Skip to main content



Deception in Reviews

As I was in class on Thursday (3/28) when we were talking about information asymmetry, it reminded me of some research that I came across from Cornell Computer Science Professors, Myle Ott, Yejin Choi, and Claire Cardie. The paper is titledĀ  “Finding Deceptive Opinion Spam” and is about an algorithm to find out whether a review is genuine or not. According to the NYTimes article, the algorithm does a decent job of selecting 90% of the fake reviews. As you can imagine, without social cues like body language and facial expressions, it’s quite difficult to spot fake reviews on the web. But apparently, there is a market out there willing to exchange real money for fake endorsements of hotels and restaurants, positive reviews to benefit the company and negative reviews for any competitors. In a day where I personally go to a bookstore and instead of looking at the testimonials of the book, I go to Amazon to read the reviews. We live in a world where network effects dictate the quality of a product and the wisdom of crowds heavily influences choices.

 

In regards to information asymmetry, I believe that this is ultimately terrible for the users who are affected by the poor quality of the product. Let me continue on the example in class, say we take the market for used cars. People will get a notion of the percentage of good cars versus bad cars by going online and reading reviews about the dealership. So, without the fake reviews, say that people believe the percentage is 50% bad cars and 50% good cars. Now with the fake reviews thrown in, people now think that the percentages are 25% bad cars and 75% good cars. They would now be willing to pay higher for a car simply based on fake reviews. Because of the fake reviews, we are going in the opposite direction of unraveling down. We are now artificially playing with the information in the market so we can still sell crappy cars but at a higher price so that the seller is willing to sell.

 

I’m not exactly what happens next though. Perhaps people come back to the site after being unhappy and write a bad review, thus canceling out the fake reviews. Maybe people are indifferent and they just allow the artificially inflated information stay in the market affecting everyone’s choices. I actually do fine this topic interesting because we are really trusting of a conglomeration of reviews, when we shouldn’t necessarily be, but then again, how else can we make an informed decision? This article made me realize how we use little information to make choices.

 

-erc73

 

http://aclweb.org/anthology/P/P11/P11-1032.pdf

http://www.nytimes.com/2011/08/20/technology/finding-fake-reviews-online.html

Comments

One Response to “ Deception in Reviews ”

  • cfs6292

    interesting post. I too read “Finding Deceptive Opinion Spam” paper, but think the application on car markets is a nice extension of the piece. Another possible result that you did not discuss is people completely distrusting all reviews. If after being unhappy and people realizing some of the reviews are false, will they not trust any of the reviews? This could result in consumers taking uninformed gambles, or maybe avoiding the market for used cars all together.

    – cfs76

Leave a Reply

You must be logged in to post a comment.