The American social scientist Donald Campbell observed that as soon as you try to apply and use indicators for social decision-making, the pressure to corrupt the measure increases and it then distorts the process it is measuring. The measure eventually becomes meaningless because it loses its value as a measure (this principle is also known as Goodhart’s Law. Nowhere has this become more true than using product reviews as a means of gauging the public’s opinion of a particular product and as a recommendation to buy it.
Amazon’s fake reviews
Over the past few months, it has been revealed that the practice by authors of writing glowing reviews of their own work on Amazon under fake accounts whilst also writing bad reviews of their “rivals” was commonplace. Crime writer R.J. Ellory and thriller writer Stephen Leather have admitted to so-called “sockpuppeting” - the practice of assuming an online identity in order to deceive readers. Even academic writers have got into the act with Professor Orlando Figes, a historian at Birckbeck College, being sued for damages for writing fake hostile reviews of books by fellow historians.
Given the risks involved with authors writing reviews themselves, it wasn’t surprising to read that positive reviews could be bought online. The New York Times reported about one service that was offering a 5-star review for $99. The same article reported a researcher Bing Liu as estimating that up to a third of all reviews on the Internet could be fake.
Academics are faking it too
Over in the academic world, there is another take on the fake review. The quality of academic journals is assessed in part by a measure of how often papers that are published in the journal are cited (referenced) by other papers. This year, 51 journals were blacklisted from an indexing service for the practice of encouraging their authors to artificially cite other papers in the same journal in which they were published, thus boosting the Impact Factor of the journal.
The process of “self-citation” to boost an individual academic’s measure of their citation rate is also a prevalent practice, with one estimate being that 36 per cent of all citations are to work done by the authors themselves. Whilst some reference of your own work is appropriate, the pressure to increase citations by gratuitous references has proved too strong.
Faking reviews on blogs and social media
Trying to manipulate online reviews for products is nothing new. Companies have paid bloggers to review products. Even a media “advertorial” could be seen in the same light as an advert masquerading as an independent review of a product. Now that product endorsement and product brand awareness have moved onto social media, the temptation to build up a “buzz” around a product by using fake accounts has proved too great for many companies. Earlier this year, Facebook acknowledged that 1.5 per cent of its accounts were undesirable accounts. On Twitter, some 27 per cent of the top 10 Twitter accounts’ followers are estimated to be fake. 1,000 fake followers could be bought for around $18.
Improving the quality of online reviews
Research has shown that reviews have a direct impact on sales. Negative reviews impact sales more than positive reviews and reviewers. Customers are affected by the reputation of the reviewer and are more swayed by reviews in early stages of the product’s lifetime when there are relatively few reviews than later on when there are more.
There are a variety of ways that sites hosting reviews could tackle the problem of fake reviews. Amazon could require that only the people who have bought the product could comment on it. They currently do indicate that a review was based on an “Amazon Verified Purchase” but don’t require it. Reputational score on Amazon is only based on the number of people who voted that a review was helpful. Not only can this be faked but also there is no reason that people would not find fake reviews helpful, since they don’t actually have to be untrue.
Tying reviews and reviewers to social networks is another way that consumers can have some sort of validation of a review’s authenticity. In fact, word-of-mouth endorsement of products and services still remains the most effective way of fake-proofing an opinion.
There is also the Rotten Tomatoes approach that takes two measures – one provided by professional critics and other from the general public.
Online reviews are not the only thing
From a consumer perspective, it is possible that we are becoming more sophisticated in our attitudes to reviews. Although negative and positive reviews do have an influence, at the end of the day, they are not necessarily going to make or break a product or service in the absence of all other factors. Fifty shades of Grey, for example, has been Amazon’s biggest selling book. On Amazon.com it has had nearly 12,000 reviews with an average rating of 3 stars – evenly split between 4 & 5 stars against 1 and 2 stars. The most useful reviews, displayed at the top of the page, are overwhelmingly negative.
But perhaps this just bolsters the case for the fakers of reviews and shows that Oscar Wilde was right when he said “The only thing worse than being talked about is not being talked about”.
David Glance is a Director at the Centre for Software Practice at The University of Western Australia