Fraudulently elevating the quality of one’s product or service is at least as old as ancient Rome when cheap wine was often passed off as expensive high quality libation which at times had fatal consequences for the imbiber. Gaming user-generated content to snooker customers may not be as pervasive but it has garnered considerable attention in the mainstream media and even occasional regulatory scrutiny.
This week’s Middle Seat column in the Wall Street Journal, an interesting lens to view the airline industry through, turns its attention to the hotel industry in an article headlined ” “The Big Flaws in Hotel Rankings”. The article lays out the array of issues that affect how a hotel is ranked as well as how some operators employ dodgy methods like paid agents to submit fake (positive) reviews for themselves and even post falsely negative reviews about competitors.
Apart from deceitful operators the traveling public is faced with puzzling rankings that are divergent and even diametrically opposite in competing websites. But as the Journal notes, Tripadvisor despite being attacked not infrequently by unhappy owners and operators for its alleged fake review content, is the review industry’s leader by orders of magnitude with over 60 million reviews. One reason for its popularity,apart from being the “first mover” is that it is open for anyone to post regardless of their having stayed at the hotel. While that makes a good TA ranking a self-evident business imperative it is unclear that the noise from planted reviews is anything more than a distraction.
Nevertheless, fake reviews are beginning to attract regulatory oversight in the US as reported in a New York Times article earlier this year where an FTC official expressed concern about advertisements passing off as editorials. Likewise in the UK’s Advertising Standards Authority forced Tripadvisor to toss out the word “trust” that was used to characterize its reviews since reviews are not verified. These worries have spawned algorithms that claim to spot deceitful reviews including one from Cornell University’s Department of Computer Science that is said to have spotted manufactured reviews in 90% of 400 reviews that they ran their program on.
However, the fears may be overdone as can be seen from one example cited in the Cornell study where two glowing reviews for the same hotel were shown with the reader being asked to tell the fake from the real. The fake was gramatically correct with no spelling mistakes while the real had a sentence that read “I can honestly stay (sic) that the James (hotel) is tops.” Most folks fill in reviews to express their feedback not to write a precis for formal review when attention to language and style are important. In the end the proof of the pudding is in the eating: were travlers unable to discern the real from the fake Tripadvisor’s growth rate likely would have been lower if not considerably so.
