After Amazon clamped down on paid reviews in 2016, the community dispersed but still flourishes, with most comments on some popular categories still fake
Washington Post : Tweets: @theharryshearer , @aaschapiro , @noupside , @jbenton , @lizzadwoskin , and @lizzadwoskin Tweets: Harry Shearer / @theharryshearer : “These days it is very hard to sell anything on Amazon if you play fairly,” So much disruption, so little time.... http://www.washingtonpost.com/ ... Avi Asher-Schapiro / @aaschapiro : So the @washingtonpost did an investigation into Amazon product reviews and, although the company has deployed AI to purge fakes, surprise surprise 2/3 of reviews for many popular products were..probably fakes http://www.washingtonpost.com/ ... Renee DiResta / @noupside : 1/ Work I did was covered by the @washingtonpost today. http://www.washingtonpost.com/ ... I'm interested in the incentive structures and unintended consequences of Amazon's global marketplace strategy. My research started not with the goal of catching bad Facebook groups... Joshua Benton / @jbenton : I would've loved to have heard the discussion among Post editors over where to put the Bezos disclaimer in this story (22nd graf) http://www.washingtonpost.com/ ... Elizabeth Dwoskin / @lizzadwoskin : For months, @noUpside lived inside the Facebook groups that exist to defraud Amazon. FB banned more than a dozen groups after our inquiry. http://twitter.com/... Elizabeth Dwoskin / @lizzadwoskin : @craigtimberg and I have an investigation into the not-so-underground economy that drives massive #'s of fake reviews on Amazon (Surprise: It lives on Facebook). More than 1/2 of all reviews for many popular products showed signs of being inauthentic. http://www.washingtonpost.com/ ...
Context & Ripple Effects
Amazon's 2016 crackdown on paid reviews scattered the brokers rather than shutting them down: the Washington Post investigation finds that despite the company deploying AI to purge fakes, roughly two-thirds of reviews for many popular products were probably inauthentic, with recruiters now organizing across Reddit, Slack, Discord, and Facebook. The inquiry had immediate fallout — Facebook moved to ban more than a dozen groups used to defraud Amazon once the reporting surfaced them.
First-order effects
- Honest third-party sellers are squeezed out of visibility: as one source put it, it is very hard to sell anything on Amazon playing fairly when rivals buy rankings, and Amazon's own AI filters are demonstrably not keeping pace with the dispersed review networks.
Second-order effects
- Enforcement migrates upstream to the recruiting grounds — Facebook's group bans show platforms hosting the schemes being forced into the policing role — while sellers escalate to click farms and even paying Amazon insiders for ranking information to stay ahead of detection.
Third-order effects
- If the pattern holds, marketplace fraud stops being a platform-moderation problem and becomes a regulatory one: the FTC's proposed per-exposure fines for bought reviews would shift liability from anonymous reviewers to the companies purchasing them, attacking the demand side that moderation keeps missing.
The trend: Marketplace trust is becoming an arms race where each enforcement wave displaces fake-review supply onto new platforms until regulators step in to price the fraud out.