Where's the logic?

[SamScribble]

Even though my ears are not Spock-shaped, I have often been accused of being uncommonly logical.

There have been times in my life when this has been useful. On one occasion, my logic was instrumental in convincing ten fellow jurors of the innocence of a chap who, only an hour or so earlier, they had been about to find guilty of murder. It was heartwarming when, on delivering our ‘Not guilty’ verdict, the judge said: ‘The decision has been yours. But I must say that I agree with it 100 percent. It seems to me completely illogical to find the defendant anything other than not guilty.’

Which brings me to the score for my entry into the National Nude Day contest. The story began with a score of 4.5-something, rose to 4.8-something, fell to just under the ‘magic’ 4.5, and then it rose again. At this point it had 50-something votes. And then it started to slide. Today, at 100 votes, it is languishing (a nice word - languishing) at 4.30.

I have no illusions about my quirky stories winning a jerk-off contest. But if the vast majority of the first 50 voters thought that it was worthy of five stars, how can the next 50 have such a low opinion? It defies logic. Just sayin’.

 

[ElectricBlue]

I have no illusions about my quirky stories winning a jerk-off contest. But if the vast majority of the first 50 voters thought that it was worthy of five stars, how can the next 50 have such a low opinion? It defies logic. Just sayin’.

I put it down to the law of averages. If you think about it, half the people you meet will be below average. You were fortunate enough to get the other half first, is all. It's probably a cyclic thing, something to do with time zones and when the below the line folk wake up.

As we all know, once you've got through the daily ups and downs enough times the maths begins to work properly. Pay no attention for a fortnight, is what I always try to do.

 

[RejectReality]

Or, as George Carlin said: Think about how stupid the average person is. Now realize, half of them are stupider than that!

Seriously, at only 100 votes, your score is likely to be a ping-pong ball in a contest. There are idiots out there doing everything in their power to muck with the scores.

Then it all washes out in the sweeps.

As often as not, trashing a story with a low vote total ends up having an opposite effect. Some troll blasts you down. The rest of the trolls don't consider you a threat any longer, and focus their attention elsewhere.

They end up hiding you from their fellow scumbags, and fewer people see your lower-scored story, resulting in fewer "liked it, didn't love it" 4 votes. The sweeps run, and suddenly you're high in the standings, and it's too late to kill you again before the results are tallied.

I put it down to the law of averages. If you think about it, half the people you meet will be below average. You were fortunate enough to get the other half first, is all. It's probably a cyclic thing, something to do with time zones and when the below the line folk wake up.

As we all know, once you've got through the daily ups and downs enough times the maths begins to work properly. Pay no attention for a fortnight, is what I always try to do.

 

[Oblimo]

There are motives at work with Lit star ratings beyond a simple, “How much did I like this story?”

There are some readers who give high averaging/ranking stories low scores, just ‘cause.

Also, the scoring survey (like most on the Web) is not statistically valid/fair or scientific, nor is it meant to be. Expecting it to behave logically is, well, illogical.

 

[Requiax]

There are motives at work with Lit star ratings beyond a simple, “How much did I like this story?”

There are some readers who give high averaging/ranking stories low scores, just ‘cause.

Also, the scoring survey (like most on the Web) is not statistically valid/fair or scientific, nor is it meant to be. Expecting it to behave logically is, well, illogical.

There’s also another type of logic which doesn’t often get considered. A lot of readers will pick to read a story which is rated highly. They reason, if other people liked this, I will too. If they then read it and find it isn’t what they expected based on the generous rating, they may feel a need to “correct” that rating and so give a lower mark than they might otherwise have given a “good, not great, but good” story they happened on without preconceptions.

In other words, the rating affects perception of the story which affects reader behaviour which affects the rating.

I wonder what would be the difference if Lit didn’t publicly show the rating for 48 hours or so after a story gets published. Not that I think this would improve or worsen things, but I’d be interested to see if reader behaviour changed...

 

[ElectricBlue]

I wonder what would be the difference if Lit didn’t publicly show the rating for 48 hours or so after a story gets published. Not that I think this would improve or worsen things, but I’d be interested to see if reader behaviour changed...

How would stories get traction? All a delay would do, as I see it, is bury more stories down in the noise - they'd fall of the front page with no "ranking". Flawed as it is, the scoring system sorta kinda works as a filter.

 

[Requiax]

How would stories get traction? All a delay would do, as I see it, is bury more stories down in the noise - they'd fall of the front page with no "ranking". Flawed as it is, the scoring system sorta kinda works as a filter.

Which is why I didn’t say it would be something I thought would make things better, more an experiment to see if people’s behaviour changed.

Unlike many here I have no particular beef with the ratings system on Lit.

But the psychology of it is interesting to me and if I controlled things the temptation to experiment would be hard to resist.

 

[Hypoxia]

I love different logics. We can build whatever logics we want. 1+1=3 for large enough values of 1. We can replace yes-no / true false binary logic with my 31-fold expansion of Jain 7-fold logic. (Any statement is maybe true and maybe false and maybe unknowable, and maybe trivial and maybe irrelevant.) Or as a learned rebbe once said, "Bullshit is bullshit but the study of bullshit is scholarship." Scores here work on their own logic. Oy.

 

[MelissaBaby]

There’s also another type of logic which doesn’t often get considered. A lot of readers will pick to read a story which is rated highly. They reason, if other people liked this, I will too. If they then read it and find it isn’t what they expected based on the generous rating, they may feel a need to “correct” that rating and so give a lower mark than they might otherwise have given a “good, not great, but good” story they happened on without preconceptions.

I think there is some truth to that. My stories typically run around 4.80-4.90 early on, drop to about 4.40-4.60, then climb to 4.70-4.80.

I think one reason why my stories start high, and this may be a factor in Sam's situation, is that followers are likely to make up a higher percentage of those first voters.

 

[lovecraft68]

I'd say the large swings in the first 50 votes are due to the fact that at fifty votes each 1/5 has a greater effect. At 100 votes there is less effect per vote so it will tend to stabilize a bit.

There's also a matter of the sweeps which in my opinion are no longer valid. I think they cut far to deeply now and seem to 'think' hardly any votes less than 4/5 are real. The huge jumps in the final sweeps and ending scores in the 4.9's were things you didn't see a few years.

Either the sweep has changed or people have gotten better at manipulating it.

I think past events with certain authors-and not just Scouries and BFW who are lit's resident boogie men should show anyone who didn't already know it that these contests are so easily screwed with that its hard to see them as anything close to legit.

That's not me saying no winners are deserving winners, but just like people are quick to point out a low score in LW isn't indicative of the quality of a story due to the trolls there I think the same of the contests a 4.9 may be a 'how the hell is this scored so high' and a 4.2 might be 'this is way better than that score'

I've fallen into that camp of if you enter just do it with the thought of the extra attention, although again like LW some of it will be bad.

Shady authors and rabid fan bases along with suspect sweeps have me unable to see these things as being any type of real contest.

 

[xelliebabex]

Even though my ears are not Spock-shaped, I have often been accused of being uncommonly logical.

There have been times in my life when this has been useful. On one occasion, my logic was instrumental in convincing ten fellow jurors of the innocence of a chap who, only an hour or so earlier, they had been about to find guilty of murder. It was heartwarming when, on delivering our ‘Not guilty’ verdict, the judge said: ‘The decision has been yours. But I must say that I agree with it 100 percent. It seems to me completely illogical to find the defendant anything other than not guilty.’

Which brings me to the score for my entry into the National Nude Day contest. The story began with a score of 4.5-something, rose to 4.8-something, fell to just under the ‘magic’ 4.5, and then it rose again. At this point it had 50-something votes. And then it started to slide. Today, at 100 votes, it is languishing (a nice word - languishing) at 4.30.

I have no illusions about my quirky stories winning a jerk-off contest. But if the vast majority of the first 50 voters thought that it was worthy of five stars, how can the next 50 have such a low opinion? It defies logic. Just sayin’.

I think (going by the contest thread) that a lot of authors have found strangeness going on this time around with the contest voting. I think the sweeps will be interesting when they come.

 

[FEELINGLUCKYPUNK]

The logic is LAUREL lets you score for any or no reason.

 

[Tio_Narratore]

I put it down to the law of averages. If you think about it, half the people you meet will be below average. You were fortunate enough to get the other half first, is all. It's probably a cyclic thing, something to do with time zones and when the below the line folk wake up.

As we all know, once you've got through the daily ups and downs enough times the maths begins to work properly. Pay no attention for a fortnight, is what I always try to do.

Not quite the law of averages. What you describe is more so the "law of medians," but that law is a tautology.

Only the median has the same number of units below as above; to the extent that the mean approaches the median, it too will split 50-50. But consider some real cases, such as the average salary in a business, no matter how large a business. There are a few - the suits - who make really high salaries, and then the workers, who makes far less. The average salary for the company will fall between the averages for the execs and the peons, and thus far more than half the salaried employees will be below average.

 

[Bramblethorn]

People vastly over-interpret score fluctuations. 100 votes is not very much and it's quite possible for a story to have major changes in the first 100-odd votes, just by chance.

Here's one I posted a few months back. I ran a computer simulation with six stories. For the first three, every vote has an 80% chance of being a 5 and 20% of being a 4, so the long-term average is 4.8; for the other three, it's 50% each for 4s and 5s, so long-term average 4.5.

Look at the green one. For the first twenty-odd votes, it's scoring about 4.7, and then suddenly it drops, and keeps dropping, down to about 4.45 at 100 votes. I could easily make up a story about that: this was a great story, earning 4.7 from the voters, until some troll noticed and started hammering it and pulled it out of the 'H' zone. But I programmed that simulation and I know there weren't any trolls. All that happened here was that it got lucky in the first 20-odd votes, scoring higher than average, and then the luck ran out and it drifted back towards its long-term average.

 

[SimonDoom]

That voting pattern suggests to me you have been deliberately downvoted by a few voters, perhaps for tactical reasons. It may be corrected with sweeps.

 

[GrandTeton]

Not quite the law of averages. What you describe is more so the "law of medians," but that law is a tautology.

Only the median has the same number of units below as above; to the extent that the mean approaches the median, it too will split 50-50. But consider some real cases, such as the average salary in a business, no matter how large a business. There are a few - the suits - who make really high salaries, and then the workers, who makes far less. The average salary for the company will fall between the averages for the execs and the peons, and thus far more than half the salaried employees will be below average.

In the example, consider what peons get as one-bombs, dragging down the average salary so what the executives get doen't seem quite so obviously obscene in comparison. The problem, which is not readily resolvable when anonymous voting is permitted (encouraged), is that the scoring system is a simple average of total points divided by total votes. A one or two is of little relevance when the story is worth, say, 3.00, but assumes huge proportions when the story is worth 4.95 or the like. Also, as has been pointed out, the fewer the votes the more influence each one has. It takes more than one or two one-bombs to seriously affect the score of a story with a couple thousand votes.

I'm amazed by some of the comments that suggest readers don't understand this. People who refuse to vote for a story they hate in the mistaken belief that something akin to a zero will be scored, when just going away is the best the author can hope for. Then there are the arithmetical illiterates who will consciously give a three or a four because the story is at least worth a vote, and so many are not, when the author would much prefer that you go away and not treat his sort of worthwhile story worse than the ones you can't even be bothered to read.

The surprise is that the scoring system works as well as it does, particularly given the relative scarcity of votes.

 

[lovecraft68]

People vastly over-interpret score fluctuations. 100 votes is not very much and it's quite possible for a story to have major changes in the first 100-odd votes, just by chance.

Here's one I posted a few months back. I ran a computer simulation with six stories. For the first three, every vote has an 80% chance of being a 5 and 20% of being a 4, so the long-term average is 4.8; for the other three, it's 50% each for 4s and 5s, so long-term average 4.5.

Look at the green one. For the first twenty-odd votes, it's scoring about 4.7, and then suddenly it drops, and keeps dropping, down to about 4.45 at 100 votes. I could easily make up a story about that: this was a great story, earning 4.7 from the voters, until some troll noticed and started hammering it and pulled it out of the 'H' zone. But I programmed that simulation and I know there weren't any trolls. All that happened here was that it got lucky in the first 20-odd votes, scoring higher than average, and then the luck ran out and it drifted back towards its long-term average.

Wow, you take this stuff seriously. I would never have the patience to spend that much time on it.

Reject Reality has charts and graphs and who knows what for his.

 

[NotWise]

There’s also another type of logic which doesn’t often get considered. A lot of readers will pick to read a story which is rated highly. They reason, if other people liked this, I will too. If they then read it and find it isn’t what they expected based on the generous rating, they may feel a need to “correct” that rating and so give a lower mark than they might otherwise have given a “good, not great, but good” story they happened on without preconceptions.

I suspect this sort of thing does happen. I posted a story to EC last winter that started low and popped up over 4.5 for a few hours before diving again. Each time thereafter until the story disappeared from the EC hub, each time it went over 4.5 it was down voted again. These were mostly 4* (sometimes 3*) votes. It happened maybe six times over the course of two days.

I don't think there was any trolling going on. The way I saw it, the readers got more critical as soon as the red H appeared and had a bigger tendency to down-vote the story. The score went over 4.5 as soon as it dropped off the EC hub, and it stayed there.

 

[RejectReality]

Haven't updated them in years. My listings were simply getting too long to deal with.

Of course, now we have a one-click download option, including favorites. ( Yay! )

Eventually, I'll redesign my template to accommodate the imported data, and probably start tracking again, at least annually.

Wow, you take this stuff seriously. I would never have the patience to spend that much time on it.

Reject Reality has charts and graphs and who knows what for his.

 

[Alice_Rosaleen]

I try not to fret over my score, even/especially if there's a contest involved. But I happen to think the five star system is bogus because if you consider it realistically, if someone stays to the end they're probably going to either give it a five or a one. Maybe a four if they're being critical that day or two if they're feeling nice, but who's really going to stay to the end of a solid three story (in their opinion)? Either they loved it so much they can't wait to vote or they hate it that much (or the author). Just my opinion, and like most it doesn't change anything, but that's why I can't take it seriously.

 

[Oblimo]

Let’s all figure out how to apply Bayesian inference to predicting the accuracy of smut Lit story ratings!

...No? ‘Kay.

 

[Bramblethorn]

Wow, you take this stuff seriously. I would never have the patience to spend that much time on it.

Reject Reality has charts and graphs and who knows what for his.

I do too, because I love that stuff. The simulation was pretty basic, less time to program than to talk about the results.

 

[Bramblethorn]

Let’s all figure out how to apply Bayesian inference to predicting the accuracy of smut Lit story ratings!

...No? ‘Kay.

Don't think I haven't thought about it

 

[Hypoxia]

It's futile to expect logical moves in LIT scores. The audience is semi-random.

 

[FEELINGLUCKYPUNK]

Recall Reuben won AMERICAN IDOL then vanished forever.