Hacker News Reader

Comment by BadCookie | Hacker News Reader

You'd think that Amazon would have some technology similar to FakeSpot's by now, but I'm not convinced that they do. I bought a product that had the special "Amazon's Choice" label on it that turned out to be a dud. I looked up the product on FakeSpot just now, and FakeSpot (correctly, I believe) identifies the reviews on that product to be unreliable (grade level "C"). Just a single anecdote, but it's clear to me that there's a ton of room for improvement on Amazon's part.

Another strange thing about the "Amazon's Choice" label is this: I've seen a product get that label in part due to a low number of returns ... but the seller doesn't accept returns! That should be a con, not a pro.

As you can probably tell, I have not been happy with the results of following Amazon's product choice advice.

View on HN

Plenty of the "Amazon's Choice" flags are...a little deceptive. Maybe not deceptive in a conventional sense, but say you see a product in the search results on Amazon and one has that flag. You click on it, because hey.

Then you notice the keywords attached to "Amazon's Choice". Oftentimes, they're strange and esoteric. Oftentimes, that flag is put on what appears to be a random off-brand item sitting right next to a much, much better product more deserving of that "Choice" insignia.

From what I've seen I'm guessing the amazon's choice tag is arbitrary and generated by some algorithm that simply matches up common search terms to the most-bought item. When it's something simple and common like "wireless mouse" it's probably alright. When it's a bit more esoteric like "male displayport socket soldered connector" or some other hopeless attempt to optimize search results and filter out noise, you get strange things.

Fakespot has lower negative impact if it mis-identifies a genuine product as grade level "C". But if Amazon does that you will certainly see people publishing stories about how their genuine product has been mis-identified by the giant called Amazon. For example today's submission:

https://bardsandsages.com/juliedawson/2018/05/06/amazon-froz...

It will be like de-monization (of YT videos) all over again.

I would imagine any sort of product review classification system would be exploitable for abuse.

Anecdotal evidence: when my mother and her team were running for student council (in college), the opposing team introduced blatantly fake votes (photocopied) in favor of her team; my mother's team got disqualified for cheating, and the opposing team won.

So I could certainly picture a way in which a mischievous seller could submit fake-looking reviews for another product and get it taken down.

There’s nothing better than the support from the college administration of such behaviour by directly acting upon it. Some of these people now sit somewhere higher up.

Part of the problem for Amazon introducing it is that it could be seen as vouching for reviews that were not detected as fake, even though they may still be fake. Unless the false negative rate is sufficiently low, it may give false confidence in products.

> but the seller doesn't accept returns! That should be a con, not a pro.

US? In the UK, the seller has to accept returns, so that removes one variable (and if the product is bad, you can just send it back).

I could see why they don't jump into the review verification side though. If they acquired one of these apps or built their own and another fake review comparison engine outperformed Amazon's, they'd have to deal with the backlash of why theirs isn't good enough.

It feels a lot like Consumer Reports to me, where the reviews of products are handled separately from the vendor.

Reply on news.ycombinator.com