I was wondering how other authors felt about story ratings. Do most care? Whats the average rating? Would really like to know what other authors think.
I was wondering how other authors felt about story ratings. Do most care? Whats the average rating? Would really like to know what other authors think.
I can't speak for authors, but as a reader, I definitely look at ratings before I open a story.
I generally will not bother with a story that is less than 5. If it is an author I am not familiar with, I will look to see what other scores that author has on their other stories.
For most authors, I will not read a story that is "In Progress". I've seen too many that have given up in the middle.
I have limited time allocated to reading, so I am getting selective.
Yeah, that's exactly why I didn't post anything until I finished it, and it's been edited. Then, I'd schedule all the remaining chapters to post as soon as I could after the first chapter was posted, if it was longer than a single chapter.
Thank you for your answer.
My stories seem to rate between 6 to 6.8. I write to please my fantasies first and hope others like it too.
I admit to being a ratings-and-download-count whore. I'm obsessed with those numbers. Since I don't try to sell my works, they are the only indicators I have as to how well, or not, I'm doing. Most of my stories end up with a rating of 7, give or take a few tenths, and I'm happy with that.
Download counts mean nothing. If you read a chapter it counts as a STORY download.
So, a story released serially, receives a "download" for each chapter posted if a reader reads each chapter as it is posted.
So, if 50 people read a story posted serially, and read each chapter as it is posted, that story receives 2500 downloads! Another 50 people read the story after it's finished results in only 50 downloads.
It's worse than that, honestly.
Picking one of my stories: Variation on a Theme, Book 6 says it has 979595 downloads.
Adding up the chapters' download counts, you get 2.1M.
The chapter marked 'Preface' has 106833 downloads, far more than any other chapter. Perhaps it's getting the credit for full-story downloads?
But if people were reading serially, I would expect the chapter download count to roughly equal the story download count. It's over 2x higher.
Even for Book 7, for which I would expect few full-book downloads (since it's nowhere near done), the total chapter download count is significantly high than the story's download count.
Big numbers better, but trying to make sense out of download count isn't worth it.
The chapter marked 'Preface' has 106833 downloads, far more than any other chapter. Perhaps it's getting the credit for full-story downloads?
The first will always be much higher. It has something to do with preloading. This is at the bottom of the chapters on the Stats page:
"Chapter download counts are not accurate due to the way the counting happens. For more info, check how stories and chapters stats are gathered" (the "how stories and chapters stats are gathered" is a clickable link)
On this website, no, it doesn't seem to matter when adjustments to the rating are automatic.
Removing malicious votes and duplicate votes to improve your score is one thing, but just removing votes automatically makes the rating unimportant to me.
On the other hand, this website feels setup to reward the authors who contribute the most, which I think is the right way to run a website.
Unlike the other website, why should someone who hasn't written a story in years be in the public eye more than someone who's written 10 stories in one year? (And I've benefitted from not writing but being seen, but that doesn't mean I think it's the best policy.)
I agree, those who actively contribute should garner most attention.
My thing about rating is why rate anyone story low? Stories I read that I don't think are good I just stop reading and move on. When I read a story I really like, I'll rate it good just to make it stand out.
That's just me, I could be wrong.
I agree, those who actively contribute should garner most attention.
So the authors producing volumes of AI slop should garner the most attention?
I read my comment, it didn't say anything about authors that use AI. That's another topic entirely and will be dealt with by the owner of this site eventually.
those who actively contribute should garner most attention.
I was commenting on the above. You didn't mention AI, but you state that those who contribute more should garner the most attention.
It could take weeks, months, or longer for an author to write and post a story. Someone using AI could do a story in a day or two. That's a lot of volume. A lot of contribution.
That was my point.
The good ol' AI topic to dismiss any all things someone doesn't want to hear.
Is AI allowed here? Should it be? If the AI was good, would that be okay? We aren't talking about AI. Bringing up AI is like bringing up, "But what if a piano falls on an author's head and they can't write for months, what then, huh? HUH!?
Contributing a lot on this website means posting often, whether that's complete stories or breaking your story into parts.
This website works best as a chapter upload site. Upload a chapter a week and you're going to be noticed.
I think you just need to upload consistently, whether it's weekly or monthly. Finishing stories regularly also helps build trust with readers, I've found. There's several authors here I don't read because they have a history of leaving stories incomplete.
Someone who uploads more consistently than someone else (more often) will be noticed more on this website.
As for leaving stories incomplete, it only takes one chapter to do that, but for the 50 chapters uploaded before that, that person is being noticed.
So when it comes to AI where do we draw the line? Is AI assisted OK? And how do we rate that?
Is AI assisted OK?
What does this even mean in the day of nearly all basic writing tools having some form of AI included, especially regarding grammar checking?
Does it include using an LLM to do consistency or continuity checks? Perform fact-checking? Updating a Wiki?
As you asked:
where do we draw the line?
That's the fundamental problem — as LLMs become more capable and are incorporated into more basic tools/software, drawing a line becomes nigh on impossible.
I set the barrier at using the AI to craft prose. Below that, it's just a tool like a spelling checker, grammar checker, or search engine.
I set the barrier at using the AI to craft prose.
What if you wrote "It was nerve-racking" and asked the AI for another way to say it? When the AI gave an alternative as "It filled her with apprehension" and you used that instead, didn't the AI write that prose?
What if you wrote "It was nerve-racking" and asked the AI for another way to say it? When the AI gave an alternative as "It filled her with apprehension" and you used that instead, didn't the AI write that prose?
It depends. 😜 What I meant by writing prose was, in effect, crafting from scratch or rewriting entire passages, not suggesting synonyms or similar idioms (things accomplished by a web search or thumbing through a thesaurus or idiom/quotations book).
This is why 'AI assisted' isn't going to work as a dividing line. It's getting to be very hard to avoid 'AI assisted' in terms of spelling or grammar checking. It's much easier to avoid 'AI assisted' in terms of rewriting.
On the other hand, sometimes that's subtle. If I write 'more clear', my grammar checker will suggest 'clearer.' If I change it, am I now 'AI assisted?' How about changing 'all of the things' to 'all the things?' What about removing or adding a comma? Technically, all of those are 'assisted' by an 'AI'.
Maybe it comes down to 'What percentage of the work was affected by AI?' With the examples I had above, it's less than 1%, probably less than 0.1%. That seems reasonable.
It's also worth noting that a human editor would have flagged any of those. At what point does the human editor become a co-author?
If a human who made X number of changes is still an 'editor' and not a 'coauthor', perhaps it's reasonable to say that the AI is also an 'editor' and not a 'coauthor.' Maybe that's a better dividing line?
Photographers in the past used to be measured by their composition and other skills. The only adjustments they had at their disposal were in the darkroom, things like cropping and burning and dodging. But today in the digital world of photography, they have tools that can make a photograph much better — or not even look like what they saw through the viewfinder.
Our world is changing.
Having a robot with an AI brain going around taking pictures is different than a person with a digital camera taking pictures and using tools to enhance the photo. Using AI to spit out a novel is the first, an author using AI to assist him is the second.
Now what happens when the AI robot takes better pictures than humans?
It's also worth noting that a human editor would have flagged any of those. At what point does the human editor become a co-author?
A good editor is a collaborator; mine does far more than make grammar and spelling corrections. My stories would be much less rich without him.
Oh, I agree, and mine do as well. But the more a good editor collaborates, the more the line blurs toward co-author.
Oh, I agree, and mine do as well. But the more a good editor collaborates, the more the line blurs toward co-author.
I think it turns on our understanding of the role.
- Developmental editors address structure and argument.
- Line editors sharpen prose and voice.
- Copy editors fix grammar, punctuation, consistency, and style guide compliance.
- Proofreaders catch final surface errors.
Mine does all four, and those are all editor roles, not author roles.
I write the prose, which is the author role.
Of those, in my workflow, Grok is a copy editor.
Are you your own line editor?
I am actually in the process of 'line editing' my stories as I have time (with help from my editor).
- Copy editors fix grammar, punctuation, consistency, and style guide compliance.
- Proofreaders catch final surface errors.
I've recently found listening to what I've written a good way to find these sort of errors. Not perfect but a nice tool.
I've recently found listening to what I've written a good way to find these sort of errors. Not perfect but a nice tool.
One of my proofreaders is vision-impaired, and she catches things everyone else misses!
My thing about rating is why rate anyone story low? Stories I read that I don't think are good I just stop reading and move on. When I read a story I really like, I'll rate it good just to make it stand out.
Using that logic, no story will stand out since only stories people like will be rated. I used to not rate a story I didn't finish, but I changed that. If I stop reading the story because it's not "my kind" of story, I don't rate it. But if I stop reading a story because it's not written well, I score it accordingly.
Well, also by my logic, if you see stories without many ratings that tells you all you need to know.
Not necessarily, in the case of older stories since until fairly recently an author could chose if they wanted to allow readers to vote or not.
If you saw that you could not rate a story then those wouldn't fall under the discussion we're having here.
I write under a different name.
At one level, I do not care about scores. On another level, if I am getting negative comments about something not understood (some of my stuff is rather techy) and the score is low, I pay attention.
That said, I write for me, an only myself. Which is evident per the some of stuff that I have written for no other reason than to get people so pissed that they can't see straight.
And yes, i am fucking drunk on a Tuesday morning, as are the dozen people around me. It's the 2d day of a funeral/wake for a good friend.
If I remember correctly, the scoring algorithm adjusts the score on every story to ensure that the average score is 6.00.
I do not understand. Are you saying that Laz enforces a normal (gaussian) distribution over the total of all story scores on the site?
You use the term 'average'. Is this referring to the medium or the mean?
I do not understand. Are you saying that Laz enforces a normal (gaussian) distribution over the total of all story scores on the site?
not over all the stories on the site. It's divided into periods that are defined by when the scoring system was changed. And the distribution he targets is not quite a strict normal distribution.
Here's the relevant FAQ entry.
It looks like it is two linear maps: one for scores above the mean, and one for scores below. If I understand it correctly, if the code determines that the current mean score is 8.00 then it would map scores between 8.00 and 10.00 linearly so they range from 6.00 to 10.00 (that is, 8.00 becomes 6.00, 8.50 becomes 7.00, 9.00 becomes 8.00, 9.50 becomes 9.00, and 10.00 stays 10.00). It would also use a different function to map scores between 0.00 and 8.00 linearly so they range from 0.00 to 6.00 (meaning that all those scores would be multiplied by 6/8). Of course, if the current mean score is actually 7.31, both mappings are a little more complicated.
You use the term 'average'. Is this referring to the medium or the mean?
No, or both. It's a weighted average, the mean of all scores except the top and bottom 5% cast, i.e. the mean of the median 90% of scores for the story. I unpick this a little in this post answering a query A.J. posed.
On a related note, I remember Ernest Bywater posting an explanation of how a '10' vote cast could reduce the score under very particular circumstances, probably in the old forums. Could anyone point me at that post?
Probably due to the issue where if a story only has minimum scores of, say '7' and a max of '8', voting '10' would then have that 10 being the extreme outlier and then discounted as part of the x% either side, which would then remove one of the'7's which would, in turn, reduce the overall average against the site median, causing the story to lose overall vote score. Possibly.
I cannot come up with a way for a 10 to lower the score.
I can give this example of a strange effect, though: With a certain number of votes, all are counted. With one more vote, the top and bottom are discarded. The change may happen on the 10th vote, which is when the 5% rounds up to 1.
Suppose a story has 9 votes: one 10 and eight 5s, for an average of 5.56. Someone casts a vote of 9, which is much higher than the 5.56, but it causes the 10 and one of the 5s to drop out of the average, so the new average is 5.50, which is lower than the previous one.
Even if dropping the highest vote happens on the 20th vote, or some other number, the same effect can happen.
I cannot come up with a way for a 10 to lower the score.
Neither can I, in fact I think I've proven it not possible, which is why I want to find EB's post - I thought I'd verified the claim with the numbers he gave.
Proof:
Assume that there are C votes that count (i.e., not eliminated as part of the top/bottom 5%), they sum to S and the lowest vote is L.
So, the pre-deskew average is S/C
(note tat the deskew function is monotonic to preserve score ordering)
Now add in the 10 vote, bringing the total votes up so another vote is dropped from each end.
The countable vote count goes up one and down two, to C-1
The sum of votes goes down by L to S-L
Now the pre-deskew average is (S-L)/(C-1)
If the score has dropped, then
S/C > (S-L)/(C-1)
which can be simplified to
S < LC
But this can never be true as S >= LC by definition
So, a 10 vote will never cause the score to drop
I agree with your calculations. I have also considered the case where the extra vote does not cause more votes to be dropped, and the case where the vote changes the story's score so that all scores get shifted. None of these scenarios seem to allow a 10 to lower a score.
I may have missed some other scenario.
It is also possible that the original claim was based on faulty reasoning. For example, if you use a different method to adjust the scores, you might be able to find such a situation, but I think I am correct about the calculations used.
Now add in the 10 vote, bringing the total votes up so another vote is dropped from each end.
I don't understand that. How would adding one vote cause two votes to be dropped? Pretty soon, the story would have no votes left that counted.
AJ
I don't understand that. How would adding one vote cause two votes to be dropped? Pretty soon, the story would have no votes left that counted.
To help mitigate the effect of 1 bombers, Lazeez implmented code that drops outliers when calculating the score, specifically the top and bottom 5% of votes.
5% of 20 = 1 so at 20 votes the top and bottom vote (2 votes) get dropped out of the score calculation.
Every additional multiple of 20 votes an additional pair of votes gets excluded from the score calculation.
I was wondering how other authors felt about story ratings. Do most care?
To be honest, I don't.
I write because I have ideas in my head that want me to tell them. And many times, I am sure what ends up in my stories can surprise them.
Of course, I also write more than a few stories with "bad endings". The main character blows it with somebody, they make a bad decision and end up worse off at the end, sometimes they even die. That is simply real life, and I have seen my scores dip when that happens.
I even laugh when many complain they did not like an ending, and I just say that they had better never watch "The Twilight Zone" then, because a hell of a lot of those stories had dark and bad endings.
"To Serve Man" and "It's A Good Life" come immediately to mind there.
Also, a great many I think will vote down a story if it does not match their particular fetish.
I even laugh when many complain they did not like an ending
I got a 3-star rating on ZBookStore on one of my novels because the main character "should have married Katie." That's a direct quote.
I actually liked Katie better too. But not my MC. So it wasn't a dark ending. It just wasn't the ending that reader wanted.
To be honest, I don't.
I wish I could say the same. But my ego is like the firebox on a steam train or Tia Billinger, both boxes need to be continually stoked for the engine to run...
If you've been on SOL a while then you know how they have worked over the years. As far as I know everything posted before Jan 2006 was just an average of votes.
After that an effort was made to control score inflation. So some formula or another was used to compare votes with current votes or something. Anyway we saw that the scoring average after that time a .8 lower. So a story that would score 9.0 before the change was made, would score an 8.2.
That seems to have been relaxed somewhat, but, probably because the data wasn't available to bring all the scoring "eras" into sync.
Anyways, if you have been around long enough to see it in action, it works just fine. Scores, along with tags are a pretty good way to find something you will probably like.
As I reader I don't care about ratings. However as I writer, I am constantly confused as to why my stories are always rated between 6 - 7. There are a few crap stories of mine that don't surprise me for low rates but the really good stories that have many downloads are still around the same mark. Either I'm just a mediocre writer, or my topics are just out on the fringe.