Publishers vs. the Censorbot


Google has entered that rarefied tier of corporations whose services or products are indispensable to American life. Its search engine is an essential tool for Internet users, and the ingenious way it matches advertisements to the contents of any given search has transformed the world of marketing. The company’s second main product is AdSense, a service whereby it posts ads on thousands of Web sites and blogs in a revenue-sharing arrangement that finances much of the online media industry. But as more and more media outlets come to depend on Google, one of its key policies is quietly, systematically punishing ambitious journalism all over the Web. If this trend continues, Google’s power and ubiquity could inadvertently cripple an independent press in the Internet era.

The genius of Google lies in its capacity to pair advertising with similarly themed editorial content. This technique is revolutionizing the very nature of advertising, as businesses exercise more control than ever over the contexts in which their messages appear. Hardware stores, for example, can now ensure that their ads are placed next to stories about home improvement. But businesses also want to control what they don’t appear next to — automakers are displeased when their ads appear next to stories about traffic accidents. In response, in 2003, Google developed “sensitivity filters” to periodically scan the Web sites of its partners in search of violence, mature content, or other unacceptable material. “They detect sensitive content that we probably don’t want to be showing advertising beside, and show public service announcements instead,” says Shuman Ghosemajumder, Google’s business product manager for trust and safety.

Unfortunately, when Google withholds advertising it also withholds the accompanying revenue, cutting off money whenever Web sites publish stories it deems too violent or tragic. Regardless of how important a story may be, the company’s algorithm pulls its advertising whenever it detects too much carnage. Asked if Google would display ads next to stories about the recent Israeli massacre of Lebanese children, for example, Ghosemajumder says, “That’s an example of something that is very difficult to find sensitive advertising [for].” The larger Google gets, and the more indispensable it becomes to news-related Web sites, the bigger this problem will become.

Earlier this year, Salon signed a small advertising contract with Google, and employees quickly discovered that whenever a story dealt with sex too explicitly, the search engine would automatically pull its ads. Salon ran stories about a Senate hearing on the effects of pornography, a study on the effect of sex on stress levels, and British attitudes toward rape victims; Google pulled its ads for each of these articles. “What we found in working with Google was that because some of our content violated its ‘family-safe policy,’ as a result we had to work with other partners such as Yahoo,” says Kathryn Surso, Salon‘s vice president of business development.

Even individual bloggers have encountered this problem. Alameda resident Bob Mendelsohn supplements his earnings from video editing and audio production with revenue from AdSense ads on his blogs, including one that profiles ski instructors. Two months ago, Mendelsohn noticed that Google had replaced one of its ads with a public service announcement. “They all had ads, but one just kept showing Katrina ads,” he says. “I couldn’t figure out why.” After reviewing the profile, he guessed that some of the words he used — teenager, husband, bachelor — could be misread as part of a porn story. “I pulled a few words out, and the next day I had ads,” he says.

On May 16, a writer for the blog posted a message complaining about a short-lived Sony copy protection scheme that surreptitiously installed spyware on its customers’ computers, with the headline “Have you boycotted Sony products yet?” Google replaced its ads with a public service announcement and canceled all ad revenue for the page. The next day, the author, who goes by the name of Kehaar, wrote a second post complaining about Google’s response. “I know we get a little rowdy over here at Silflay Hraka, but I didn’t think we put up anything too objectionable,” the blogger wrote, adding that he or she might consider switching ad networks.

Few bloggers rely on ad revenue to pay their bills, and Salon‘s advertising base is sufficiently diversified that dropping the occasional Google ad doesn’t hurt it. But for smaller Web news outfits, losing Google revenue is much more serious. According to the publisher of a prominent news Web site who agreed to speak only if granted anonymity, his company recently signed a premium Google advertising contract that now accounts for a third of his site’s revenue. A few months ago, his Web site ran a series of stories about a major bombing in Iraq. Within hours, he says, Google’s ads vanished from his home page, and so did all the revenue they generated. “They said we had the word ‘kill’ on our site, and that killed the ads,” the publisher said. “I wrote them and said that would be very difficult for a news site, which would often use the word ‘kill.’ They said, ‘Those are the rules.'”

On July 7, the Web site linked to a story from The New York Times about racist groups and skinheads joining the army. Once again, the publisher claims, Google wiped out his ads. As near as he can tell, the use of the word “skinhead” must have run afoul of Google’s ban on Web sites that publish racist content. In both cases, the ads vanished for several weeks. The publisher estimates that his company lost at least $7,000 while waiting for Google to reevaluate his content again and bring back the ads.

When the publisher contacted Google and asked for explicit guidelines about what constitutes illicit content, company representatives refused. “I asked them for a set of keywords, and they wouldn’t give me one,” he says. “I don’t know what the words are; we just have to approach it by toning down the language in our articles. … It’s just ridiculous. I don’t think the [advertisers] are going to have a problem with us reporting the news. … But they’re Google, and we’re a small site. So we’ll have to conform to their regulations if we want their money.”

Ghosemajumder says Google constantly refines its method of matching stories to advertisers, whittling down the number of topics that would kill ad revenue, and adds that publishers should always alert the company when they think an ad has been inappropriately pulled. But he refuses to offer a list of forbidden words, claiming that Google’s vetting procedure is much more complicated: “We’re not trying to create very specific rules so much as we’re trying to determine, ‘What is the topic of a particular story such that viewers would have a negative reaction?'” In the end, he acknowledges, some stories may be too unpleasant to be paired with paying advertisers.

As the Internet becomes more important for newspapers, and publishers from The New York Times to Village Voice Media look to Google for scarce online revenues, serious journalism about international conflict, crime, or any of the less-pleasant aspects of human nature could find itself gasping for support. If stories about the Arab-Israeli conflict, the occupation of Iraq, or the genocide in Darfur are destined to appear without advertising, editors and publishers may conclude they can no longer afford to cover the world’s most important stories. In an era of targeted, contextual advertising, what will happen to the stories no sponsor wants to touch?

This article was published originally on August 2, 2006, in the East Bay Express.