Buying from a “fulfilled by Amazon” trusted vendor won’t necessarily help either, because Amazon also bins all products with a given UPC into a single bucket, no matter who the seller is.
7 months ago, 624 comments were posted in a discussion about Amazon returns neither being painless nor simple. My favorite is this write up noting that returns are only processed if your predicted future revenue exceeds the return requested: https://news.ycombinator.com/item?id=27333273
But in general, Amazon won’t honor their purchase protection policy for high value items without extensive argument and delaying tactics, which puts shoppers at risk for precisely the purchased that most need protection.
So, I opened up the OpenAI playground to see if GPT-3 could do better. I prompted: "Q: Is a product called "KXZM 24V LED Strip Light Natural White 600LEDs, 16.4ft Flexible SMD2835 High Brightness 4000-4500K No-Waterproof IP33 LED Tape Lights" waterproof? A:" and it answered "It's not. As you can see, this product is not waterproof. It says so in the product description." (GPT-3 is always a bit mean to me, but in this case, I obviously deserved it. What else could "no-waterproof" mean, ya idiot ;)
My conclusion is that GPT-3 could easily filter out products that don't actually match a search query, and prevent disreputable sellers from abusing naive keyword matching, making the shopping experience better for everyone. And, Amazon obviously has the talent and compute resources to build their own model like GPT-3. But they don't do it! They don't care if you buy the wrong thing or that the search results are irrelevant. As long as you click something and buy it, they make money.
I don't know how this works out long term, but I'm guessing that people don't notice they're not waterproof, and Chinese Factory A that did keyword optimization makes more money than Chinese Factory B that sprayed the LEDs with a conformal coating. (If the return rate gets too high, you just pick 8 more random all-caps letters and start a new company.)
Simple use of an API could solve this problem in seconds. Customers would be happier. We wouldn't be rewarding manipulation. It's easy. And yet, they don't. They just "select * from products where keywords like '%$1'".
My conclusion is that Amazon simply doesn't care. They're #1, and you can't get better than #1, right? If they kill a few people in a pandemic with counterfeit masks, who cares? #2 DREAMS they were killing people in a pandemic!
If Congress doesn't act, plaintiffs' attorneys will. It'll be the McDonald's coffee case all over again. They're not just asking for it, they're begging.
Until that happens, they simply DGAF. As you say, they're #1, and in Amazon-world, that means you stop trying.
Let's be clear about something
GPT-3 is state of the art. It is really good. It cost A LOT to develop.
I don't think Amazon replicating it is trivial.
I'm pretty sure I've gotten counterfeit GE light bulbs from them. It was an appliance bulb, packaging looked right, the bulb died within a day. I found another at a local grocery store, same packaging, slightly different tint to the glass, and it's been going for years now (in an oven that I frequently run at 550 F).
f3 - Fight Flash Fraud¶
f3 is a simple tool that tests flash cards capacity and performance to see if
they live up to claimed specifications. It fills the device with pseudorandom
data and then checks if it returns the same on reading.
F3 stands for Fight Flash Fraud, or Fight Fake Flash.
1) If anything, the company leadership has demonstrated their amorality when it comes to exchanging dollars for things people want
2) The company leadership doesn't feel the need to be particularly hand-holdy to a public that is still running them dry on Ivermectin as fast as it comes in.
... but there's nothing really stopping the government from saying "If you're going to sell medical supplies you must be regulated as a medical supplier" other than lack of will.