Algo-Driven News with a Human Face
posted by Frank Pasquale
Chistopher Steiner’s new book on algorithms looks interesting. (One nugget: Many companies now use software to analyze the emotional tone of customers calling in for customer service help. Sound emotional, and you’ll get routed to the more empathic call center workers.) It’s part of a growing literature on algorithms both online and off. As we search for reliable information on algorithms, they in turn may well be driving even our awareness and discussion of them. It’s another way technology shapes values, rather than being influenced or constrained by them. Consider a recent feature on an increasingly algorithm driven news industry:
Google News-powered results, Google says, are viewed by about 1 billion unique users a week. . . . Which translates, for news outlets overall, to more than 4 billion clicks each month: 1 billion from Google News itself and an additional 3 billion from web search. . . .
Google News’s head of engineering[] summed up the challenge: “How do I take a story that has 20,000 articles, potentially, and showcase all of its variety and breadth to the user?” . . . . Google [is] symbolic of a broader transition: producers’ own grudging acceptance of a media environment in which they are no longer the primary distributors of their own work. [It] suggests an ecosystem that will find producers and amplifiers working collaboratively, rather than competitively. And working, intentionally or not, toward the earnest end that Schmidt expressed two years ago: “the survival of high-quality journalism.”
When Google News launched in 2002, it’s worth remembering, it did so with the following . . . declaration: “This page was generated entirely by computer algorithms without human editors. No humans were harmed or even used in the creation of this page.” Since then, as news publishers have emphasized to Google how human a process news production actually is, the company’s news platform has — carefully, incrementally, strategically — found ways to balance its core algorithmic approach with more human concerns.
The article frames Google’s approach as a series of magnanimous concessions to squabbling journos—the commodity “paint” (as Lessig christened them a few years ago) artistically arranged by Google into a picture of the world. Like most business news nowadays, the commercial concerns about fairly dividing the pie of digital advertising dominate. But I have to wonder about Google News’s public role, and how it could potentially be manipulated. How do stories make it to the top of the Google News front page? How important is the sheer number of mentions of a given story, as opposed to, say, the authority of the news outlets promoting them? (For example, how long should the 47 percent meme dominate presidential news coverage?) And finally, as direct human interventions into the page increase, what are the standards for raising or lowering the prominence of the story? Will news outlets be able to pay for premium placement, like hotels appear to be doing?
If all those decisions are made behind closed doors at the Googleplex (or the Twitterdome, Facebookistan, or wherever your favorite intermediary is), expect increasingly vertiginous online sense making. At its best, Google News could be one more indicator of journalists’ sentiment and interest, helping gatekeepers decide what’s the most important news and ordinary readers tame information overload. As personalization continues apace, rival services will develop different “theories of you” to decide what to present, as Eli Pariser explains:
Google’s filtering systems . . . rely heavily on Web history and what you click on (click signals) to infer what you like and dislike. These clicks often happen in an entirely private context: The assumption is that searches for “intestinal gas” and celebrity gossip Web sites are between you and your browser. You might behave differently if you thought other people were going to see your searches. But it’s that behavior that determines what content you see in Google News, what ads Google displays—what determines, in other words, Google’s theory of you.
The basis for Facebook’s personalization is entirely different. While Facebook undoubtedly tracks clicks, its primary way of thinking about your identity is to look at what you share and with whom you interact. That’s a whole different kettle of data from Google’s: There are plenty of prurient, vain, and embarrassing things we click on that we’d be reluctant to share with all of our friends in a status update. And the reverse is true, too. I’ll cop to sometimes sharing links I’ve barely read—the long investigative piece on the reconstruction of Haiti, the bold political headline—because I like the way it makes me appear to others. The Google self and the Facebook self, in other words, are pretty different people. There’s a big difference between “you are what you click” and “you are what you share.”
As time wears on, the real “news” will be those few items that break through the “filter bubbles” of a critical mass of the populace. In Gary Shteyngart’s novel Super Sad True Love Story, a service called “CrisisNet” provides urgent updates that everyone needs to know. Meanwhile, the New York Lifestyle Times masters the most profitable ways of grabbing the attention of High Net Worth Individuals. Somehow I think a novel like his provides a scenario analysis of the future of news more prescient than most algorithmic predictions of “present and future business models to monetize the newspaper industry.”
X-Posted: Madisonian.
September 21, 2012 at 8:42 pm
Posted in: Cyberlaw, Google & Search Engines, Intellectual Property
Print This Post
Responses (2)
A.J. Sutter - September 21, 2012 at 11:12 pm
The conjunction of “filter bubbles” and “intestinal gas” is indeed disturbing. I wonder, though, if issues like the persistence of the 47-percent meme are really so new. Until the 1990s, the persistence of that meme would have been up to a handful of newspapers and a few TV networks — still quite oligarchic.
And things could be worse. Take, e.g. Japan. Certain issues are kept out of the media through the use of “press clubs”: if you want to get information from a government agency or large corporation, you have to be invited to their press briefing, and the only way to keep getting invited is not to cross them. I’ve also heard from TV station staff that the government will threaten stations with license revocation if certain stories are broadcast. Recently we heard from the husband of a relative about goings-on in Soma City, one of the cities worst affected by the Fukushima Dai-ichi nuclear meltdown; he is a banker there. It seems the government has come up with a deeply stupid and unfair compensation system for the city’s residents, which has resulted in the abandonment of many businesses while people spend windfalls on gambling and other amusements. None of this is reported. As he put it, “This was so outside my expectation, I couldn’t even dream up a search term would let me find it on Google.”
Another development here: A DVD called Naito Sunakkaa (“Night Snacker”), by a comedian who travels around Japan visiting hole-in-the-wall bars known as “snack” bars (sunakku), has become a huge hit. (Such bars are favorite hang-outs of what used to be called the Lumpenproletariat, though there are fancier ones, too.) The comedian, whose stage name is roughly something like Velvet Scrotum, has appeared on talk shows decrying peoples’ tendency to use social media to seek advice about their problems in life. Better, he says, to visit a “snack” and discuss it with people there — you’ll always get advice, and from a real person.
I don’t think there has ever been a moment in history when a reader would have had full access to 20,000 different versions of a story. Access to news has always been limited and filtered. Some things you can only learn by being there yourself. The Internet does create a problem about the funding of in-depth journalism, but as for the broader issue you describe in this post, a lot of the problem is us: in less than 20 years, we’ve somehow come to expect that the virtual is, or ought to be, a substitute for the real — or even a heightening of it. And it’s neither of those things.
Frank Pasquale - September 22, 2012 at 1:44 pm
I hope the “Night Snacker” TV show makes it to the US (though the latest Japanese export has been disappointing: http://en.wikipedia.org/wiki/MXC).
You’re right: the question here, as always, is “compared to what.” No Nirvana Fallacies allowed. Bracha and I attempted to address that question in our paper in 2008, which I will excerpt below. At this point, I don’t have a dog in the fight as to whether the Internet public sphere is better or worse than older ones. (Even Habermas had to back down from a lot of critiques in Structural Transformation of the Public Sphere; see, e.g., http://www.nyu.edu/ipk/calhoun/files/calhounIntroductionHabermasAndThePublicSphere.pdf). I do want to point out how current systems are more opaque than they need to be, just as James Rule showed “dramatic discrepancies between … surveillance potentials—-one an ultra-sophisticated reality, the other grossly underdeveloped”–are not dictated by technological capacity, but rather, by power (at http://www.democracyjournal.org/22/the-whole-world-is-watching.php?page=all)
From “Federal Search Commission” (2008):
“Yochai Benkler defends the relative superiority of Internet speech (in comparison with old broadcasting and print models) by analyzing patterns of information flow and visibility on the network. The web, he explains, consists of multiple levels of clusters of interlinked websites; local clusters based on topic, interest, or similar criteria coalesce to form higher-order clusters. These high-order clusters are characterized by a very small number of highly visible sites and a multitude of nearly invisible ones; lower-level clusters have a small number of dominant sites too, but visibility and exposure is much more broadly and evenly distributed among the other websites.”
“The optimists concede that the old intermediaries or their Net-replicas will maintain some level of power but argue that there are also new and non-trivial alternatives for effective speech. These alternatives are claimed to constitute a much more decentralized and open model and significantly ameliorate many of the ills identified by critics of mass-media.”
***
However, “very few entities control the critical junction of Internet communication, and this situation generates problems similar to those diagnosed in broadcasting long ago. These new gatekeepers can directly manipulate the flow of information–suppressing some sources while highlighting others–whether on the basis of intrinsic preferences or in response to inducements or pressures by others. Second, the hierarchical ranking system, at least in its current one-size fits all form, has a strong bias toward majority preferences. The majority bias partly overlaps with a dominance of well-financed and commercial speakers. Third, the system tilts toward consumerist content both because consumption-oriented content-producers can more successfully induce manipulation and, more importantly, because search engines have an interest in channeling users toward sites with which they cooperate under various commercial schemes.”
***
“According to Benkler, the web functions as a decentralized, peer-based filtering system: lower-order clusters, where a large number of various speakers enjoy exposure to a community of intense-interest individuals, organically elevate a small number of sites to the attention of higher-order clusters. At the most general level, a power law distribution dictates that a small fraction of all websites receive most of the visibility. To the extent that the small group of winners was produced by the decentralized filtering system just described (and not picked by a few powerful players catering to the lowest common denominator), it should not be troubling. A mix of democracy and merit trumped plutocracy. By derivation, the same consoling logic applies to the structural bias of search engines. Search engine algorithms may give a high weight to the preferences of relatively few dominant websites in determining their rankings. But those preferences, and hence the search engine ranking that assigns them a high weight, are, to a large extent, a product of a bottom-up, “democratic” filtering system.”
“Yet this logic does not apply to more targeted manipulations by search engines. When a search engine specifically decides to intervene, for whatever reason, to enhance or reduce the visibility of a specific website or a group of websites, the decentralized filtering system may be circumvented. Instead of reflecting the synthesized results of a bottom-up filtering process, the search engine imposes its own preferences or the preferences of those who are powerful enough to induce it to act. The aggregate result of specific interventions of this kind by search engines that determine which content reaches viewers may be prejudicial to the democratic aspiration of a free, open, and diverse expressive sphere.”
Leave a Reply