Yes, the Toplists are Broken

[RejectReality]

I'm actually not sure how many times in a 24 hour period it updates? I'd believe any number between 2 and 6, but I think 3 or 4 is probably the right answer. Confounding is that I have things to do other than constantly refresh the toplists, and because the toplists are so manicured right now, an "update" can happen without anything obvious changing. You'd need hourly data for several days to be certain what the schedule was - and it might actually not be the same every day (for example: a 5-hour rotation would cause it to change at different hours on different days). But it is often different if I check in the morning and then again in the evening.

---

The final 1 is almost certainly the completion of a "pair."

As you say, there's a lot of ways to get to 4.8 at 106 votes from 4.91 at 100 votes. But since we're dealing with a trolling pattern, it would make the most sense for that to be following the pattern. And the pattern is that the troll hammer comes in pairs of 1* hits. A 1* rating followed by a second 1* rating a short time later. Since you have a snapshot at 106 votes and another snapshot at 107 votes, it would make the most sense if that last rating is something already generated by the troll vote pair having been initiated.

Dropping from 4.91 to 4.8 in 6 votes implies that your story picked up a total of 18 stars across 6 votes. That's three 5* ratings and three 1* ratings. But the 1* ratings come in pairs, so if you only got three of them, you'd expect another 1* rating to be incoming shortly after.

Getting 2 pairs of troll votes on a modest delay is actually entirely consistent with the process being automated. Especially if the pair completed after you were no longer even being displayed on the toplist. A human wouldn't bother chasing down a story that was no longer a threat, or even be able to find you, but an attack script would run to completion.

You picked the one combination that works with your theory.

Why would a middle chapter pick up three fives when nothing else in the series was voted on? It's not something special. It doesn't have one of the rare, full sex scenes of the series. This chapter is the very definition of a middle chapter. Found this cool thing on the toplist, voted five on it and decided not to read/vote on any more of them? Once? Maybe. Twice? I suppose. Three times? Doesn't compute.

18 stars is also six threes. That's the kind of votes a seasoned troll from a category with fans of stories a decade old with relatively low vote totals would cast. Then the oddball 1 hangs out there off-script.

There's no reason for a script to be dropping any fives as part of its calculation aiming for 4.84 either. And if there was, it failed spectacularly.

It's just toplist trolling.

 

[RocketGrunt]

I just imagine they keep getting hit, until the refresh.

Well, the key information is that the second pair of 1* attacks completed apparent AFTER the refresh.

While the trollhammer system is sophisticated enough to evade the site's sweeps for inauthentic votes, it's not sophisticated in any other way. Pairs of 1* reviews may be the maximum they've figured out how to get away with, but they aren't actually required on the low-vote total stories that come into the attack protocol for crossing the threshold of getting the minimum number of votes. Like, sure if there's a story with thousands of votes that bubbles back out of the 4.84 crush because of the weight of many many 5* ratings, it might take a few 1* hits to knock it down again. But the 100 and 200 vote stories usually don't even need a single 1* bomb. At 100 votes, a single 3* rating drops a 4.85 to a 4.83. These double taps are often HUGE overkill.

This is one of the reasons I think it's very likely to be automated. It's already kind of hard to imagine a real human being logging into 86 different chapters of Three Square Meals multiple times each so that they can drop enough 1* hits to drop them all from 4.85 or more to 4.84 or less. But also there was just genuinely no need to send two double-taps against One Whore's Town, Chapter 03. It entered with only 100 votes, a single double tap would drop it from 4.91 to 4.83. But the fact that the 4.91 story with a hundred votes ate two double taps and the 4.86 stories described in the original post only seem to have gotten hit with a single double tap each indicates that there's a rule being followed.

The constant use of excessive force is evidence that the actions aren't being considered by a human, and the swiftness and uniformity of the troll hammers is evidence that they are being automated. The fact that there are different rules being used on different boards is consistent with it being one person and also consistent with it being a group of different people. But the fact that the hammer falls exactly the same whether it's on the Gay Male board where the hammers only target the toplist or the Taboo/Incest board where the hammers target anything with a Red H indicates that it's the same system being used whether or not it obeys different masters.

 

[Cacatua_Galerita]

As further evidence of the Toplists brokenness. Here’s the progression of a highly rated story since before the times of the “4.84” regression…

Slingshot was published into Sci-Fi & Fantasy on January 22nd, 2024. From my (limited) data points, it appeared on the Sci-Fi Top List with a rating of 4.94 from 1651 votes.

Entering the Top List (and thus “exposing it to a wider audience”) did not condemn it. It gained at least a further 613 votes while retaining its 4.94 rating. Only in late 2024 did its ratings take a downturn.

The subsequent 269 votes were an average 4.66 stars, bringing it down to 4.91 overall.

The next 524 votes collectively rated the story at 4.50 stars. Meaning, from November 2024 to present, this ‘top of the top list’ story has been rated as ‘below average’ (for a Sci-Fi story).

Here are the data points.

Date Rating (Votes) Delta As per:
04/05/24 4.94 (1651) Debut at #8 Link
05/11/24 4.94 (1817) Up 7 to #1 Link
07/16/24 4.94 (2075) Dn 2 to #3 Link
09/09/24 4.94 (2264) Up 2 to #1 Link
11/20/24 4.91 (2533) Stdy at #1 Link
04/19/25 4.88 (2800) Dn 6 to #7 Link
06/13/25 4.88 (2852) Dn 2 to #9 Link
07/30/25 4.86 (2912) Dn 4 to #13 Link
02/13/26 4.84 (3057) Dn 111 to #124 Link

Conversely, there have been winners.

Three Square Meals, a 145-chapter series published between 2015 and 2023, had zero chapters listed in the Sci-Fi Top 50 on April 5th, 2024. By April 19th, 2025 it had four chapters in the Top 50. Today, 36 of the Top 50 stories are various chapters of Three Square Meals. No further chapters have been added to Literotica since 2023 (although, many continuation chapters are now available on Patreon, as pay-per-view).

Obviously, this is not proof that Three-Square-Meals has a group of fan-boys that are manipulating the Top List – some chapters of that story have regressed to 4.84 as well - but it shows that the Lists are becoming less diverse, thus, less useful overall.

I have read Slingshot and enjoyed it. This is the type of story Literotica should be placing front-and-center for their Sci-Fi readers. So, for those that argue that ‘fuck-the-authors’ is the site motto because readers are the only thing that counts... Why does this site let compelling content, like Slingshot, get spat out the arse-end of the Top Lists and into obscurity, only to be replaced by more chapters of the same old story? This is a reader/consumer issue for the site to address, not just an author gripe.

Today, as a reader, if you want to find a good story, I suggest you go to some archived Top List, because the live-site Top Lists are a hot mess.

Last thing to consider. Everything is being manipulated these days. Headlines, share markets, Amazon product ratings, political polls, market indexes, electricity spot prices, inflation rates – everything that can be squeezed for profit or influence, is being squeezed. I even wrote a story themed around dark money influence. So when I, or others like EmilyMiller, suggest improvements to the voting system, we’re not just trying to protect ‘our precious’ – we’re trying to protect every author’s stories from the same enshittification we see happening everywhere else. We’re not trying to gain personal advantage, we’re trying to strengthen the site against (inevitable) manipulation.

 

[RocketGrunt]

You picked the one combination that works with your theory.

Why would a middle chapter pick up three fives when nothing else in the series was voted on? It's not something special. It doesn't have one of the rare, full sex scenes of the series. This chapter is the very definition of a middle chapter. Found this cool thing on the toplist, voted five on it and decided not to read/vote on any more of them? Once? Maybe. Twice? I suppose. Three times? Doesn't compute.

18 stars is also six threes. That's the kind of votes a seasoned troll from a category with fans of stories a decade old with relatively low vote totals would cast. Then the oddball 1 hangs out there off-script.

There's no reason for a script to be dropping any fives as part of its calculation aiming for 4.84 either. And if there was, it failed spectacularly.

It's just toplist trolling.

A majority of the votes cast on Literotica are 5*. If people think something is good, they give it a 5*.

Your story, prior to being exposed to toplist trolling, received 5* ratings as more than 90% of all ratings. If it was viewed and rated by an extra group of people (for example: because it was briefly displayed on the top of the SF/F toplist), you would expect to receive extra votes that were also more than 90% 5* ratings. People who read your story think it's good, and people on Literotica don't need to think a story is amaze-balls terrific good to rate it 5*.

So if you got an extra block of votes and you wondered what the ratings given were, the most likely is just whatever implies the largest number of literal 5* ratings. Because we know that 5* ratings are more than 90% of the organic ratings that story receives.

The fact that this assumption paints a picture that fits EXACTLY with the votes over time graph of the previously studied trolling attacks shown earlier is something I find persuasive.

 

[EmilyMiller]

So when I, or others like EmilyMiller, suggest improvements to the voting system, we’re not just trying to protect ‘our precious’ – we’re trying to protect every author’s stories from the same enshittification we see happening everywhere else. We’re not trying to gain personal advantage, we’re trying to strengthen the site against (inevitable) manipulation.

You mention politics being manipulated. It’s a political tactic to create an alt universe based on straw-men and cognitive bias, leavened with twisted and cherry-picked versions of what the person you disagree with actually said, followed by demonizing your opponent, ideally personally. What counts is not facts or reasonableness, it’s owning the other side by any means possible.

This tactic has migrated to AH big time in the last few years.

 

[ThatNewGuy]

So when I, or others like EmilyMiller, suggest improvements to the voting system, we’re not just trying to protect ‘our precious’ – we’re trying to protect every author’s stories from the same enshittification we see happening everywhere else. We’re not trying to gain personal advantage, we’re trying to strengthen the site against (inevitable) manipulation.

Why does this site let compelling content, like Slingshot, get spat out the arse-end of the Top Lists and into obscurity, only to be replaced by more chapters of the same old story? This is a reader/consumer issue for the site to address, not just an author gripe.

Very well said. I also read, enjoyed, and commented on Slingshot. You know how I found it and why I read it? It was #1 on the SF&F all-time top list. It's a disservice to readers when a story that is rated 4.94(!) after 2,200(!) votes gets buried. Is it the end of the world? Not even close. Is it a problem that would be nice to have fixed? Absolutely.

Like you, I'm not trying to protect "my precious" here. My stories will get bumped further down their respective top lists if this issue is fixed.

Others are correct that this thread is unlikely to solve anything, but that's fine. Screaming into the void has always been a part of the AH community. The best any of us can do is to send Laurel and Manu a respectful PM asking if they have any insight into the cause of the top list flattening and whether anything is in the works to address it.

 

[Dodola]

enshittification

My favorite new word

Edit: *favourite,, for those who's eyes twitched a little when they saw this

 

[RejectReality]

A majority of the votes cast on Literotica are 5*. If people think something is good, they give it a 5*.

Your story, prior to being exposed to toplist trolling, received 5* ratings as more than 90% of all ratings. If it was viewed and rated by an extra group of people (for example: because it was briefly displayed on the top of the SF/F toplist), you would expect to receive extra votes that were also more than 90% 5* ratings. People who read your story think it's good, and people on Literotica don't need to think a story is amaze-balls terrific good to rate it 5*.

So if you got an extra block of votes and you wondered what the ratings given were, the most likely is just whatever implies the largest number of literal 5* ratings. Because we know that 5* ratings are more than 90% of the organic ratings that story receives.

The fact that this assumption paints a picture that fits EXACTLY with the votes over time graph of the previously studied trolling attacks shown earlier is something I find persuasive.

Middle chapters do not behave like self contained submissions. Being highlighted by the toplist or a contest win may cause a view spike, but votes only spike if the preceding chapters do as well. 3 5s on a single middle chapter simply doesn't fit readership patterns.

 

[RocketGrunt]

Middle chapters do not behave like self contained submissions. Being highlighted by the toplist or a contest win may cause a view spike, but votes only spike if the preceding chapters do as well. 3 5s on a single middle chapter simply doesn't fit readership patterns.

Why do you believe that?

The votes spiked. There was a 1 vote on that chapter from a normal readership of your series starting at chapter 1, right? That may have been triggered by you advertising it on this thread, or it may have been completely random. It's one viewer following normal read and rate patterns. Whether they were linked from this thread or found your series by a tag search or just got it as one of their random sci-fi suggestions is not something we're going to know unless they raise their hand.

And then you got 7 ratings specifically on that chapter that are one hundred percent obviously there because your story popped up at the top of the toplist, right? Some number of them are troll hammers, but why is it so bizarre to think that some of them might be actual people who clicked on it to read because it was advertised by the toplist as being of high quality?

Any number of "real" votes are overwhelmingly likely to be 5*. More 5* votes are cast than all other ratings combined. Not just on this story, but overall across the entire site. It's why I was so flippantly dismissive of the idea of dropping the "top 5% of votes." Every vote in the top quintile is an identical 5* vote, but every vote in the middle quintile is also an identical 5* vote. For a well regarded story like your own, more than half the votes in the bottom quintile are an identical 5* vote.

On the other hand, 3* reviews are incredibly rare. One person graphed their votes on this thread and identified only a single 3* vote in the entire time series (and believed that it came from a snarky forum poster in response to a statement that they never got 3* ratings at all). Certainly, the last time I am sure that I got a 3* rating on any of my stories, it came with a bewildering comment that they didn't like that my Pokemon fanfic kept using the phrase "It's super effective."

Troll votes, organic votes, almost none of them are 3*. People have made memes about it. So if we see 6 votes and the total is 18*, it is OVERWHELMINGLY more likely that three of them are 5*s and three of them are 1*s than that it is six 3*s. I have 59 stories with over three hundred votes each, and NONE of them have six 3* ratings on them.