Support for Indie Authors discussion

I'm not sure this is in the right folder as it is not something for SIA to investigate, but I think it is worth knowing. It relates to what they call 'star factories' etc.

In my weekend newspaper there was an article about fake reviews - review factories, bots, and leaning on or tempting reviewers to give 5 star reviews or not at all. Some of this relates to book reviews.

I quote:

"Amazon gave the most comprehensive response to any suggestion of laxity (in policing fake reviews). 'Amazon invests significant resources to protect the integrity of reviews in our store...

" 'Even one inauthentic review is one too many.' It said that it uses machine learning to analyse all incoming reviews and regularly works with social networks to break up review groups. Since 2015, it said, it had brought lawsuits against more than 1,000 defendants for abusing the review system and in 2018 it spent more that £319m on stopping reviews abuse, fraud and other forms of misconduct."

Worth knowing that Amazon are defending those of us who don't have many reviews but they are all genuine.

Good to know.

Good to hear though it'd be nice if they acted just as strongly against book stuffing and other shady practices aimed at mining the KU prize pot.

Hmm... Tomas. A very good point.

I agree Amazon should be stringent, but they sometimes delete valid reviews. That is a problem and hurts authors.

Amazon does a good job. Do the other online sellers do anything? I don't know, just wondering.

I am glad Amazon is making an effort to police this, but some of us get caught in their net when we've done nothing to justify reviews being deleted.

I've also seen books with exactly 50 reviews that basically all say "this was a great book. I really liked it." but nothing actually specific about the story. Seems like too much of a coincidence to me to be a coincidence. It makes me sick when Amazon deletes a review that is Purchase Verified and gives good reasons why they liked the book. It really makes no sense. It's happened several times to me.

There's not much we can do about it, though.

Anna Faversham wrote: "I'm not sure this is in the right folder as it is not something for SIA to investigate, but I think it is worth knowing. It relates to what they call 'star factories' etc.

In my weekend newspaper ..."

We need a like button! Great post.

Thanks, Robert. There's a lot more that could be said on the subject of fake reviews and also on the unjust removal of true reviews. I know a prolific reader who had all her reviews removed. She fought back and they were reinstated.

But I feel grateful that Amazon are at least being more active than some.

Anna Faversham wrote: "But I feel grateful that Amazon are at least being more active than some."

I just wish they were more active here. I find it disconcerting that people can rate books without leaving any reviews whatsoever. Their shop, their rules, but it seems to me that it's a pretty easy way to game the system.

Robert wrote: "Their shop, their rules, but it seems to me that it's a pretty easy way to game the system."

Hi Robert. Do you think being able to leave a rating without a review leads to higher false ratings? Or, does the requirement on Amazon to have a review in order to rate lead to less ratings/reviews for authors?

Where is the risk perceived to be higher?

Eldon wrote: " Hi Robert. Do you think being able to leave a rating without a review leads to higher fa..."

That's hard to say. A while back, I did some comparisons between reviews of best sellers here and on Amazon. On Goodreads, the books have a lot more ratings than reviews, by a factor of 30. There are also about 4 times as many reviews of the same book here compared to Amazon. This makes sense, because best sellers can be obtained anywhere.

Among best sellers, the average rating at Amazon tends to be higher than it is here, but not by a lot. A better analysis would be a sampling of self-published books. The problem there is that they have far fewer reviews. For a reasonable statistical analysis, you need at least 30 random samples. You also have to look at the distributions (histograms) for conformity to a standard curve, which for customer ratings tends to be a skewed normal distribution weighted toward the higher ratings.

So I would want at least 30 randomly selected indie books with at least 30 reviews each to perform an analysis. I would compare the Amazon averages to the Goodreads averages with and without the reviewless ratings.

My god, who wants to do that?! I retired from an engineering job for a reason!

Now that that is out of the way, I'm back to speculation. I do know that Amazon moderates their reviews closely, whereas it's kind of wild and woolly here.

The issue is that anyone can leave a reviewless rating here, without having to show that they read any part of the book. It's thus a lot easier to game the system on Goodreads, but as to the ultimate effect this has, I don't know. For bestsellers, the averages on Amazon and Goodreads are statistically different, but not by a lot. For indie publications with far fewer ratings and reviews, I'm just taking an educated guess.

Robert wrote: "Eldon wrote: " Hi Robert. Do you think being able to leave a rating without a review leads to higher fa..."

That's hard to say. A while back, I did some comparisons between reviews of best sellers..."

Well, Robert, that is a well thought out and technical point of view. Since you lost me around histogram, I'll take your word for it lol :)

Haha, Eldon, I was going to say the same thing! Robert would be a great professor of statistics and probabilities!

Leah wrote: "Haha, Eldon, I was going to say the same thing! Robert would be a great professor of statistics and probabilities!"

Indeed Leah ;)

@Eldon, a histogram is just a bar chart. Every book has one. On Amazon, you see it when you scroll down to customer reviews. On Goodreads, it pops up when you click on the link that says "rating details." A normal distribution is the bell curve you probably know about. A skewed normal distribution has a bell shape but it looks distorted, like someone pushed it away from the center. In stats, we use something called a T distribution to approximate a normal distribution, as you are always working with a limited number of samples.

Robert, where were you when I took stats and probabilities?? You may have made that class more interesting. I may have even gotten higher than a C!

Leah wrote: "Robert, where were you when I took stats and probabilities?? You may have made that class more interesting. I may have even gotten higher than a C!"

It's getting hard for me to remember where I was when I was doing anything! I taught math to high schoolers in the mid-2000s, but that didn't last long. I loved the kids, but the lesson prep was murder.

Robert, that already happens to me and I just turned 40. lol

Leah wrote: "Robert, that already happens to me and I just turned 40. lol" 40 is nothing. I just turned 70. I wasn't even into bifocals until I turned 43. Wait until you're in your 50s and you start having senior moments.

But we digress. I am not sure what's under the hood of Amazon's process for weeding out specious reviews, but I'm certain statistics is at its heart. All AIs operate on probabilities. I just wish Goodreads would take a cue from their parent company and do something about the number of bogus ratings and reviews here. A bit of analysis makes them easy enough to spot.

Hi - So I'm a newbie. I just joined the Goodreads world about a month ago. Out of curiosity, how would one spot "bogus ratings and reviews"? And, what are you considering bogus? Like some bot creating them, or paid reviews? Really, just trying to educate myself here. Right now I'm aiming to self-publish my first book in January and am trying to better understand all the ins and outs.

Isabel wrote: "Hi - So I'm a newbie. I just joined the Goodreads world about a month ago. Out of curiosity, how would one spot "bogus ratings and reviews"? And, what are you considering bogus? Like some bot creat..."

Amazon does a pretty good job of weeding out phony reviews, but Goodreads doesn't moderate much of anything. I've seen quite a few "reviews" here where the reviewer did not read the book at all, but left a (typically bad) review based on the blurb or even just the title. Evidently, this is fine with Goodreads. You can read their reviewer guidelines to get an idea of what they deem acceptable, and it's not very encouraging for authors.

I've also seen reviews that don't leave any real text, e.g., the reviewer just copied and pasted parts of the book blurb or even plagiarized another review. Also, if you see something like "Great book, I liked it" as the only text, it's pretty suspicious. And for reviewless ratings -- there's no real way to tell if they're phony or not. If a flurry of them come in all at once, it's suspicious, but there is no way to prove anything.

Had I known about these things at the outset, I may not have set myself up as a Goodreads author at all. I don't want to discourage you from publishing your book -- by all means, do it! But I would be wary of listing it here until you've gotten enough reviews on Amazon or another site, one with actual moderation, to get an idea how well your book is really being received.