Proposal For Voting Reform - For All To Review

[Guest]

To the Hermit:

I confused you and the Earl, and have clarified my last post.

You really do seem to have trouble listening, and aside from my confusing you and Earl, I see no mistakes in WH's or my comments.

If you want to mobilize, you need to learn to be concise, as I was in three proposals to abate the "1" problem.

Your claims are not quite accurate, nor is your last post.

The first three, you say, would virtually eliminate the impact of fraud. Let's see.
1) was to have ratings 3-5, with half steps, 3 being the new "awful" instead of "1".

WH has commented, as have I. You preserve your "hot" only assuming it's not recalibrated, as it certainly would and should be.
It's supposed to isolate a top 5%, or less.

To achieve that under your 1) would require making H awarded on the basis of, say, 4.75. At that point, the impact of a "3" is exactly the same as that of a "1" before.

2) is to show distributions. I have agreed with this. It would allow an author to see what the non-1 votes amount to. The author could try to see the center of the bell curve, if there is one.
Or calculate a median score.

3) was to lop off the bottom votes (option 1). This ignores fraudulent top votes. It's the reason I said your intent simply seems to be to raise. Lopping off top and bottom, you mention as option 2, and I favor that, though you don't.

In sum, 2) and 3) second option would help. If you would stick to those, in concise postings, you might get support.

It's to be noted that measures such as I described would have to be added to genuinely have an impact against a serious ratings vandal.

Lastly, the claim, in your last post, that it takes 180 fives to counteract 5 ones, is simply not true, if you are talking about remaining at or above the 4.5 level, i.e., maintaining a "H" rating. The correct number is 35 fives (to counteract 5 ones) . In short, a "1" has **seven** times the power to lower, as does a "5" to raise, above 4.5. Your claim is apparently that the 1 has **thirty five** time the power, and that's an error.

In general, the idea of bad guys with A bombs and good voters with pee shooters is colorful, but as WH and I have shown, if there is a "top" rating (average), then the lower votes necessarily have disproportionate effect, and your manipulations don't avoid that. The only genuine cure is to make sure the "1"s are given sincerely, by real people, an issue I dealt with in my last post.

 

[Pure]

The above posting to the Hermit (about proposals 1-3) is by me, Pure. Sorry about the computer glitch.

 

[Weird Harold]

Re: Check Your Facts

I am responding to some of Pure and Weird Harold's comments.

I suspect from some other comments made by you and Harold that you do not truly appreciate the disproportionate power now given to the tiny minority that periodically throw out 1-votes like rice at a wedding.

Pure addressed the statistical flaws in your arguments and suggestions, so I will refrain from repeating them. She (?) seems to have a firm grasp on statistical principles that you apparently lack. Probably better than my understanding, although I'm not completely ignorant about how statistics can be manupulated.

Weird Harold seems to suggest that nothing should be done because nothing has been done in the past. That Laurel won't act in order to preserve 'free speech'. If true, the free speech currently being protected is the same free speech practiced by 19 terrorists on 9/11. Yet we give the pea-shooters to the good voters [5-vote] and A-bombs [1-votes] to the terrorists who are casting votes simply to destroy. Furthermore, it is highly improbable that the fraud votes are evenly distributed.
...
As for Laurel not acting in the past… that is possibly because this community has never really pushed for reform.

First, your comparison to 9/11 is tacky, overblown and irrelevant. It's NOT likelyto win you any converts.

First, I'm not saying that "nothing should be done because nothing has been done." Far from it in fact. If you had read my post closely, you would note that I alluded to "the changes that have been made in the past" and "participating in this same discussion about once a year since I found Lit. Some changes have resulted from the discussions, and it's possible that further changes are in the works and being considered.

Laurel has acted in the past -- both to make changes to the voting system to help eliminate fraud and to specifically reject all of the things you propose plus several other schemes that you seem to have missed.

This is Laurel's site, and I will NOT continue to push her for changes she has considered and rejected.

When and if I come up with some inspiration to improve the integrity of the voting system, I will directly e-mail Laurel with the suggestion instead of airing the proposal in public to give the trolls advance warning.

There are things being done behind the scenes to eliminate fraudulent votes and if you feel your story has been maliciously attacked in a direct attack on you or can convince her that fraud is involved in an attack to promote a particular story for the monthly contests, she can and often will investigate and remove votes -- The result might NOT be what you want, because she removes fraudulent five votes too and they are oftenmore numerous than fraudulent ones.

FWIW, I have searched out several of the authors who are complaining about their stories losing the coveted "H" bcause of "Fraudulent voting." (BTW, I couldn't find anything by you -- cre to provide me a link via PM?) In my personal opinion, I can't see why any of them ever had an "H" rating in the first place.

Most used too much passive voice and were consequently "flat and uninteresting" as Purdue University's page on passive voice predicts. All had a profusion of long sentences that are hard to follow, misused and misspelled words, very little dialogue, and what dialogue they do use is hard to follow and stilted.

A couple of the stories were good "stories" but the poor execution would keepme from ever giving them a five -- a cuple were worththe fours I gave them and the rest rated a three at best. (which I did NOT give them becuse I was too bored to make it worth logging back onto vote.)

 

[karmadog]

Harold's post illustrates a strong point: If anything, the scores are skewed higher by the current system.

Stories that deserve a low score don't get them because those who might vote leave the story before getting to the bottom of the page. In this case, it doesn't take many friends to reward a story that has little merit.

On the other hand, They probably won't be good enough to make it to the top list. The problem with the '1' vote as I see it, is that a story that makes the toplist, is immediately hit.

I put a poem in my sig line that made it to the toplist without being hit. But it stayed there for a very short time before being knocked off. This didn't bother me too much, as, quite frankly, it didn't deserve to be considered that high. But what if it did?

A poem or story that is not on the toplist, and not on the 'new' page, may have a hard time drawing readers. To me, that is the worst thing about the troll vote. I just like to have my stuff read. It's nice to get good votes or useful criticism, but I value my read count far more highly.

BTW, I am NOT complaining. I think Laurel has a terrific site here.

 

[KillerMuffin]

some bastard gave me 1s because he went through every story in the Celebrity top list and gave them all 1s and screwed my scores.

When'd this happen? My celeb story hasn't moved--vote wise--since January. It's on the top list. Granted it's at #60, but it's still got the H.

 

[Weird Harold]

Harold's post illustrates a strong point: If anything, the scores are skewed higher by the current system.
...
A poem or story that is not on the toplist, and not on the 'new' page, may have a hard time drawing readers. To me, that is the worst thing about the troll vote.

I'd quibble that it's not thesystem but the dynamics of voter psychology that skew the intial votes higher than a story objectively deserves.

Your point about the effect on readership is what is really at the heart of these periodic "discussions" about voting "fraud." Your tactic of "shameless self-promotion" is as good a method of keeping your readership up as any possible reform to the voting. Every story eventually falls off the top story lists, but keeping the link infront of people who don't visit the top story lists (like me) keeps a trickle of interest alive.

 

[KillerMuffin]

Laurel is busy. Period. She currently runs three active sites. She currently reads about 6 hours worth of porno every night. She also has a full time job.

She has plans for voting reform anyway. If anyone were interested in this topic, then one would have searched. She's mentioned things along the lines of what imdb does. Like all things, the programming takes time and costs money. Everything was put on hold last year--in case it slipped everyone's notice, she also has to deal with the consequences of COPA. Why on earth should she sink funds out of her own pocket for a site that may end up shutting down?

NONE of the reforms I saw ever considered the reader in this. Other than as an evildoer. What's the vote ratio now? For every 100 reads you get one vote? Every 1000 reads? And you want to make it more difficult for a reader to vote? Think of it from the average Lit-goer. The one who shows up with his cock hanging out of his zipper looking to get off and then go watch Letterman. He may or may not vote. A couple of clicks is one thing, it's easy enough to do. Adding to that lowers the number of votes.

I also firmly believe that a lot of these "1" votes people are screaming about weren't even "1" to begin with. How do you know it was a one? How do you know it wasn't a three? Take a drop from 4.67 to 4.25 in one vote. What was the number that dropped it? A one? Wrong. Look at some math. 5+5+4= 14/3= 4.67 5+5+4+3= 17/4= 4.25 You see a drop like that in one vote you're going to assume a one, aren't you?

I firmly believe that a lot of us believe we're getting knocked down when that's nothing more than pure vanity talking. After all, it's just not possible for a story to hit the top lists, get a couple of 3's because the readers oh my gawd! didn't like the story, and drop. It has to be the mostly mythical one bandit. Unless you're running over 100 votes, it's not that difficult to drop your score pretty dramatically with a few honestly earned threes.

 

[Octavian]

Compensating for the influence of ‘1’s in the voting

It seems to me there is a need to give some examples.

Say a story has 130 votes with an average of 4.70 and then it gets zapped with 3 ‘1’s.
The new average is 4.62. It would need an additional 37 ‘5’s to compensate.
If the same story got hit 5 times the average would go down to 4.56 and it would need 63 ‘5’s to restore it to its original score. It is precisely this abuse that Hermit was addressing with his proposals to reform the voting.

But if one takes Weird Harold seriously why bother? Hardly any of us who are complaining have stories of any merit anyway.

As I said earlier it’s your loss.

 

[Weird Harold]

But if one takes Weird Harold seriously why bother? Hardly any of us who are complaining have stories of any merit anyway.

I never said the stories didn't have any merit, I said they weren't exceptional or what I would consider deserving of the coveted H. I however am only one vote and don't control what gets the coveted "H" ratings, although I can give anyone who is interested detailed reasons for not giving a five.

 

[TheHermit]

The Math - numbers don't lie

The average person will not intuitively understand the power of the numbers in the ranking formula used at this site. In the example I posted of a story after 45 votes here is the math:

Assume the story has a rank of 4.88889 [which rounds up to 4.89]. The votes cast are:

1=0 2=0 3=0 4=5 5=40 Total votes cast = 45

The program needs points to feed into it's rank calculation this is done as follows;
[ 4 (value of vote) x 5 (number of 4-votes) = 20 points ]
[ 5 x 40 = 200 points]
Rank = 20 points + 200 points / 45 votes cast = 4.88889


Along comes the troll, it casts 5 1-votes. Note: the numbers come out the same whether is occurs over an hour, a day, or a month.

1=5 2=0 3=0 4=5 5=40 Total votes cast = 50

[ 1 (value of vote) x 5 (number of 1-votes) = 5 points ]
[ 4 x 5 = 20 points]
[ 5 x 40 = 200 points]
Rank = 5 points + 20 points + 200 points / 50 votes cast = 4.5


At this point in order for this story to regain the 4.89 it had earned requires 180 5-votes

1=5 2=0 3=0 4=5 5=215 Total votes cast = 225

[ 1 (value of vote) x 5 (number of 1-votes) = 5 points ]
[ 4 x 5 = 20 points]
[ 5 x 215 = 1 075 points]
Rank = 5 points + 20 points + 1075 points / 225 votes cast = 4.8889


If the suggestion to change the lowest vote's value to a 3 was implemented the instead of falling to 4.5 here is what would have happened:

Along comes the troll, it casts 5 1-votes. Note: the numbers come out the same whether is occurs over an hour, a day, or a month.

3.0=5 3.5=0 4.0=5 4.5=0 5=40 Total votes cast = 50

[ 3 ( value of vote) x 5 (number of 1-votes) = 15 points ]
[ 4 x 5 = 20 points]
[ 5 x 40 = 200 points]
Rank = 15 points + 20 points + 200 points / 50 votes cast = 4.7 [versus 4.5 now]


At this point in order for this story to regain the 4.89 it had earned requires 85 5-votes [versus 180 today]

3.0=5 3.5=0 4.0=5 4.5=0 5=125 Total votes cast = 135

[ 3 (value of vote) x 5 (number of 3-votes) = 15 points ]
[ 4 x 5 = 20 points]
[ 5 x 125 = 625 points]
Rank = 15 points + 20 points + 625 points / 135 votes cast = 4..8889

This is why the simple substitution of a low value of 3.0 versus the currnet 1 would instantly reduce the impact of fraud voting by over 50%.

This is why many stories that deserve a little time on the Top List often disappear so quickly, robbing them of their chance to be read and possibly develope a following.

Furthermore, the 3.0 to 5.0 might mean that many people who vote a 5 today, might opt for the 4.5. No way to know, but ranking would not necessarily rise. If a story is truly average, more poeple might vote a 4.0 because it would not hurt a story the way a 3 does today. TH

 

[Weird Harold]

Re: The Math - numbers don't lie

The average person will not intuitively understand the power of the numbers in the ranking formula used at this site.

The ranking formula is a simple average of the votes -- what's to "inuitively understand?"

You didn't answer my point that all this has been presented to Laurel and specifically rejected!

If you can come up with a NEW argument or suggestion, then do so, but no matter how outraged you are, every proposal you have made has been considered and rejected.

If you can't see the flaw in simply changing the scale and ignoring positive fraud, then perhaps you need to take a course in statistics.

Oh, by the way, did I mention every suggestion you've made has beenpreviously considered and rejected!

 

[Guest]

Re: The Math - numbers don't lie

Hello, Hermit,

It's unfortunate that you do no read more carefully.
It's reasonable to consider the power of a "1" in relation to maintaining a 4.5 or "h" rating, which is what you're screaming about.

IF you're talking about maintaing some higher rating (in your example, 4.89 then the power increases (number of 1's it takes)
AS MUCH AS YOU LIKE, as the rating is made higher.

So here is some math that will really fuel your grievance.
Assume a person has 19, fives, and 1 four, for an average of 4.95.
At his point, a five draws up by only .05, while a one lowers by 3.95, and the power of the one is *seventy nine* times the power of the five--double the power in your example, and it would take about five TIMES seventy-nine 5's --395 fives to repair.
OH MY GOD! WE'VE GOT TO FIND OSAMA. AND TAKE OUT IRAQ.

So the impact can be made as disastrous as you please. Indeed, if someone has 20 fives, and a perfect 5 average, then it will take infinitely many ones to recover, or, given the rounding, SEVEN HUNDRED AND NINETY EIGHT fives to counteract a SINGLE one, or about FOUR THOUSAND fives to counteract FIVE ones.

HOLY SMOKE, OSAMA HAS TUNNELED **INTO** THE WHITE HOUSE!

The rest of your math isn't worth much comment, since you ignore what WH and I have proposed. It's a bit like saying, if new families and working couples can't afford homes when their joint incomes are under $50,000 per year, then the fed should print up money and give each couple a bag of new bills in the amount of $20,ooo, to put toward a home. Think. What's wrong with this???

What's done or not done in the past is not of that much concern and I won't get into predicting Laurel, saying she'll change or not.
She, that is literotica, depends on a satisfied bunch of customers.
Hence if there were a major problem and a substantial bunch of people advocating a simple solution ---not something as nighmarish as the Hermit's-- it would likely be considered and even done. And a solution would have to be far simpler, and defended in fewer, less hysterical words to attract mass attention.

The average person will not intuitively understand the power of the numbers in the ranking formula used at this site. In the example I posted of a story after 45 votes here is the math:

Assume the story has a rank of 4.88889 [which rounds up to 4.89]. The votes cast are:

1=0 2=0 3=0 4=5 5=40 Total votes cast = 45

The program needs points to feed into it's rank calculation this is done as follows;
[ 4 (value of vote) x 5 (number of 4-votes) = 20 points ]
[ 5 x 40 = 200 points]
Rank = 20 points + 200 points / 45 votes cast = 4.88889


Along comes the troll, it casts 5 1-votes. Note: the numbers come out the same whether is occurs over an hour, a day, or a month.

1=5 2=0 3=0 4=5 5=40 Total votes cast = 50

[ 1 (value of vote) x 5 (number of 1-votes) = 5 points ]
[ 4 x 5 = 20 points]
[ 5 x 40 = 200 points]
Rank = 5 points + 20 points + 200 points / 50 votes cast = 4.5


At this point in order for this story to regain the 4.89 it had earned requires 180 5-votes

1=5 2=0 3=0 4=5 5=215 Total votes cast = 225

[ 1 (value of vote) x 5 (number of 1-votes) = 5 points ]
[ 4 x 5 = 20 points]
[ 5 x 215 = 1 075 points]
Rank = 5 points + 20 points + 1075 points / 225 votes cast = 4.8889


If the suggestion to change the lowest vote's value to a 3 was implemented the instead of falling to 4.5 here is what would have happened:

 

[Pure]

The above response to the Hermit and the threat of OSAMA is by me, Pure. I wish I knew why after logging in and posting, the post comes "unregistered" and I've been automatically logged out, and have to log in again. I suppose the tentacles of Al Qaeda reach very deep.

J./

 

[Uther_Pendragon]

Re: BIGGEST BANGS FOR THE BUCK

Suggestion 1

Without disrupting your current ranking in any meaningful way, the 1 to 5 voting values could be switched to these values: 3.0 3.5 4.0 4.5 5.0













Advantages:

This is highly likely to be dirt-cheap as well as easy to implement. The new and older ranking numbers will co-exist perfectly. The few stories ranked below 3.0 today might rise over time, but so what.

Some folks presently wish there was a vote between a 4 and 5 - they want to be able to say it's good enough for an H, it IS above average, but it could be better. People aren't voting in some cases because they can't justify a 5, but don't want to 'hurt' a story with a 4, so you might get actually more voting.





None that I can see. [/B]

Sorry for how this response shows up. The original is incredibly badly formatted. This part of the proposal would not affect the scores on puture stories in any meaningful way. There are five possible scores now; there would be five possible scores then. What it would change is the relationship between scores on past stories and scores on future stories.

 

[Uther_Pendragon]

I tried to respond to the first proposal, but that proposal
was formatted so that it filled the page.
What the proposer doesn't understand is simple arithmetic.
So long as you have five values, and those values are
an equal distance apart. Then you have the same system
for comparing stories. You can have the top value 5, and
the bottom 1. You can have the top value 100 and the
bottom value 96. The scores are different, but the
comparative values are all the same.
Now, if you say that the lowest vote on the new system
is 3, then you slant the comparison between votes on the
old system and votes on the new system in favor of stories
voted on the new system. You haven't, however, changed
the system permanently.
|
Now, the suggestion to put words in place of numbers makes
a lot of sense. Stories On Line gave guidance for using
its numbers, and the average scores changed.

 

[its Leslie]

6 Hours of porn literature a day !!!!

God I sure don't think I could read that much.
Or handle that much

Well maybe if I had my wife sitting in my lap slowly gyrating in a sexy nighty while I read

Wherever this argument goes, I sure hope Laurel gets to get off after reading all that porn daily .

 

[Weird Harold]

Now, the suggestion to put words in place of numbers makes
a lot of sense. Stories On Line gave guidance for using
its numbers, and the average scores changed.

I believe that is one of the pending changes KM mentioned. It requires reprogramming the script and reformatting the voting box. Not a difficult change by itself, but it's part of an upgrade package for the entire site that is still in-work, or on hold pending the resolution of the COPA legislation -- I'm not sure which.

 

[TheHermit]

Killermuffins

Your idea that some people might be thinking a 3-vote was a 1-vote could be happening. However, the author's who have sent me copies of spreadsheets they used to track the actual votes know when 1-votes occur. If you sample often enough, and your computed ranking number rises and falls when Lit.org's does, they can be sure the estimated vote distribution entered into the spread sheet is accurate.

The suggestion to display the vote distribution is intended to eliminate this and many other misunderstandings. Speaking of which, I direct you to a new thread I came here tonight to post.

I don't understand why people attempting to offer constructive suggestions are viewed as attackers. The voting system is fraud prone and could be better.

The suggestions offered up front were collected from many voices, in many threads. And I documented many of the sources.

I have designed computer systems for years, my time and ideas normally cost a lot. The time invested in consolidating the scattered opinions and preparing a viable recommendation consumed many hours that I could have well used elsewhere.

Laurel is no doubt a terrific person and this is a fine site. If this was a crap site, I wouldn't have wasted a minute on it. Maybe the COPA you mentioned will hurt, maybe not. A disaster could wipe out anyone or anything tomorrow, but people generally proceed as if there is a future.

The cost to program, test and implement the Suggestions 1~3 I still recommend should be 4 to 8 hours TOTAL, plus be compatible with today's system - no one would have to do a single thing differently. One person-day to counteract the Troll Factor is not a kingly sum. A key reason for my expending the time to write the ideas was so EVERYONE, except the trolls, would waste far less time, especially Laurel.

If the voting relegates the fraud to a minor annoyance, author's can focus on writing, readers can find the better stories, and Laurel can worry about more productive pursuits.

The suggestion of a suggestion box is an example of that. When the same issues and ideas a raises time and again one of 2 things happens. Either no one replies to and the author gets frustrated, or the same messages have to be re-typed.

How many times have you typed that Laurel works full time, or that she runs several sites? Maybe that should be part of a FAQ or concerns forum. Then people have a chance to read what's been stated before, or more experienced people can direct newbie's to a thread that fills in background.

A forum for ideas/concerns that have been raised and answered is intended to hopefully lessen the re-inventing of the wheel. If Laurel actually decides once and for all that none of the ideas the community has put forth will not be adopted, then why not put the suggestions and an official reply in a public vault or FYI paper on voting issues for future reference.

 

[tenyari]

You will never truly fix a voting scheme.

That said, I'm a renderosity artist, and one thing they do is put the name of who cast the vote next to each vote and a small comment.


I would suggest something like that.

Instead of just typing in a number, require a minimum of a 25 and up to 80 character comment for the vote to be cast, and then list all the votes, along with who cast them and what they said.

The name would include a link to that person's profile.

This would of course only allow for people with membership to vote...


What is the true utility of a vote though?

Personally as an author I get nothing from the numbers. I get a lot from the feedback I am sent. That's where people speak their minds and I can find out -WHAT- they liked or didn't like and why. I get specific points to address, correct, or continue.

The number? That's just a ranking that helps to get me more hits from the people surf by order of ranking.

On renderosity, They installed a system where an author could get the comments without a vote, or a viewer could leave a comment without voting.

You can do that here of course by sending feedback rather than voting. The difference there is that all the comments are public, listed below the image.

The big advantage of their system is that the moderators can quickly tell who's giving an honest ranking and who's trying to trash the system. In fact the entire community can see it.

The disadvantage is that fights break out from time to time over the ratings or the comments and when you know who to target for what you're mad about it can get nasty and public. Late 2000 I believe one of the artist's got her first '9' ranking out of 10, and she got so mad she demanded the person ranking her below perfect be banned, and together with her clic of followers began trashing the person for a few weeks publicly. Several people came out saying the problem wasn't the '9', it was her insistance that she always get '10s' and the interesting nature of who had been giving her all those 10s...

Fortunatly, no-one here seems to believe they must always get a '5' on every story, though I wonder if there are people who believe a story must be downrated not because it has flaws, but because they feel letting it have that '5' is somehow wrong. I don't know the community well enough to answer that yet.



********************

Another idea:

Drop the lowest 10% of a ranking, and the highest 10%.

If a story gets 20 votes, you drop the lowest 2, and the highest 2, then compute the ranking.

If it had 3 5's, it would now only have one. But likewise if it had 3 1'.

Rubbing off a bit of the extremes as flukes.

 

[Coolville]

Yeah, yeah, yeah, I know. It's my standard reply to all these voting reform threads. Sorry, but...

shouldn't you people be writing?

You become a better writer by writing, not by worrying whether or not cockhard2003 in shitsville, arkansas voted 1 or 5.

I'll shut up now...

 

[Weird Harold]

You will never truly fix a voting scheme.
...
What is the true utility of a vote though?

Personally as an author I get nothing from the numbers. I get a lot from the feedback I am sent. ...

The number? That's just a ranking that helps to get me more hits from the people surf by order of ranking.

********************

Another idea:

Drop the lowest 10% of a ranking, and the highest 10%.

"Regularization," or dropping upper and lower votes, is one of the most common suggestions, and one Laurel opposes. I pesonally think you're right, it is the best solution to get a "truer" ranking fo stories.

Your comments about the utility of a vote is really more to the point; The purpose of voting here at Lit is just a short way of recommending a story to others. There are monthly contests, but the votes for contenders are checked for fraudulent voting before the winners are selected. Beyond the first months votes, there is no serious value to them except to rank the story against others. Your story about renderosity's problems clearly illustrates that some authors are far to egotistic to accept any ranking system that doesn't favor their work.

 

[TheEarl]

KM: Really? I thought it was every story (I know some other people who it happened to in the Celeb top list). It would have been about May-June kinda time.

The Earl