Movie Studios Build Buzz With Fake Film Tweets
It’s movie night. Which film should you see? Your decision might be swayed by fake tweets from studios.
Based on the research of Andrew Whinston
Movie fans shouldn’t believe everything they read on Twitter.
In recent research, Information, Risk, and Operations Management Professor Andrew Whinston, director of the Center for Research in Electronic Commerce at Texas McCombs, finds evidence that studios have been using phony tweets to build buzz for their new releases.
The spammers can use fake accounts or seed real accounts with overly glowing reviews. The effects are the same: steering moviegoers to films they might not otherwise want to see.
They’re getting less guidance from newspapers, which are cutting back on paid film critics, says Whinston. “Today, it could be the company itself generating fake tweets to get things going.”
His findings offer lessons for social media, on how to recognize and prevent faux film chatter, as well as for moviegoers, on how to avoid it.
Disturbances in the Twitterverse
Opinion spam is not new. Scholars have analyzed fraudulent writeups of restaurants on Yelp and hotels on TripAdvisor. But none have previously looked at online movie reviews, says Whinston, though rumors sometimes surface:
· Netflix was suspected of using fake Twitter accounts to spread memes about its 2018 release “Bird Box.”
· The 2018 film “Gotti” got 6,900 audience reviews its opening weekend on the film site Rotten Tomatoes, despite showing in only 500 theaters.
Twitter is particularly vulnerable to manipulation, Whinston says. It’s easy to create fake accounts, and false followers can be bought for pennies apiece. A recent University of Southern California study estimated up to 15 percent of the platform’s accounts were fictitious.
Fingerprints of Fake Tweets
Whinston says it’s hard to separate bogus reviews from real ones. So instead, with Shun-Yang Lee of the University of Connecticut and Liangfei Qiu of the University of Florida (both McCombs doctoral students at the time the research was conducted), he looked for statistical patterns in movie tweets.
If studios were manufacturing positive reviews, he reasoned, they would show up mostly before opening day. That’s when studios spend 90 percent of their ad budgets, hoping to attract a rush of early viewers.
To confirm their hunch, the researchers examined two 2012 films — “At Any Price” and “Act of Valor” — that were both hated by critics. They analyzed each tweet about the films, 60 days before and after release, and scored it on a scale from 0, for most negative, to 1, for most positive.
Overall, tweets were indeed more positive before a movie’s release. Average scores plunged between 10 and 25 percent on opening days, once real viewers began to share their thoughts online, suggesting the numbers had been pumped up beforehand.
“The studios can’t keep up with the sudden influx of genuine tweets right after a movie’s release. If a movie’s no good, real tweets will overtake the fake ones.” — Andy Whinston
Next, the researchers widened their focus to 482 U.S. films released in 2012 and 2013. They found an average drop of 5 percent in positive sentiment on release day. The difference is enough to sink a film from the middle of the pack in audience rankings to the bottom quarter.
Movies that faced competition from pictures with similar release dates and genres suffered even steeper declines. To Whinston, that suggests higher levels of tampered tweets.
“In those cases, there’s more incentive for studios to promote their movie, so that consumers choose to see it rather than a competitor’s movie,” he says.
Though the tweets he analyzed were six years old — a common lag time in academic research — Whinston says the practice continues, as the examples of “Bird Box” and “Gotti” suggest.
“Consumers need to be wary and realize that studios have a lot of incentives to put up fake tweets.” — Andy Whinston
Restoring Consumer Confidence
Like fake news, the long-term danger of fake reviews is the erosion of public trust in online information, Whinston says. (Twitter acknowledged the wider problem last year, suspending more than 70 million of its 336 million accounts.)
But Twitter could do more to prevent fraud before it occurs, he says. The movie-rating site IMDb discourages duplicate accounts and inflated reviews by limiting each registered user to only one vote per movie. The user can change a vote, but it overwrites their old one.
Are IMDb’s ratings more reliable than Twitter’s? The researchers compared both sites. IMDb showed less change in sentiment before and after a movie’s release, suggesting it was less susceptible to manipulation.
Consumers should look for such safeguards when they’re reading online reviews, he says, for example, certifying whether a commenter actually bought the product.
Filmgoers can also trust online reviews from people they already know. Whinston points to the neighborhood chat site Nextdoor. Before allowing a new user to register, it verifies their identity through a postcard, a phone number, or a billing address from a credit card.
Nextdoor offers local information on movies and restaurants that users know is legitimate, because it comes from their neighbors. Unlike film reviews from unknown Twitter users, says Whinston, “they’re hard to fake.”
“Sentiment Manipulation in Online Platforms: An Analysis of Movie Tweets” was published Oct. 15, 2017 in Production and Operations Management.
Story by Steve Brooks
