why doesn't Amazon do something about fake reviews?

[thelurkinghorror]

@CookingWithGas

I got an ad on Facebook for a knife. Long story short, it’s a decent but mediocre knife with a price higher than it’s worth and good marketing. I read a review on that knife in Wired, and they said right out of the box it’s as good as knives costing four times as much, but he consulted with two experts in knifemaking and they ran hardness tests. They said the real test of a knife is how well it holds up after a year, after you have to sharpen it and so on. The Amazon reviews were generally glowing–because people reviewed it the first week they bought it. They still hadn’t found out that it gets harder and harder to keep an edge on it.

JD Power’s business model. “This car is highest ranked in *initial *quality.”

[DrDeth]

@Cheesesteak

I was looking for some phone accessories recently and found a whole class of items that routinely had ~100 fake 5 star reviews. All reviews posted within a week or two of each other, all 5 star, any new reviews were 1-2 stars complaining about the product being worthless garbage. I needed a whole new filter for judging the reviews.

How do you know they are fake? Were they verified purchases?

[race_to_the_bottom]

@Chronos

Personally, the way I prefer to use reviews is to look at the bad reviews (there are bound to be some), and see what’s bad about them. If they’re just about trivial things that I don’t care about (or which I do care about, but have decided are acceptable tradeoffs), then it’s a good product.

I pay attention to well written negative reviews.

[Jragon]

It’s fundamentally the same problem Youtube, Twitter, and Facebook face to varying degrees.

Can you imagine the sheer infrastructure that would be necessary to audit every slapfight on Twitter to properly verify who in a given thread is running afoul of the rules? You’d need to have staff familiar with almost every type of argument that can occur on the site, not to mention in many cases the tweet histories of the parties involved to properly moderate disputes and sanitize content. Not to mention, even if you hire the 15k people needed to manually screen all this content, you need to somehow enforce a company-wide code of ethics on them to determine what sort of thing should or should not be moderated in a certain way. Same for Facebook.

This is also an issue with Youtube and false flags and copyright claims, or spurious TOS violation flags.

The way these sites have elected to combat this problem is some combination of throwing up their hands and yelling “fuck it”, and algorithmic moderation with minimal human oversight. The algorithmic moderation in particular is known for doing some absolutely daft, arguably actively socially harmful things (like that one time on Youtube when it was flagging LGBT content or a dozen other cases I could mention).

Amazon has similar problems where it has a mind boggling number of items, and a mind boggling number of user reviews submitted every hour. And, sure, perhaps you can spot a lot of them from a mile away, but I bet you can’t for a large percentage (perhaps more than you suspect), especially if you’re not personally familiar with that particular item.

It’s a hard problem, and I don’t blame them for not solving it. I do feel like some accountability on the part of these massive social media sites and storefronts is needed, but it’s also not really a simple problem to solve. As mentioned they have some measures in place, but throwing all the IP filtering and algorithmic pattern recognition in the world at it isn’t going to make more than a large dent in the problem. The only thing that could is a whitelist system like verified reviewers, and if you make that too hard to achieve you’ve arguably defeated the purpose of a “user review” system and just made it an “Amazon company-approved reviewer” system. Which isn’t terrible… but let’s just say there’s good reason to maybe be a bit suspicious of reviewers the people trying to sell you something explicitly endorse.

[Cheesesteak]

@DrDeth

How do you know they are fake? Were they verified purchases?

I’m not sure if they were verified, but they were all glowing reviews, all submitted within a recent two week period, and any review submitted after that period described a product that flat out didn’t work.

It’s an iPhone adapter, it wasn’t invented a month ago, so why aren’t there reviews older than 3 weeks?

So, to answer your question, I don’t know they were fake, but the situation fairly well screamed fake.