Amazon and the Case of the Unjustly Deleted Review

There has been a lot of discussion on the interwebs about Amazon deleting author reviews. Lots of authors are upset, and rightfully so. Being an author myself, I can imagine how frustrating it would be to have good reviews deleted from my upcoming book. However, being an Information Technology professional who specializes in data, Big Data, metadata, etc., I can also appreciate how difficult this task is for Amazon.

Amazon doesn’t have enough time or money to hire and train a staff dedicated to manually policing every review and ensuring that only those reviews that are in violation of the rules are deleted. Therefore, they must automate those efforts.

Let me be clear: this is VERY difficult to do.

I have no inside knowledge of how Amazon is doing this, but based on my own industry experience, my assumption is that they are using a score based approach. By score based, I mean that they are programmatically scanning the reviews and using computer algorithms to generate a score value when certain targets are found.

For example, combinations of words, phrases, and other targets will generate a score. “I am an author” could generate a point. A five (or one) star review could generate a point. A URL to another book could generate a point. If the reviewer’s name is found in Amazon’s author database, it could generate a point. Some targets will generate multiple points, others fewer.

Once enough points add up to a predefined threshold, then the review will be targeted for deletion. This deletion could also be automated, or the review could then be funneled into a queue where a human would examine it (although it doesn’t appear as if that is happening).

As you can imagine, this process isn’t foolproof. In fact, we have all seen enough examples to know that just about anyone could unwittingingly generate a review that hits the threshold. Alternatively, it is just as likely that reviews which should be deleted won’t be because they won’t generate a high enough score.

Is there a better way that Amazon could do this? Aside from the human staff I mentioned earlier, the answer is no. At least there isn’t a method that I’m aware of.

One could justifiably argue that Amazon has implemented immature algorithms in a knee-jerk response to paid reviews, or that they should have done more extensive testing before implementing these algorithms in the marketplace, but what’s done is done. Now all Amazon can do is learn from their mistakes and refine their scoring algorithms. Eventually these algorithms will become “pretty good” at their assigned task, just as Google has become pretty good at getting the answers to your search requests, but these algorithms will never be perfect. At least not until computers start thinking for themselves.

And then we will have a whole new bucket of problems to deal with. 🙂


10 thoughts on “Amazon and the Case of the Unjustly Deleted Review

  1. I wondered how Amazon was able to properly assess thousands upon thousands of reviews. Reminds me of the automated software used by recruitment companies designed to reject resumes! Thanks for the insightful comments – very useful.

  2. Certainly makes it seem less of a witch-hunt! I am not a published author yet, but hope to be soon – do you think these algorithms will find the reviews I have already left, and maybe delete them because I now hit the ‘author’ points?

  3. Thank you for that clear and simple explanation, Shawn – understandable even to non-techies like me. So Amazon’s decision-making process was not quite as random as it first seemed, though it is still a bit sinister! But knowledge is power…

  4. As an author who just had one of my best 5 stars reviews deleted because the reviewer referenced another book in her review (or at least that’s what I think happened; there was no other reason for the review to be deleted), here’s a crazy idea for Amazon: DON’T DELETE ANY REVIEWS!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s