Susan Crawford is an eminent telcoms scholar, former government official (who resigned because of corruption in telcoms policy) and the author, recently, of an important book on telcoms corruption and net neutrality called Captive Audience: The Telecom Industry and Monopoly Power in the New Gilded Age. This book has scared the pants off of big telcos.
Their anti-Net-Neutrality front groups like NetCompetition, Broadband For America, and Media Freedom have been smearing Crawford and her book since it was published, and now, at least 31 people have posted highly similar one-star reviews of her book to Amazon, quoting talking points from these organizations. Most of these reviewers are not in Amazon's "real name" program, and the ones that are work for big telcos and the think-tanks they fund. Mike Masnick investigated the reviews in detail and it's pretty clear that nearly all the five-star reviews are from legit, named, disinterested parties (albeit with a few people who have a dog in the fight, like activists and scholars, and a couple more who say they are trying to balance out the one-star smears); meanwhile, nearly all the one-star reviews are from shills or telco people.
America has some of the worst Internet infrastructure in the developed world, and it's getting worse year by year. It's thanks to the crooked phone companies and their corrupt pals in Congress, the state houses, and the regulators. These titans have the country by its nervous system, and they're so afraid of criticism that they engage in petty, corrupt astroturfing to attack books that call them out. Now is a great time to buy Captive Audience: The Telecom Industry and Monopoly Power in the New Gilded Age, read it, and give it an honest review.
My second check was to look at whether or not the reviews had one of the following three criteria: they were a verified purchaser, they were enrolled in Amazon's "REAL NAME" program, or they had reviewed other products besides just Crawford's book. While this is a rather crude measure, I figured that having any of those things be true at least suggested that there was a real person behind the review. Having none of those three things might still mean they were a real person who legitimately bought the book and was giving a legitimate opinion, but at the very least it couldn't be proven. To give the benefit of the doubt to the Crawford haters, here I added back in the known policy wonks — who were basically the only 1-star reviewers to qualify as humans under these criteria. Without this, I think only two of the remaining 26 reviewers could meet the criteria. In the end, even with the known DC policy insiders, only 11 out of the 31 reviews, or 35% met the criteria.
Of the 5-star reviews, 80% met the criteria. And, even this is somewhat misleading. Of the reviews that did not meet the criteria of provably human, nearly all of them mention that they're leaving a 5-star review solely to counteract the obvious shill 1-star reviews. I think that's counterproductive in many ways, but it suggests that those reviews weren't directly "shill" reviews, but rather response to astroturf reviews.
As a further check, I compared the average number of other products reviewed by each group — the 1-star reviewers and the 5-star reviewers. That really wasn't a fair fight. The average number of "other" reviews by those who gave Crawford's book a 1-star review: 1.4. And that's almost entirely due to one person, Richard Bennett, who has 24 other reviews. Of those who gave it a 5-star review, the average is 113.9. Yes. 1.4 vs. 113.9. Okay, the 5-star reviews are also skewed heavily by one reviewer, Loyd E. Eskildson who has over 4,000 reviews. So, to be fairer, I cut out that outlier and the 5-star reviewers still had an average of 13 other reviews (I didn't even bother to take out Bennett's outlier on the 1-star reviews). Using the other (probably better) tool, we could also compare the median other reviews for each group. For the 1-star reviews, it will surprise no one to find out that the median is 0. For the 5-star group, the median is two.