Should the sites like Scribble Hub use the 5-star rating system?

[RepresentingWrath]

In that case, I have a new idea.

A plutocratic hellish Star Duel system

Users have a bank of stars 5 out of 5
if they give a novel 1 star they will have 4 out of 5 stars left
and only one star gets regenerated per week
if a user wants to give something 5 stars they will use up all their star stockpile

the only way to give out more stars is to pay a microtransaction to Tony to "top up" your stars
you can also pay to increase your star capacity, so you can stockpile more stars

Another way to get stars is to get another user to give you their stars voluntarily....

OR!

A "Star Duel"
You can challenge another user to a "Star Duel"
in which the winner will bet and get the loser's stars
in a duel using dice and cards.

there is also an option for a
"There could be only one" Highlander-style duel
that you go all in and take over the other person's account.
And if you challenge an author to a "Highlanders Star duel", you get all their novels as well
and you are now forced to write all their novels forever.

If you want to give a rating but author doesn't consent you have to win against the author in Yu-Gi-Oh.

 

[owotrucked]

In that case, I have a new idea.

A plutocratic hellish Star Duel system

Users have a bank of stars 5 out of 5
if they give a novel 1 star they will have 4 out of 5 stars left
and only one star gets regenerated per week
if a user wants to give something 5 stars they will use up all their star stockpile

the only way to give out more stars is to pay a microtransaction to Tony to "top up" your stars
you can also pay to increase your star capacity, so you can stockpile more stars

Another way to get stars is to get another user to give you their stars voluntarily....

OR!

A "Star Duel"
You can challenge another user to a "Star Duel"
in which the winner will bet and get the loser's stars
in a duel using dice and cards.

there is also an option for a
"There could be only one" Highlander-style duel
that you go all in and take over the other person's account.
And if you challenge an author to a "Highlanders Star duel", you get all their novels as well
and you are now forced to write all their novels forever.

If you want to give a rating but author doesn't consent you have to win against the author in Yu-Gi-Oh.

If the author wins, then the reader receive that rating instead

 

[beast_regards]

Duel system

Illegal in Japan.

You couldn't carry the long scissors.

You couldn't challenge people to duels.

I believe you couldn't be mean to people on the Internet either.

...and you certainly can't challenge people on the Internet to the duel using scissors.

 

[Tyranomaster]

How about profiling readers by placing them on a latent space according to the ratings they give to different genre and lumping them into groups. Then each book receive multiple ratings weighted according different groups. That way you could have "NTR enjoyers and cucks give this book 4.7 star, but everyone else rate it 1.24" or something

The approximate order of operations for *Patent Pending* (JK) "TruescoreSR* (Truescore Story Rating) is as follows for any new score being put up:

1. Adjust Author score based on their nominal score range - Examples: If they have 7 reviews all in the range of 1-3, adjust scores to be 1-5. If all reviews are the same score, they all default to that score, and a heavy weighting penalty is applied (maybe adjust weight down by 50% to start).
   
   
2. Apply Genre tag average weighting to rating - Examples: NTR underperforms the average site rating across all NTR stories, as such, scores are adjusted upwards somewhat. Isekai overperforms, and is lowered slightly. This is averaged with a weighting taken by reverse calculating averages across stories every so often (perhaps once a month).
   
   
3. Edit: Apply an account score weighting based on Genre accuracy as well, some people are better at rating certain genres than others after all (perhaps they keep reading things they hate, idk)
   
   
4. Apply review count weighting accordingly - Set arbitrary review limit to reach full weighting, and adjust accordingly (lets say 10 ratings gets you to full weight on this variable, and you get 10% more weighting for each rating until reaching it)
   
   
5. Apply account wide accuracy rating - Arbitrary variable multiplies your rating accuracy vs the average rating. When you're off by too much on multiple stories it applies a small penalty to your account's ratings in the future. This is also weighted based only on stories that have Green Quality Truescore ratings (which indicates a 90% certainty in score based on types of reviews and total count.)
   
   
6. Periodically adjust and tune parameters provided in these different variable spaces until all numbers work well across genres. With multiple weightings and variables applied on both account and genre levels, it should be easily possible to fine-tune the numbers to encourage good review practices.

Truescore accuracy would be determined based on the accuracy rating of the accounts that are giving reviews (based on their downweight % versus maximum), and averaged with an inverse rule to bring accuracy up with more account reviews.

That's the gist of my idea.

 

[Valmond]

Something worth considering about a like and dislike system is that it takes longer to converge on a meaningful value, so stories with fewer ratings just have meaningless ratings for longer. This happens with steam games too. It takes longer to gain traction. That means more effort on advertising, though I still think it's probably better than 1-5

Yeah, it does.

 

[RepresentingWrath]

If the author wins, then the reader receive that rating instead

This is fucking genius idea! I don't remember who it was, uber? Whatever, there was this story of a taxi service making another rating. Drivers could rate customers. We should do it on SH! Authors can now rate readers!

 

[Valmond]

If you want to give a rating but author doesn't consent you have to win against the author in Yu-Gi-Oh.

Better get out that OTK deck again.

 

[Deleted member 266]

We should do it on SH! Authors can now rate readers!

Yes!

That is why we workshop ideas!!!

 

[Prince_Azmiran_Myrian]

This is fucking genius idea! I don't remember who it was, uber? Whatever, there was this story of a taxi service making another rating. Drivers could rate customers. We should do it on SH! Authors can now rate readers!

Rate commenters, too

 

[unlaumy]

In fact we should add the scale to 10-star, or even 100-score while we're at it

 

[Valmond]

Onlookers: Why are there only good ratings?

Insider: It is because everyone was banished to the Shadow Realm.

 

[Tyranomaster]

The approximate order of operations for *Patent Pending* (JK) "TruescoreSR* (Truescore Story Rating) is as follows for any new score being put up:

1. Adjust Author score based on their nominal score range - Examples: If they have 7 reviews all in the range of 1-3, adjust scores to be 1-5. If all reviews are the same score, they all default to that score, and a heavy weighting penalty is applied (maybe adjust weight down by 50% to start).
   
   
2. Apply Genre tag average weighting to rating - Examples: NTR underperforms the average site rating across all NTR stories, as such, scores are adjusted upwards somewhat. Isekai overperforms, and is lowered slightly. This is averaged with a weighting taken by reverse calculating averages across stories every so often (perhaps once a month).
   
   
3. Apply review count weighting accordingly - Set arbitrary review limit to reach full weighting, and adjust accordingly (lets say 10 ratings gets you to full weight on this variable, and you get 10% more weighting for each rating until reaching it)
   
   
4. Apply account wide accuracy rating - Arbitrary variable multiplies your rating accuracy vs the average rating. When you're off by too much on multiple stories it applies a small penalty to your account's ratings in the future. This is also weighted based only on stories that have Green Quality Truescore ratings (which indicates a 90% certainty in score based on types of reviews and total count.)
   
   
5. Periodically adjust and tune parameters provided in these different variable spaces until all numbers work well across genres. With multiple weightings and variables applied on both account and genre levels, it should be easily possible to fine-tune the numbers to encourage good review practices.

Truescore accuracy would be determined based on the accuracy rating of the accounts that are giving reviews (based on their downweight % versus maximum), and averaged with an inverse rule to bring accuracy up with more account reviews.

That's the gist of my idea.

Oh, forgot to say that the non-truescore rating should also still be visible (and it's used on the backend anyway, so it has to stick around.

 

[RepresentingWrath]

Yes!

That is why we workshop ideas!!!

Rate commenters, too

In fact we should add the scale to 10-star, or even 100-score while we're at it

Yeah! Rate those readers and commenters on a scale of 0 to 100.

 

[Prince_Azmiran_Myrian]

In fact we should add the scale to 10-star, or even 100-score while we're at it

The scale should also be non-linear, so that the common gathering points would be 13, 43, 68, and 96. Just to annoy people's OCD.

 

[RepresentingWrath]

The scale should also be non-linear, so that the common gathering points would be 13, 43, 68, and 96. Just to annoy people's OCD.

Don't forget spelling mistakes!

 

[unlaumy]

The scale should also be non-linear, so that the common gathering points would be 13, 43, 68, and 96. Just to annoy people's OCD.

Exactly. We should make the rating system like the academic grading system. We can also add @Tyranomaster 's idea:

Apply Genre tag average weighting to rating - Examples: NTR underperforms the average site rating across all NTR stories, as such, scores are adjusted upwards somewhat. Isekai overperforms, and is lowered slightly. This is averaged with a weighting taken by reverse calculating averages across stories every so often (perhaps once a month).

Every genre or tag has a weighted neutral/passing score. Because NTR is opressed in this site, their neutral score should be around 20-30. And so on and so on.

 

[CharlesEBrown]

In fact we should add the scale to 10-star, or even 100-score while we're at it

-3 to 17. With five being the best.

 

[beast_regards]

Yeah, it does.

This is because you could review only the games you have purchased.

Even on Amazon, you could review the products you purchased.

However, you could start reading any story on the Scribble Hub anytime you wish.

I didn't even encounter stubbed stories, not to mention there isn't the ranking list that forces to notice the novels in certain order (there is ranked per tag, yes, but it doesn't order stories unless you try really really hard, rather than being default more of view)

If the Scribble Hub worked on the Steam ranking model, the stories would stabilize on the certain value relatively quickly.

Not to mention, the Scribble Hub algorithm isn't rating based as on the Roadkill Route, which means all rating would be more informative, and clearly explained, instead of being 90% anonymous.

 

[Anonjohn20]

This is because you could review only the games you have purchased.

But they sometimes hide reviews from players who acquired the game through a key (I saw my Helldivers 2 review get censored) and players who got a game for free (either from a giveaway or as a gift from a friend).

Even on Amazon, you could review the products you purchased.

This is not true; I've reviewed products that I bought elsewhere, so Amazon had no way of knowing if I owned the product.

However, you could start reading any story on the Scribble Hub anytime you wish.

I assume you meant rating, not reading, but it's also not true; it needs to be on your reading list.

all rating would be more informative, and clearly explained, instead of being 90% anonymous.

That would lead to more vindictive authors. They are already whining that they want negative reviews removed; imagine if they started whining that negative reviewers should be banned.

 

[Tyranomaster]

If the Scribble Hub worked on the Steam ranking model, the stories would stabilize on the certain value relatively quickly.

Not to mention, the Scribble Hub algorithm isn't rating based as on the Roadkill Route, which means all rating would be more informative, and clearly explained, instead of being 90% anonymous.

Because there is no price penalty to a review though, this would also open SH ratings up to easier manipulation by bad actors. With the ability to simply paraphrase a requested bad review from GPT with a link, people can write *informed* bad reviews without any anonymous positive ratings to counteract.

Non-reporting bias would be heavily increased if reviews were mandatory, concentrating rating power in fewer hands. I don't think it's apparent that switching to this style of system would be an improvement at all.

On the contrary, I think the majority of proper reviews on stories are actually lower rated than the stories themselves (after a threshold), because people are most motivated by hate rather than appreciation. I think more authors would go underwater with this change than not regardless of malicious intent.