Will We Ever Be Able To Trust Online Reviews?


‘Fake Reviews Are Still Rife’

Three summers ago, in August 2012, one of the hottest stories of the year came from the book-retail sector. The author John Locke had confirmed to The New York Times’ David Streitfeld that he had paid for reviews, lots of them, starting with 50 reviews of his books for $1,000. The story, The Best Book Reviews Money Can Buy, was one of Streitfeld’s best-written articles — nuanced, spun out in deepening stages of revelation. Locke, the first writer credited with selling 1 million ebooks on Amazon, was the author of How I Sold 1 Million eBooks in Five Months. That book had not mentioned buying fake reviews, as Streitfeld deftly pointed out.

Streitfeld would follow that outing with another story the following month, His Biggest Fan Was Himself, about UK author R.J. Ellory’s sock-puppetry in which he wrote glowing reviews of his own work under pseudonyms online and gave one-star reviews to other authors’ work.

News of these and other cases played out over weeks: It might be one thing to suspect that the bathroom scales you wanted to buy had a heavy round of consumer-review interest. It was another thing to have respected authors you had read and appreciated now confessing that they’d lied to their readers.

One of the better things relative to these events was the publication of the book Virtual Unreality: Just Because the Internet Told You, How Do You Know It’s True? (2014, Viking) by New York University journalism professor Charles Seife. As Dwight Garner wrote at the Times, Seife was commissioned by Wired to investigate allegations of bad practice on the part of Jonah Lehrer.

In Virtual Unreality, Seife writes that information online “is all a war over who gets the ability to affect your reality, to shape your social interactions, to manipulate your beliefs and control your behavior.” Fake reviews might have a lot of effect on your behavior. Big star ratings might manipulate your beliefs about a book or its author.

Although Amazon began removing some reviews deemed bogus and announced tougher oversight, that August-September period in 2012 remains a turning point. In books, consumer-written reviews have never again been as readily trusted as they once were.

Last fall, as the Alliance of Independent Authors (ALLi) launched its “Ethical Author” campaign at The FutureBook Conference in London, the third point of the Ethical Author Code directly addressed the problem:

Reviewing and Rating books: I do not review or rate my own or another author’s books in any way that misleads or deceives the reader. I am transparent about my relationships with other authors when reviewing their books.

This past April, Amazon sued several Web sites for, it said, producing and selling fake reviews. Reuters’ report of this was carried in the Times:

Amazon said the bogus reviews undermine a system that the Seattle-based online retailer launched 20 years ago to help shoppers using its website decide what to buy…Amazon said the defendants are misleading customers, and through their activity generating improper profit for themselves and a “handful” of dishonest sellers and manufacturers.

London Calling

Last Friday (19th June), the UK’s Competition & Markets Authority (CMA) released a 71-page “Report on the CMA’s Call for Information” on Online Reviews and Endorsements (pdf). The report marks the launch of an investigation in the UK into online fake reviews. It’s unknown whether this investigation involved Amazon; no specific companies have been named.

But — unrelated, and also on Friday — Ben Fox Rubin wrote at CNET.com about Amazon rolling out “a big change to its customer reviews system in the US, introducing a new machine-learning platform it developed in-house to surface newer and more helpful reviews.”

A part of the new program, Rubin writes, will involve adjusting the weightings of reviews in Amazon sales-page star ratings. Criteria in those adjustments are to include such factors as how new those reviews are; whether verified Amazon purchasers have written them; and whether other site visitors commend those reviews as helpful.

Rubin interviews Amazon’s Julie Law in consumer retail public relations. She tells him: “The system will learn what reviews are most helpful to customers…and it improves over time,”

And as it’s deployed, Rubin writes, the new approach is expected to be largely undetectable for a while. It’s in the training stage:

The change, which started Friday, will probably go unnoticed at first, as the e-commerce giant’s new platform gradually starts altering the star ratings and top reviews on product pages.

When Consumers Are The Critics

Reviews written and posted by customers differ from typical professional reviews in a couple of key ways that can affect their role in the marketplace: (1) the sheer number of voices reviewing things, and (2) a motivational intent on the part of many reviewers.

  • In the world of professional review, say, of computer software, there certainly are plenty of sources for buyers to use. CNET, mentioned above, is one. But even with as many online and offline publications as there are to choose from, those numbers are nothing compared to how many users may feel themselves qualified to issue opinions in online settings.  Faced with thousands upon thousands of rank-and-file customer reviews, Amazon at times applies one or more of its reviewer badges (“Top 10 Reviewer” or “Vine Voice”) in an effort to signal reviews created by shoppers who are vetted by the company.

There’s nothing innately wrong with there being many opinions available. If anything, this should be useful and it should help shoppers avoid “filter bubble” effects of hearing from the same parties over and over. On the other hand, there are fewer chances to identify a small group of trusted influencers, as it once was when mainstream critics were the prime sources for appraisals. There was a time when “I never agree with anything that critic writes” was a perfectly good way to evaluate that critic’s reviews.

“By the time everyone realized how open it was to abuse, the concept of people reviewing like this…was just part of our culture, and so very difficult to turn back the clock.”
Author Jeremy Duns
  • The second distinguishing characteristic of many consumer-written reviews is that motivational factor: the reviewer may want to persuade you one way or the other. Strictly speaking, a professional reviewer will not tell you whether to buy a product. He or she lays out her or his opinions and stops right there, allowing the reader of the review to make any decisions. You can see this, for example, in the longtime work of ConsumerReport.org reviewers. Testing results are offered and the reader is left to decide if or how to use the guidance. The lay reviewer, on the other hand, frequently seems to want to sway other shoppers. “Don’t Buy This!” or “Buy Three Of These Now!” are the result. And this emotional potency, this zeal to motivate a shopper, may have little to do with accurate and helpful information. It may simply reflect a review writer’s desire to influence other consumers.

These two challenges are inherent in the nature of the commercial world’s adoption of consumer-written reviews. They’re amplified by the gravity of purposeful falsehood in fake reviews.

‘Most Authors Feel A Little Bit Hostage’

At The Bookseller in London, my colleague Lisa Campbell has been in touch with author Jeremy Duns, who surfaced Ellory’s sock-puppetry in 2012. In CMA probe ‘should clean up fake book reviews,’ Duns tells Campbell that “fake reviews are still rife, and pretty much any author you speak to has experienced it.”

He adds, “or taken part in it.”

In his thoughtful comments to Campbell, Duns says:

I think fake reviews are damaging to the publishing industry: in essence, big bookselling sites set up the ability for anyone to review books but didn’t in my view think it through properly, and by the time everyone realized how open it was to abuse, the concept of people reviewing like this had become so embedded that it was just part of our culture, and so very difficult to turn back the clock.

I think most authors feel a little bit hostage to this situation. Unfriend someone on Facebook and perhaps in a few days you might find a mysterious new reviewer giving your books one star: they [online reviews] make a difference, as the CMA points out, but are extremely open to abuse.

The CMA report, which seems to be aimed not at publishing per se but at commerce in general in the United Kingdom, lists “a number of practices that concern us.” Excerpting and quoting them here:

  • Businesses writing or commissioning fake positive reviews about themselves to boost their ratings on review sites relative to rivals.
  • Businesses or individuals writing or commissioning fake negative reviews.
  • Review sites “cherry-picking” positive reviews, or suppressing negative reviews that they collect and/or display, without making it clear to readers that they are presenting a selection of reviews only.
  • Review sites’ moderation processes potentially causing some genuine negative reviews not to be published.

There’s  useful “60-Second Summary” of the report with guidance for businesses that host product or service reviews. Among the advice:

Be clear about where reviews come from, and how they are checked before publishing…Publish all reviews, even negative ones, provided they are genuine and lawful…Explain to consumers the circumstances in which reviews might not be published or might be edited (for example if they include swearing, abusive language or defamatory remarks)…Make sure there is no unreasonable delay before reviews are published, so consumers get the current picture…Disclose commercial relationships with businesses that appear on your site, and explain how this may affect the review ratings and/or their ranking…Clearly identify advertising or paid-for content…Have appropriate procedures in place to detect and remove fake reviews, and act promptly in response to reports of suspected fake reviews.

The report carries results gleaned from a consumer survey commissioned by the CMA. In some of the responses lie important points.

For example, the report tells us:

Results from the consumer survey we commissioned suggest that 54% of UK adults read online reviews. Online reviews appear to be read for one-off purchases, ‘experience’ goods (goods that the consumer cannot easily assess for quality before they buy) and/or more expensive goods or services – hotels and travel, for example. Across the six broad sectors that we looked at, we estimate that £23 billion [$36 billion] a year of UK consumer spending is potentially influenced by online reviews.

And there’s a considerable trust factor indicated by the survey’s respondents to the CMA:

Between 76% and 80% of consumers (across the six sectors) who had used reviews before making a purchase thought that it was either ‘very likely’ or ‘fairly likely’ that the reviews they read were written by genuine consumers.

So it would seem that publishing industry insiders and authors struggling with star ratings and consumer-written reviews might share a common interest with the CMA and others interested in efforts to clean up, to any degree possible, the image that consumer-written reviews may have and the shadow that image can cast on those touched by them.

When the historical fiction author Jane Steen spoke to me last year here at Thought Catalog about her concept for what would become ALLi’s Ethical Author campaign, three of eight critical areas of concern she listed for me were paid reviews, reciprocal reviewing arrangements, and sock puppet accounts.

What has become one of the fundamental elements of contemporary online retail, the consumer-written review — digitally enabled and, in and of itself, a great idea — still remains one of the most vexing issues of modern commerce, especially as authors’ reputations and relations with their readers become more important seemingly by the day.

As Duns says to Campbell at The Bookseller:

Carrying out this one recommendation — checking reviews before published, for defamation, that the person appears to have at least read the book — and indicating how they have been checked clearly to readers — would be a massive step forward.