Of late, it seems like there are an awful lot of companies (Google News, Findory, Netvibes etc) betting that folks would love an online news reading experience that is customized based on browsing history. And while the "customer's like you also prefer" approach has generated truckloads of cash for Amazon and a handful of other online retailers, I'm not sure if its the right approach to customizing news.
First of all, retailers are working with much stronger signal. A purchase decision indicates a level of commitment to a product far greater than that of a casual click on a clever headline.
Second, recommendation engines show you as you are, not as you wish to be. Though it pains me to admit this, given the choice between "Britain sends note to Iran amid threats of legal action" and "Cops: Man caught with 1,500 undies," I seldom select the serious story. But please understand, that - in my own mind - I am a serious news consumer, concerned only with the most significant of world events. Were a news recommendation engine to suggest stories based on my penchant for exploding whales, IT mishaps, and gentle giants saving dolphins, I would haughtily puff out my cheeks and derisively dismiss these childish recommendations. Like I said, I am serious news consumer.
So maybe the answer is to make recommendations that readers don't recognize as such. By making the reader feel that they've serendipitously stumbled upon an irresistible bit of fluff, media sites will get the traffic they need and users can preserve their fragile self image.