I'm sure Amazon has done studies on what the effect is.
It could be that products with fake positive reviews sell better individually but the presence of fake reviews in general reduces sales overall. But the former is really easy to measure and the latter is much harder.
Yep. It's easy to A/B test a narrow feature over a short timespan, but it's hard to test the multi-year effect of pollution -- by the time you learn the answer, you've already lost.
It could be that products with fake positive reviews sell better individually but the presence of fake reviews in general reduces sales overall. But the former is really easy to measure and the latter is much harder.