Skip to content

Battling Black Boxes

For a researcher, it’s hard to imagine life without search engines. I’m consistently amazed at the range of information they bring to us. And they make many other parts of life easier as well.

But I worry about growing reliance on search engines. How is my worldview being biased by them? Do search engines have an interest in getting certain information to me? Or blocking me from seeing other information?

These may seem like speculative worries, but a recent book on Google (by eminent search blogger John Battelle) gives some concrete examples. For example, he tells the story of the owner of 2bigfeet.com (a seller of large-sized men’s shoes), whose site was knocked off the first page of Google’s rankings by a sudden algorithm shift in November, 2003, right before the Christmas buying season. Moncrief attempted to contact Google several times, but “never got a response.”

Google, of course, has every right to change its search algorithm. It needs to–search engine optimizers are constantly trying to game the system. But should the algorithm be a black box, as defenders of search engine autonomy maintain? What if web-reliant businesses begin to suspect a “churn” in the top organic rankings that is designed to force them to buy “adwords” to maintain a stable position at the top of certain pages? Might there be some room for some agency to look into complaints?

I think there might be. In the next few days, I’ll try to give some reasons why, and how it might work. For now, I just ask you to consider this a subset of a larger set of quandaries created by “black box” sources of information in commercial life. For example, Altria has used trade secret law to block disclosures of cigarette ingredient levels. Credit bureaus have repeatedly resisted calls for transparency in FICO scoring, despite worries that these scores could be racially driven. Ratings agencies like Moody’s completely missed the Enron debacle, and it’s hard to hold them accountable.

True, these companies’ business models depend on IP protection for crucial data. But protection against gaming or unfair competition should not be a shield against all scrutiny of whether such companies’ practices actually amount to what they are represented as.