What would Google do? AOL has the answer: the algorithm as editor
AOL is making plans for its post-Time Warner life that show just how news could be organised if you started with a blank canvas and two words: user data:
In December, when it becomes a stand-alone company, AOL will begin to tap a new digital-newsroom system that uses a series of algorithms to predict the types of stories, videos and photos that will be most popular with consumers and marketers.
The predictions, it says, are based on a wide swath of data AOL collects, from the Web searches people make on its site to the sites visited by subscribers to its Internet services.
The system is designed to track breaking newsand trends and identify the best times to write about seasonal events, such as Halloween or Monday Night Football.
Based on these recommendations, the company’s editorial staff, which totals about 500, will assign articles to a network of free-lancers across the country via a new Web site called Seed.com. AOL says it now works with about 3,000 free-lancers, but it is hoping to sharply increase that number through the Web site, which is open to anyone looking to submit a story.
It’s brave stuff. For years we’ve heard traditional publishers state flatly that, while user data is useful, they would never think of handing over the editorial agenda. Whether that’s pride, vanity, professionalism, or all three, AOL doesn’t have it.
And I lied: it’s not two words on that blank canvas, but 4: user and advertiser data. The article goes on:
AOL says it will pay free-lancers based on how much its technology predicts marketers will pay to advertise next to their articles or videos. It says that will range from nothing upfront, with a promise to share ad revenues the article generates, to more than $100 per item.
In addition to selling standard ads to run alongside the story or video on a Web page, AOL says it will offer custom content. For instance, AOL says, if its algorithms show consumers are searching for information about the Zhu Zhu Pets robotic hamster, a retailer could pay AOL to sponsor an article about where to find the hot toy. Some traditional media outlets, including magazines and TV studios, offer similar services.
This is Google’s auction-based contextual advertising model applied to journalism, essentially matching supply and demand from readers and advertisers to set the market rate. The one variable that is notable by its absence is the supply of journalists: AOL don’t say whether payment rates will go up if no one decides to volunteer their writing for a mere ’share of ad revenues’ (I’m guessing in that instance one of AOL’s editors will have to write it themselves – but at least they’ll be being paid. Hopefully.)
Indeed, with an upper rate of ‘more than $100 per item’ you wonder how large the supply of writers will be – yes, there’s lots of people writing for nothing online, but they generally write out of choice and for pleasure, not based on the arbitrary demand of an algorithm. And clearly, based on the number of editors they look set to employ, AOL are not expecting writers with great knowledge and talent (the payment of journalists also sounds similar to the content factories of the search engine optimisation industry).
Ryan Singel points out that Demand Media are already doing something similar. That’s true, but AOL have access to data that Demand could only dream of, along with a number of growing brands.
Ultimately, it’s a clever idea, but one that looks like it has already been taken to an extreme too far for advertisers who like to see their brand next to quality journalism. A lot rests on whether AOL can manage the churn of contributors, and the bottleneck of editing, long enough for advertisers to get used to the model. It’s a peculiarly new media model, with its own downfall built in.
