Over the weekend, Twitter exploded with anger directed at Amazon.com. The bookselling giant had effectively blacklisted GLBT (gay, lesbian, bisexual and transgender) books by blocking them from search results and sales rankings.
We're not just talking about explicit books, either. The affected books included scholarly works, award-winning novels and even Brokeback Mountain.
Depending on whom you believe, there are three possible explanations for the sudden disappearance of more than 50,000 titles: human error, hacking or a sinister plot.
Amazon says it was a mistake, a "ham-fisted" and "embarrassing" error where books were accidentally given the wrong classification.
Hacker "Weev" says it was him, claiming on LiveJournal (NSFW) that he took advantage of a flaw in Amazon's system for reporting inappropriate content. "The thing about the adult reporting function of Amazon was that it was vulnerable to something called "Cross-site request forgery'," he says. "This means if I referred someone to the URL of the successful complaint, it would register as a complaint if they were logged in. So now it is a numbers game." Weev claims to have hired an army of troublemakers to flag thousands of books as inappropriate.
Last but not least, there's the conspiracy theory. Amazon, the theory goes, is pandering to the US religious right. The "category error" explanation is a whitewash, and the whole thing was deliberate. If it weren't for those pesky Internet users, Amazon would have got away with it.
So which is true? Given the choice between human error, hacking and conspiracy we'd tend to go for the human error angle every time, but of course it's entirely possible that Amazon was indeed the victim of a hacking attack and that its PR people are pretending otherwise.
If that's the case, it demonstrates a big problem with any user reporting system: it can be abused. Whether it's taking the mickey out of David Hasselhoff or acting out of political or religious views, it doesn't take much effort to recruit a bunch of like-minded people and flood reviews with nonsense or flag perfectly decent content as inappropriate. That's why it's important that humans, not computers, review any inappropriate content flagging.
If it's a conspiracy, then the episode shows the power of social networking. In the space of 24 hours the #amazonfail discussion on Twitter became big news, with thousands of people spreading the word.
If you subscribe to the conspiracy theory Twitter users, bloggers and other outraged Amazon watchers are the Scooby-Doo to Amazon's cackling baddie, the pesky kids without whom Amazon would have got away with it.
If it is just a mistake, however, then the Twitter storm shows the downside of social networks: tune into the conversations and you'd be convinced that Amazon didn't just make a mistake, but that it has got a problem with gay people.
The story is already a PR disaster, generating huge amounts of news coverage, and even if it is a genuine mistake there will be plenty of GLBT people whose trust in Amazon has taken a big knock.
If ecommerce firms want to be sure that their actions aren't being misinterpreted or that their mistakes aren't going to blow up into a big scandal, they need to be monitoring the conversations - or better still, joining in.