Monday, June 25, 2012

Are There Sophisticated Fake Reviews and Reviewers?

Quite often, before choosing a restaurant, or buying somewthing, or hiring someone to work on my roof, I check the web for reviews. Occasionally I find a suspicious pattern—lots of very positive and rather generic reviews, along with a smaller number of very negative reviews that sound as though they were written by real people—which suggests that the positive reviews are fakes, written by someone working for their subject. On one notable occasion, that pattern warned me off of a roofing firm that, I concluded, was the third generation of a serial scam. There exists at least one online firm whose business is improving a firm's online reputation—I do not know enough about them to say whether they restrict themselves to honest ways of doing it.

Recently, looking for a furniture store, I came across what I thought might be a less clear version of the same pattern. The reviews were on Yelp, which lets you  click on the name of the reviewer and see the rest of his reviews. If all of one reviewer's reviews are in praise of a single firm, one might suspect that he works for them—and his job is writing favorable reviews. The reviewers I was looking at did not fit that pattern.

Which started me wondering  how sophisticated the people who sell the service of improving a firm's reputation might be by this time. Are there some who deliberately create believable reviewers, write real reviews of a number of products or businesses, and then sell the service of having the same reviewers write glowing reviews for anyone who will pay for them?

Anyone know?

14 comments:

Steve said...

Indeed, yes - and not just for reviews of goods and services, but for comments on political blogs. I've seen spam from the sellers of such services, but I'm not sure how their business model works (for one thing, I can't see anyone with a grain of sense giving them sufficient bank details to be paid)

RP Long said...

My sister works in advertising and confirms that there are, in fact, people who are paid to write positive reviews - and also folks who are paid to write negative ones.

Eric Goldman said...

A business can buy a positive review written by a human on Mechanical Turk for a couple of dollars. Eric

G.A. said...

If you search the uber popular site fiverr.com for "testimonials" or "reviews" you will get about 5,000 people offering to write or do video testimonials for your product or service... And this is only 1 website! I only read reviews that have been filtered by Testimonial Shield.

Douglas Knight said...

One not very sophisticated thing that definitely happens is using a single reviewer account to write bought reviews of different companies. They know that you're clicking to look at all reviews by the same person.

Anonymous said...

@ G.A. I had never heard of Testimonial Shield before. What a great idea and necessary service (I can't believe they've been around for 6 years) and just in time with the explosion of review sites and the ability of anyone to post reviews it's nice to know that someone is out there weeding out the reality from the B.S. Does anyone know how they determine the difference between real and fake reviews?

Anonymous said...

does this mean that the free market doesn't work?

Unknown said...
This comment has been removed by the author.
Richard Gilbert said...

Angie's List is good for finding actual service providers for exactly this reason. Their reviews are by members only, and companies cannot be members, nor can members write reviews for their own companies. Because the membership is associated with payment info, it is more secure. No doubt there are ways around it, but many of the easy ways wouldn't work.

jdgalt said...

There seem to be enough outside people willing to create fake reviews that it probably isn't necessary for the company to hire it done themselves.

Some 20 years ago, there was a scandal at my home town paper that hardly anybody heard about (I got it from one of their columnists). Their head restaurant reviewer went to a very expensive restaurant, was quite disappointed, and wrote a review saying so. The paper's editorial staff changed the review to say the place was perfect. A reader went there, was also quite disappointed, and wrote back to the reviewer saying so.

The reviewer then made the mistake of telling the truth in print. (The letter and reply were printed on a Saturday, which I assume is how it got through.) Result: the reviewer was fired, and no explanation was given to the public.

This illustrates the folly of one of the mainstay ideas shared by a lot of libertarians: reputation is NOT enough to prevent fraud, because it's too easy to fake it, change 'nyms and walk away from it -- or even earn a good rep and then cash it out, as Bernie Madoff did. There has to be law enforcement against fraud, or commerce with strangers is simply not viable.

Anonymous said...

so as an economist you have to judge the system through understanding what its doing in the long term. are there equilibriums, how stable are they and what determines which equilibrium the system will tend to after shock? there are clearly many reviews which are bought by companies, both to destroy the competition and to improve their own reputation. but on the other hand not all people are stupid, many see which reviews are bad or not, and there are ways of checking the honesty of the reviewers. so in the end there are some people who are influenced by dishonest reviews, it is not obvious which way though. in general i would say that the equilibrium is not first best, ie its not as efficient as a situation where there is perfect information and costless negotiations. however the 2nd best, which we have, is probably the best that can be achieved. its also interesting to see whether the 2nd best equilibrium tends towards first best if technology changes. i would say probably not. we just have to accept that in the real world competition works but its not perfect and it cant be made perfect, but thats better than any other alternative. dave?

גליה סגל הלוי said...

One possible solution is the "social proximity" gauge, such as this: http://www.meezoog.com/zp/go.php?to=static&name=social_proximity&src=lobby&language=English

The idea is to calculate, for every pair of users in the system, the *subjective* trust between them, based on the social paths that connect them.

For example, if I join and invite my brother, I can say that I trust him 100%, so his reviews will appear as trustable to me. But, if another person registers to the site and is not connected to me or my brother, then his reviews will appear 0% to me. If he intives his own brother, then probably their reviews will appear trustable to each other, but not to other people outside the clique.

It can be proven that this sys

Anonymous said...

uh, yes. it's called rent-seeking socially (if not acceptable) then tolerated sociopathy. you think the guys making collection calls are all bunnies and rainbows and sunflowers?

Anonymous said...

OnTheMedia blog had a story on Public Relations for Dictators: "they would write and place op-eds in American newspapers. And what they do is, is that a staffer at APCO would write it. Then they’d go out and recruit an academic or somebody at a think tank to put their name on it. And then they would go place it in an American newspaper, so it would look like some independent, thinking human being, as opposed to a paid flack for a dictatorship."

http://www.onthemedia.org/2012/jun/15/pr-dictators/transcript/