Comment by alehul | Hacker News Reader

Comment by alehul | Hacker News Reader

Parent Full thread

> we also ensured that 99.9% of all pages viewed by our customers were to products that had never received a counterfeit notice of infringement

While this initially seems impressive (i.e. 1 in 1,000 products are fake), Amazon has a power-law distribution to its sales, and thus likely to its views [1] [2].

This means that, as long as the products on the left side of the power-law aren't counterfeit, this claim can be true.

As a result, if you venture into niche product spaces, beware that you're looking at much higher odds of a counterfeit.

[1] http://ebusiness.mit.edu/research/papers/2010.09_Brynjolfsso... [2] https://www.marketplacepulse.com/articles/marketplaces-power...

Also, absence of a counterfeit notice does not mean the product is necessarily genuine. It seems to me that actively preventing counterfeits would entail much more than responding to notices.

Also whitewashing as it hides the scale of the problem - even if you do everything to make sure you get an original product, Amazon might still send you a fake or opened & repackaged item due to their mixing own & other vendors' stock.

Really appalling all the way and nonetheless profitable - so Amazon will not change until they get sued for billions.

> As a result, if you venture into niche product spaces, beware that you're looking at much higher odds of a counterfeit.

Yes, very good point. Stating by views is tricky. What percentage of listings? Huge numbers of items there are in the long tail, that's one of the claimed advantages of amazon, availability of weird things, and it's great. Is it 90% of all listings are in the long tail? And presumably nearly all fakes are there since high profile fakes would be discovered quickly. If 1% of views are rare long tail items and all fakes are there and .1% of views are fakes, then perhaps around 10% of long tail items are fakes. Of course these numbers are speculative, but amazon could answer many questions by disclosing more stats than % of views.

Reply on news.ycombinator.com