A
alwayswantedto
Guest
<deleted>
Last edited by a moderator:
The scoring still has me confused...
- Thread starter nakdsub
- Start date
Darkniciad
Literotica Guru
- Joined
- Jun 5, 2005
- Posts
- 7,944
The same would apply to a ten point system as well. The other site where I post has a 10 point system, and the admin has put a complicated mathmatical formula in place that nobody can understand in order to spread the scores ( which are clustered at 8-10 for the most part ) out.
The formula takes into account the scores of other stories posted during the same time frame as when yours posted, so your score is dependant upon everyone elses, and can change at any time without a single vote being cast on your story.
It's screwy.
It's an internet poll. It's always going to be skewed high ( unless the person in question is hated by enough people ) 5 points... 10 points... doesn't matter. You're at the mercy of human nature.
The formula takes into account the scores of other stories posted during the same time frame as when yours posted, so your score is dependant upon everyone elses, and can change at any time without a single vote being cast on your story.
It's screwy.
It's an internet poll. It's always going to be skewed high ( unless the person in question is hated by enough people ) 5 points... 10 points... doesn't matter. You're at the mercy of human nature.
CarlusMagnus
Literotica Guru
- Joined
- Feb 6, 2011
- Posts
- 1,183
No one? PM me the URL where I can find the formula.
As to the OP's problem:
With 381 votes, any total number of stars from 1603 to 1605 gives a score of 4.21, while with 383 voters, any total number of stars from 1603 to 1606 gives a score of 4.19. (Analysis of my own scores strongly indicates that a five in the third digit results in upward rounding of the second digit; it really is rounding, and not truncation.) So there are several scenarios under which the reported drop in score could happen, even if no Hoovering is going on. And if a sweep is in progress, all bets are off.
Those interested in this discussion who haven't already seen it may be interested in my How-To note on analysis of scores: http://www.literotica.com/s/how-to-analyze-your-scores. Warning: If you don't want to look at algebra, don't bother.
Okay, I'm starting to see what you mean now...
Thanks for taking the time to answer my question. Again.
Thanks for taking the time to answer my question. Again.
Thanks for the link to your analysis...
Very interesting and easy to read. Thank you.
No one? PM me the URL where I can find the formula.
As to the OP's problem:
With 381 votes, any total number of stars from 1603 to 1605 gives a score of 4.21, while with 383 voters, any total number of stars from 1603 to 1606 gives a score of 4.19. (Analysis of my own scores strongly indicates that a five in the third digit results in upward rounding of the second digit; it really is rounding, and not truncation.) So there are several scenarios under which the reported drop in score could happen, even if no Hoovering is going on. And if a sweep is in progress, all bets are off.
Those interested in this discussion who haven't already seen it may be interested in my How-To note on analysis of scores: http://www.literotica.com/s/how-to-analyze-your-scores. Warning: If you don't want to look at algebra, don't bother.
Very interesting and easy to read. Thank you.
Darkniciad
Literotica Guru
- Joined
- Jun 5, 2005
- Posts
- 7,944
Understanding the math is one thing. Understanding how the displayed score has anything to do with how people actually voted is another. Understanding why your score is dependant upon someone else's work is yet another.
The formula isn't actually posted anywhere, either. It's just a text description of the generalities. There's not even a time frame for "in the same time period as yours".
*shrug* Everyone's on the same scale, but it's still screwy.
The formula isn't actually posted anywhere, either. It's just a text description of the generalities. There's not even a time frame for "in the same time period as yours".
*shrug* Everyone's on the same scale, but it's still screwy.
No one? PM me the URL where I can find the formula.
scouries
Literotica's #1 Author
- Joined
- Apr 5, 2005
- Posts
- 4,989
Hey man, I just did the 5 star analysis for my best seller ”Daddy,” I Whispered. Here’s what I came up with:
[size=+3]
Five-star Analysis:
According to part a) of the Five-star Principle, the number of five-star votes that this story has received is no smaller than (4.52 – 4.005) × 33,936, or 0.515 × 33,936 = 17,477.04. Therefore, my story has to have at least 17,478 five-star votes.
On the other hand, part b) of the same Five-star Principle says that my story has no more than 33,936 × (4.52 – 0.995)/4 five-star votes. That's 33,936 × 3.525/4 = 29906.1 and I can't have any more five-star votes than that. But, again, fractions of a vote aren't possible, so the highest number of five-star votes I could have on this story is 29,906.
Conclusion: The number of five-star votes I have for this story falls somewhere in the range of whole numbers that begins with 17,478 and ends with 29,906.
Unfortunately not very helpful in coming up with the exact number! I'm still left with an over 12,000 vote range! My own observations lead me to believe that 73-76% of the votes cast on the story were fives (24,774-25,792).
Which isn’t that bad!!!!
[/SIZE]
[size=+3]
“Daddy”, I Whispered
http://www.literotica.com/s/daddy-i-whispered
Five-star Analysis:
According to part a) of the Five-star Principle, the number of five-star votes that this story has received is no smaller than (4.52 – 4.005) × 33,936, or 0.515 × 33,936 = 17,477.04. Therefore, my story has to have at least 17,478 five-star votes.
On the other hand, part b) of the same Five-star Principle says that my story has no more than 33,936 × (4.52 – 0.995)/4 five-star votes. That's 33,936 × 3.525/4 = 29906.1 and I can't have any more five-star votes than that. But, again, fractions of a vote aren't possible, so the highest number of five-star votes I could have on this story is 29,906.
Conclusion: The number of five-star votes I have for this story falls somewhere in the range of whole numbers that begins with 17,478 and ends with 29,906.
Unfortunately not very helpful in coming up with the exact number! I'm still left with an over 12,000 vote range! My own observations lead me to believe that 73-76% of the votes cast on the story were fives (24,774-25,792).
Which isn’t that bad!!!!
[/SIZE]
sr71plt
Literotica Guru
- Joined
- Jul 18, 2006
- Posts
- 51,872
CarlusMagnus
Literotica Guru
- Joined
- Feb 6, 2011
- Posts
- 1,183
I think 73-76% fives is very unlikely. Taking 75% as the fraction of fives would mean that you could have at most about 640 fours---and that only if all the rest (over 7800) were threes.
Darkniciad
Literotica Guru
- Joined
- Jun 5, 2005
- Posts
- 7,944
This is one figure you probably don't want to dispute he-who-shall-not-be-named on. When you cast most of the votes on a story yourself, you probably have a good idea what those votes were 
He's just not taking into account how many of his bogus IPs have been caught and the votes removed.
He's just not taking into account how many of his bogus IPs have been caught and the votes removed.
CarlusMagnus
Literotica Guru
- Joined
- Feb 6, 2011
- Posts
- 1,183
My guess is that He-Who-Must-Not-Be-Named doesn't really want a "true gauge".
CarlusMagnus
Literotica Guru
- Joined
- Feb 6, 2011
- Posts
- 1,183
This is one figure you probably don't want to dispute he-who-shall-not-be-named on. When you cast most of the votes on a story yourself, you probably have a good idea what those votes were
He's just not taking into account how many of his bogus IPs have been caught and the votes removed.
Oh, I didn't dispute the figure. I just pointed out how unusual the distribution of votes would have to be if the figure's correct.
nomennescio
Virgin
- Joined
- Jul 16, 2009
- Posts
- 4
I was going to say something along the lines of what this article describes. A 1-5 scale is simply not a productive way of arriving at well-distributed rankings. While there's some variation between categories, at least in those I read the entire scope of remotely readable stories are to be found in the 4-5 range. As a consequence, if you enjoy a story at all, the only thing you really can vote is a 5, as anything else would drop the story (often sharply) towards the no-man's-land below four.
Rather than expanding the possible rankings (would anyone really vote a story a 3.5? I don't think I've ever used 3, myself), I'd suggest the reverse - give only two rankings, up or down. Like or dislike. No more hemming and hawing about whether a story really deserves a five or whether you ought to give it a four (but then its score will drop, and you did like the story...) - you just click the little plus. No more wondering what you ought to vote a story whose spelling mistakes got too much in the way for you to focus on the action (does that really deserve a one? maybe I should just give it a two...) - you just click the little minus. Nice. Simple. Easy. Score will tell you what percentage enjoyed it of those who bothered to vote.
I'm sure no such change will be implemented, but it seems considerably better than the current system.
Darkniciad
Literotica Guru
- Joined
- Jun 5, 2005
- Posts
- 7,944
The problem, as with any suggested change of the point range, is that there are thousands of stories with an average of hundreds of votes each already in the system.
Every single one of those votes is going to have to be converted. Going with a 10 point scale means you have to double the existing scores, and that means that no existing votes are going to be odd numbers. Old stories are going to be on a completely different scale from anything new coming out. How do you convert to a thumbs up/down? Change all the 1s and 2s to down and all the 4s and 5s to up? What do you do with 3s then?
Voting would have to be shut down while the conversion was in process, once you figured out what formula to use. Nobody's going to like that, and it's not going to be quick.
Then there's the inherant problems of a thumbs up/down system. There's a reason that places like Facebook are refusing a "thumbs down" They've learned from the places that have implemented it that it's abused -- badly.
The change to the star voting system increased the number of people voting, and marked 4 as a good score. That was step one. The only logical way to decrease the vote inflation is to put in place a second H level, or eliminate the Hs completely.
And in the end -- as always -- its an internet poll. No matter what you do, people are going to muck with it. There's no way to stop it, and only a few ways to mitigate it.
Every single one of those votes is going to have to be converted. Going with a 10 point scale means you have to double the existing scores, and that means that no existing votes are going to be odd numbers. Old stories are going to be on a completely different scale from anything new coming out. How do you convert to a thumbs up/down? Change all the 1s and 2s to down and all the 4s and 5s to up? What do you do with 3s then?
Voting would have to be shut down while the conversion was in process, once you figured out what formula to use. Nobody's going to like that, and it's not going to be quick.
Then there's the inherant problems of a thumbs up/down system. There's a reason that places like Facebook are refusing a "thumbs down" They've learned from the places that have implemented it that it's abused -- badly.
The change to the star voting system increased the number of people voting, and marked 4 as a good score. That was step one. The only logical way to decrease the vote inflation is to put in place a second H level, or eliminate the Hs completely.
And in the end -- as always -- its an internet poll. No matter what you do, people are going to muck with it. There's no way to stop it, and only a few ways to mitigate it.
nomennescio
Virgin
- Joined
- Jul 16, 2009
- Posts
- 4
Conversion would indeed be rather a headache, though I don't see any reason it couldn't be quick - if there's one things computers are good at, it's rapidly crunching large piles of numbers. Votes on the 5-scale, you're right, don't have clear analogues on a like/dislike scale - indeed, that's rather the point, since if they did have clear analogues then the two scales would be functionally interchangeable anyway.
If I were the President-for-Life of Literotica, I'd probably just translate the current score into a percentage (4 = 80%, 4.5=90%, etc), multiply that by the number of votes to get 'likes,' and call it a day. Individual votes don't get individually translated, but the basic idea gets across, even though the story probably wouldn't have wound up with that resulting score if the upperdown system had been in place from the beginning.
Regarding 'thumbs down' abuse - true, something of the sort is inevitable. There are always trolls. But the current system is worse for that, not better. If 4.0 is the breakpoint for a moderately enjoyable story, then one troll voting 1 has three times the weight of someone voting five, in terms of the impact on the average score. I did consider the possibility of having no downvote at all, just a lone 'Like'. You could sort of try to abstract an indicator of quality from that, by comparing the number of views to the number of likes; it's not the worst idea in the world. But overall, I do think you get more genuinely useful information by adding the downvote.
And about the 'H' - I rather strongly disagree that adding or removing icons is going to make a significant difference to vote inflation. All scored voting is subject to the effect we see here, regardless of whether or not there are extra icons associated with high scores. People, broadly considered, do not offer a vote at all unless they feel strongly about the object, resulting in a bimodal distribution around 1 and 5, plus a scattering of 4s; the link alwayswantedto posted above talks a little about this. As long as you give people a spread of scores, most of that spread is going to go unused.
If I were the President-for-Life of Literotica, I'd probably just translate the current score into a percentage (4 = 80%, 4.5=90%, etc), multiply that by the number of votes to get 'likes,' and call it a day. Individual votes don't get individually translated, but the basic idea gets across, even though the story probably wouldn't have wound up with that resulting score if the upperdown system had been in place from the beginning.
Regarding 'thumbs down' abuse - true, something of the sort is inevitable. There are always trolls. But the current system is worse for that, not better. If 4.0 is the breakpoint for a moderately enjoyable story, then one troll voting 1 has three times the weight of someone voting five, in terms of the impact on the average score. I did consider the possibility of having no downvote at all, just a lone 'Like'. You could sort of try to abstract an indicator of quality from that, by comparing the number of views to the number of likes; it's not the worst idea in the world. But overall, I do think you get more genuinely useful information by adding the downvote.
And about the 'H' - I rather strongly disagree that adding or removing icons is going to make a significant difference to vote inflation. All scored voting is subject to the effect we see here, regardless of whether or not there are extra icons associated with high scores. People, broadly considered, do not offer a vote at all unless they feel strongly about the object, resulting in a bimodal distribution around 1 and 5, plus a scattering of 4s; the link alwayswantedto posted above talks a little about this. As long as you give people a spread of scores, most of that spread is going to go unused.
Darkniciad
Literotica Guru
- Joined
- Jun 5, 2005
- Posts
- 7,944
There was a significant change in the average score just from implementing the star system.
While the scores will always be inflated, removing the Hs, a visual cue that - like it or not - is heavily utilized as a determination of whether a story is worth reading or not, or adding a second H level from 4.0-4.49 would mitigate some of the knee-jerk 5 voting. The stars proved that enough readers do take note of and utilize something that tells the author they're doing a good job, without necessarily saying it's heaven-sent.
As to converting...
You have to track the individual votes in order to remove the multi-votes. Otherwise, the abuse already present mutliplies exponentially.
The conversion from ID# to Friendly URLs took a couple of days, and the site lagged like a mother-fucker.
That was a single change on each story. Now multiply that times the average of a couple hundred votes, and consider that you're messing with an active statistic, and not a static database reference that is created once and never changed. The whole time you're converting, new votes are coming in. My submissions alone over the last 3 months have accumulated nearly 10k votes.
I suppose you could shunt new votes to a different table, then move converted votes to that table as you went along. At the very least, you're going to have to remove references to score from everywhere it appears ( which is now, almost everywhere ) while it's going on, because the scores will be a clusterfuck during that process and people would lose their minds.
Not that they aren't going to lose their minds when scores vanish.
You'd also have to turn off toplist generation.
The list goes on and on.
You're talking about a major change to an ENORMOUS database, and almost every single piece of code that runs the website - Live, with thousands of visitors online at any given moment actively participating with the very data you're attempting to change.
And the end result would be an equally flawed internet poll, abused by trolls and cheerleaders alike, telling us nothing more than we have now, and causing authors to have fits.
Cost vs. benefit analysis takes about a nanosecond. It's not even remotely worth the tremendous time and effort that would be necessary.
While the scores will always be inflated, removing the Hs, a visual cue that - like it or not - is heavily utilized as a determination of whether a story is worth reading or not, or adding a second H level from 4.0-4.49 would mitigate some of the knee-jerk 5 voting. The stars proved that enough readers do take note of and utilize something that tells the author they're doing a good job, without necessarily saying it's heaven-sent.
As to converting...
You have to track the individual votes in order to remove the multi-votes. Otherwise, the abuse already present mutliplies exponentially.
The conversion from ID# to Friendly URLs took a couple of days, and the site lagged like a mother-fucker.
That was a single change on each story. Now multiply that times the average of a couple hundred votes, and consider that you're messing with an active statistic, and not a static database reference that is created once and never changed. The whole time you're converting, new votes are coming in. My submissions alone over the last 3 months have accumulated nearly 10k votes.
I suppose you could shunt new votes to a different table, then move converted votes to that table as you went along. At the very least, you're going to have to remove references to score from everywhere it appears ( which is now, almost everywhere ) while it's going on, because the scores will be a clusterfuck during that process and people would lose their minds.
Not that they aren't going to lose their minds when scores vanish.
You'd also have to turn off toplist generation.
The list goes on and on.
You're talking about a major change to an ENORMOUS database, and almost every single piece of code that runs the website - Live, with thousands of visitors online at any given moment actively participating with the very data you're attempting to change.
And the end result would be an equally flawed internet poll, abused by trolls and cheerleaders alike, telling us nothing more than we have now, and causing authors to have fits.
Cost vs. benefit analysis takes about a nanosecond. It's not even remotely worth the tremendous time and effort that would be necessary.
nomennescio
Virgin
- Joined
- Jul 16, 2009
- Posts
- 4
Can't say I agree with a great deal of your reasoning, but I respect your opinion.
A
alwayswantedto
Guest
<deleted>
Last edited by a moderator:
DG Hear
My Friend Kipper
- Joined
- Jun 14, 2005
- Posts
- 6,689
I was just sitting here thinking. First of all nothing is going to change but I was wondering if we just got rid of the 1 vote, which is considered a troll vote most of the time and added a 4.5.
That way even if the story isn't perfect voting 4.5 would not hurt the possible H. Voting would be 2 3 4 4.5 5. All stories already here would be left alone. It would only affect new stories. Of course if manangement got rid of the one votes (if there are any left) it would raise the score of the remaining stories.
Just a thought.
DG
That way even if the story isn't perfect voting 4.5 would not hurt the possible H. Voting would be 2 3 4 4.5 5. All stories already here would be left alone. It would only affect new stories. Of course if manangement got rid of the one votes (if there are any left) it would raise the score of the remaining stories.
Just a thought.
DG
Darkniciad
Literotica Guru
- Joined
- Jun 5, 2005
- Posts
- 7,944
Keep in mind that we have no idea how the database is structured. It may not be as simple as swapping them. Voting started in the dark ages of the internet, long before best practices were worked out.
The reason I argue so hard against suggestions like this is how much work they're going to require on Lit's part vs. a very limited ( or no ) reward.
Adding another H level is probably within the scope of the existing code and likely configurations of the database, requiring limited coding time. Ditto for removing the Hs. The changes everyone saw in the incoming scores from the star voting system is precedent that a visual change like this has an impact on reducing auto-five voting.
You have to go practical, simple, and within the scope of what's already here, if you want to have any hope of seeing it actually happen.
The reason I argue so hard against suggestions like this is how much work they're going to require on Lit's part vs. a very limited ( or no ) reward.
Adding another H level is probably within the scope of the existing code and likely configurations of the database, requiring limited coding time. Ditto for removing the Hs. The changes everyone saw in the incoming scores from the star voting system is precedent that a visual change like this has an impact on reducing auto-five voting.
You have to go practical, simple, and within the scope of what's already here, if you want to have any hope of seeing it actually happen.
scouries
Literotica's #1 Author
- Joined
- Apr 5, 2005
- Posts
- 4,989
angie I'd worry less about the score and more about the feedback
photo by magnus…
magnus I think 73-76% fives is very unlikely. Taking 75% as the fraction of fives would mean that you could have at most about 640 fours---and that only if all the rest (over 7800) were threes…
My guess is that He-Who-Must-Not-Be-Named doesn't really want a "true gauge"…
Oh, I didn't dispute the figure. I just pointed out how unusual the distribution of votes would have to be if the figure's correct…
photo by magnus…
magnus I think 73-76% fives is very unlikely. Taking 75% as the fraction of fives would mean that you could have at most about 640 fours---and that only if all the rest (over 7800) were threes…
My guess is that He-Who-Must-Not-Be-Named doesn't really want a "true gauge"…
Oh, I didn't dispute the figure. I just pointed out how unusual the distribution of votes would have to be if the figure's correct…
Angie is 100% right on – It’s why [SIZE=+2] ScouriesWorld[/SIZE] has been a long time supporter of the removal of the red H. It’s why many years ago we created the two most informative and respected FEEDBACK threads on the site – the MOST COMMENTS and the MOST VOTES threads. It’s why we continually emphasize the MOST FAVORITED (stories and authors) lists.
As to our aspiring mathematician he seems, like so many of his ilk, to be unable to make the intellectual jump from the raw numbers to an intelligent analysis of them. Technicians like he should be content to simply provide the data and not try to interrupt them. Leave that to the experts my good man!
fyi this story has garnered more “1” votes than any story on the site. In fact for a period of time it was under relentless attack from a group of fanatical arab muslims – an attack that was only stopped when the QUEEN turned of all voting from Iran and Libya!
As to our aspiring mathematician he seems, like so many of his ilk, to be unable to make the intellectual jump from the raw numbers to an intelligent analysis of them. Technicians like he should be content to simply provide the data and not try to interrupt them. Leave that to the experts my good man!
fyi this story has garnered more “1” votes than any story on the site. In fact for a period of time it was under relentless attack from a group of fanatical arab muslims – an attack that was only stopped when the QUEEN turned of all voting from Iran and Libya!
TxRad
Dirty Old Man
- Joined
- Jan 13, 2005
- Posts
- 45,152
Poor delusional scouries. He actually thinks someone believes his crap. 
Anyway, most every thing he said in the above post is a lie. Laurel and manu could care less who 1 bombs his crappy stories. As for turning off all voting from a region, that isn't even possible with the software they have here from what I've seen.
Sorry scouries, more of your crap that don't float. Take your sore thumbs and go bomb the stories in the contest again. You've been doing it since day one. Not just mine but everyone's.
As for your fake lists, you only made those up so you could associate your crap and your female alt's crap with the good writers here on Lit. Not to mention the lists are far from complete or honest. People you don't like or who write in categories you don't like, never make the lists and when they do, you change the criteria. Just more lies and crap from the king of lies and crap.
Anyway, most every thing he said in the above post is a lie. Laurel and manu could care less who 1 bombs his crappy stories. As for turning off all voting from a region, that isn't even possible with the software they have here from what I've seen.
Sorry scouries, more of your crap that don't float. Take your sore thumbs and go bomb the stories in the contest again. You've been doing it since day one. Not just mine but everyone's.
As for your fake lists, you only made those up so you could associate your crap and your female alt's crap with the good writers here on Lit. Not to mention the lists are far from complete or honest. People you don't like or who write in categories you don't like, never make the lists and when they do, you change the criteria. Just more lies and crap from the king of lies and crap.
Last edited:
sagethorns
Really Experienced
- Joined
- Jun 6, 2006
- Posts
- 281
I must admit that as a reader with limited time I look to the H of new stories for the categories I'm interested in that day. That definitely makes me miss some very interesting and good stories but I have no other means of avoiding some of the truly awful stories out there.
As an author all I care about is that my story isn't too adversely affected by troll votes and is fairly represented by those that have voted.
Is it naive of me to think that removing the H, posting the current score and number of views/votes next to the story is enough?
As an author all I care about is that my story isn't too adversely affected by troll votes and is fairly represented by those that have voted.
Is it naive of me to think that removing the H, posting the current score and number of views/votes next to the story is enough?
