Joyce and more Dan Brown.

I completely agree.

Bringing it back to the title of this thread, the NYT bestseller lists are manipulated in ways not entirely dissimilar to the Literotica lists. I've read accounts of EJ James paying people to go to the bookstores she knew the NYT used to determine its lists and buy all the copies of 50 Shades. I suspect there are various other tactics used to influence the lists, as well as 'sweep-like' countermeasures to lower the manipulation.

 

[Bramblethorn]

This "gap" of 1,485 votes cannot be explained simply by someone manipulating scores. If you look at other categories, you will see similar large gaps between stories (look at #6 & #7 or #13 & #14 in Lesbian Sex) with the same score. This only started happening when the flattening began, so there must be some correlation.

But it didn't.

Lesbian Sex toplist, March 1 2015. (I had to try several times to get through the 503 errors, Wayback has had problems with denial-of-service attacks and I guess it's happening again.) Here are the stories with a score of 4.83:




Story #4 has 4.83 from 4728 views; story #5 has 4.83 from 436. That's a bigger gap than the one you're pointing at, from more than 10 years ago.

Here are the 4.82s from that same list. Again, a huge gap in votes between #8 and #9.


And the 4.81s:




So whatever the reason is for having large gaps between top and second-top story at a particular score, it's not a new phenomenon.

IDK why the distribution is like that, but I can think of one possible mechanism that doesn't involve anything other than honest voting and what's publicly known about how the toplists are supposed to work; I'll put that in a separate post when I have time.

 

[TheExperimentalist]

Fun fact: Ostriches don't actually stick their heads in the sand.

This random animal fact has been brought to you by a weirdo

I'd bet that in your stories, they prefer to stick their heads somewhere else, amiright?

 

[Bramblethorn]

IDK why the distribution is like that, but I can think of one possible mechanism that doesn't involve anything other than honest voting and what's publicly known about how the toplists are supposed to work; I'll put that in a separate post when I have time.

So, I ended up running a simulated scenario to confirm that this kind of vote gap can arise naturally, just from the way voting and toplists work on Lit.

Basic assumptions:

• Every story has a "quality" which represents the average score it would get if infinitely many people voted on it.
• Stories with higher quality are less common: for every 3000 stories in the range 4.5-4.6 there are about 400 in the range 4.6-4.7, 50 in the range 4.7-4.8, 7 in the range 4.8-4.9, and just one over 4.9. (Exponential distribution, beginning at 4.5, scaling parameter 0.05, no values above 4.99 permitted; stories below 4.5 aren't modelled as they're not relevant to toplists.)
• Every day, some number of new stories are posted. (For this run I just set that number to 1, remembering that we're not simulating all stories, just the ones over 4.5.)
• New stories get 20 votes on their first day (exact number doesn't matter too much since I don't care about exactly simulating what happens in their early days; it's just easier if I don't have to deal with zero-vote stories.)
• Old stories get votes at a fairly low rate (set to 0.05 per day, on average).
• Stories that are currently in a toplist, as assessed by Lit criteria, get extra votes: extra 0.5 votes/day for #1 on the toplist, decreasing to an extra 0.01 for stories in positions 51-250.
• Every vote is either a 4 or 5, randomly determined according to story quality: e.g. if the quality is 4.8 then it has an 80% chance of getting a 5 and a 20% chance of getting a 4.
• Repeat for ten thousand days.

At the end of the run, the top-25 list looks like this. (IRL we never know "quality"; it's what the score is trying to measure. But I'm including it here because it's helpful for understanding what's going on.):

Position, Votes, Score, Quality
1, 4.97, 2718, 4.979
2, 4.96, 1287, 4.965
...
17, 4.84, 219, 4.820
18, 4.84, 200, 4.866
19, 4.83, 1527, 4.823
20, 4.83, 522, 4.817
21, 4.83, 391, 4.805
22, 4.82, 1062, 4.823
23, 4.82, 544, 4.818
24, 4.82, 403, 4.825
25, 4.82, 402, 4.790

In the 4.83 band, story #19 has 1527 votes, and #20 has just 522, a similar gap to the one @BobbyBrandt highlighted. There's also a bit of a gap in the 4.82 band, though not as drastic.

So, having confirmed that this kind of gap can happen without any kind of vote-rigging shenanigans...why is it happening?

TLDR: some stories are genuinely really good, and they stay in the toplist (or they did back before the systematic downvoting that triggered this thread), and so they keep on getting more and more votes.

Other stories aren't quite good enough to make it into the toplist just on their merits, but they get pushed over the line by luck. The lower the vote count, the more influence luck has on a story's score; for stories around the 4.80 range, with 500 votes, there's about 0.04 points of potential "noise" in their score at any given time. A 4.8 won't score very highly in the top list, but a 4.84 will, so that luck can have a big effect on their placement. Of course, higher placement means more votes, which will eventually pull those stories back down to their long-run average and out of the list.

So at any given time the top list is a mix of a few long-term high scorers, which will have very high vote counts because of their longevity, and relatively low-vote stories which are only there until their luck runs out.

Is that what was happening in the real toplists? I don't know. There could have been other things going on. But gaps of this kind aren't inherently suspicious.

 

[Areala-chan]

What a crappy outcome

Weirdly enough, I've found it liberating.

 

[EmilyMiller]

Weirdly enough, I've found it liberating.

Not having to live up to expectations?

 

[Areala-chan]

Not having to live up to expectations?

More like just not having to worry about which of my stories is going to get bombed next.

 

[SmilingLez]

So, I ended up running a simulated scenario to confirm that this kind of vote gap can arise naturally, just from the way voting and toplists work on Lit.

Basic assumptions:

• Every story has a "quality" which represents the average score it would get if infinitely many people voted on it.
• Stories with higher quality are less common: for every 3000 stories in the range 4.5-4.6 there are about 400 in the range 4.6-4.7, 50 in the range 4.7-4.8, 7 in the range 4.8-4.9, and just one over 4.9. (Exponential distribution, beginning at 4.5, scaling parameter 0.05, no values above 4.99 permitted; stories below 4.5 aren't modelled as they're not relevant to toplists.)
• Every day, some number of new stories are posted. (For this run I just set that number to 1, remembering that we're not simulating all stories, just the ones over 4.5.)
• New stories get 20 votes on their first day (exact number doesn't matter too much since I don't care about exactly simulating what happens in their early days; it's just easier if I don't have to deal with zero-vote stories.)
• Old stories get votes at a fairly low rate (set to 0.05 per day, on average).
• Stories that are currently in a toplist, as assessed by Lit criteria, get extra votes: extra 0.5 votes/day for #1 on the toplist, decreasing to an extra 0.01 for stories in positions 51-250.
• Every vote is either a 4 or 5, randomly determined according to story quality: e.g. if the quality is 4.8 then it has an 80% chance of getting a 5 and a 20% chance of getting a 4.
• Repeat for ten thousand days.

At the end of the run, the top-25 list looks like this. (IRL we never know "quality"; it's what the score is trying to measure. But I'm including it here because it's helpful for understanding what's going on.):

Position, Votes, Score, Quality
1, 4.97, 2718, 4.979
2, 4.96, 1287, 4.965
...
17, 4.84, 219, 4.820
18, 4.84, 200, 4.866
19, 4.83, 1527, 4.823
20, 4.83, 522, 4.817
21, 4.83, 391, 4.805
22, 4.82, 1062, 4.823
23, 4.82, 544, 4.818
24, 4.82, 403, 4.825
25, 4.82, 402, 4.790

In the 4.83 band, story #19 has 1527 votes, and #20 has just 522, a similar gap to the one @BobbyBrandt highlighted. There's also a bit of a gap in the 4.82 band, though not as drastic.

So, having confirmed that this kind of gap can happen without any kind of vote-rigging shenanigans...why is it happening?

TLDR: some stories are genuinely really good, and they stay in the toplist (or they did back before the systematic downvoting that triggered this thread), and so they keep on getting more and more votes.

Other stories aren't quite good enough to make it into the toplist just on their merits, but they get pushed over the line by luck. The lower the vote count, the more influence luck has on a story's score; for stories around the 4.80 range, with 500 votes, there's about 0.04 points of potential "noise" in their score at any given time. A 4.8 won't score very highly in the top list, but a 4.84 will, so that luck can have a big effect on their placement. Of course, higher placement means more votes, which will eventually pull those stories back down to their long-run average and out of the list.

So at any given time the top list is a mix of a few long-term high scorers, which will have very high vote counts because of their longevity, and relatively low-vote stories which are only there until their luck runs out.

Is that what was happening in the real toplists? I don't know. There could have been other things going on. But gaps of this kind aren't inherently suspicious.

That is the problem with the current rating system Literotica uses. A story with over 20,000 views and 1,000 five star ratings (without any 1 bombs) has a lower placement than a story with 50 views and 50 five star ratings. Definitely easy to manipulate a poorly written story with 50 views to be on the top list. Won't last very long because people will read it and recognize the (lack of) quality and vote it down. The inverse of what happens when a good story gets one star ratings by troll bombers. The playbook for a malignant narcissist who craves the little red hot tag

 

[Bramblethorn]

That is the problem with the current rating system Literotica uses. A story with over 20,000 views and 1,000 five star ratings (without any 1 bombs) has a lower placement than a story with 50 views and 50 five star ratings. Definitely easy to manipulate a poorly written story with 50 views to be on the top list. Won't last very long because people will read it and recognize the (lack of) quality and vote it down. The inverse of what happens when a good story gets one star ratings by troll bombers. The playbook for a malignant narcissist who craves the little red hot tag

The all-time top lists usually require a minimum of 100 votes, which goes some way to mitigate this, but it's not a complete fix. At least that kind of situation usually self-corrects pretty quickly.

 

[SmilingLez]

The all-time top lists usually require a minimum of 100 votes, which goes some way to mitigate this, but it's not a complete fix. At least that kind of situation usually self-corrects pretty quickly.

Then we will adjust the numbers. A story with over 20,000 views and 1,000 five star ratings (without any 1 bombs) has a lower placement than a story with 100 views and 100 five star ratings. And I agree, correction comes swiftly

 

[SkyBubble]

the only realistic methods to guard against it would alienate a large portion of the readership.

What would that be?

 

[SkyBubble]

So, it hasn't started yet?

Sorry, couldn't resist, but you might want to edit the year in your post to something in the past.

Unless you're psychic.

 

[bi_cathy]

I’ll have some thoughts to share about this once I’m done publishing Strings Attached.
For now, I’m just screenshotting and enjoying the rise and fall. Celebrating the top spots each of the different chapters has been getting, grumbling at the pettiness, and using it all *for science*. It’s a fun experiment between math and observing the inevitability of human behavior.

 

[onehitwanda]

What would that be?

registration required to vote and a public voting record.

 

[EmilyMiller]

and using it all *for science*. It’s a fun experiment between math and observing the inevitability of human behavior.

I’d caution you using words like science and math here, that’s menfolks’ talk. Don’t you worry your purty little head about it…

 

[MelissaBaby]

Monday, after a sweep:



Today:





Surely, just a coincidence.

The same coincidence, over and over again, story after story...

 

[EmilyMiller]

Surely, just a coincidence.

Ten coincidences equals a pattern

 

[SmilingLez]

It all makes me think that the all-time top list is obsolete because the site can't protect the stories on it. The site might be able to protect future highly-rated stories by taking the list down.

100% agree same with the Poetry top list. My 2025 free verse entry for the Ode to Mickey Spillane, was the top of all time for several months with a 5.0 rating. When I heard about the one bomb trollers, I immediately took voting off