On May 13, the European Union's Court of Justice decided that a Spanish national had a right to have removed from search engine results references to a decade-old newspaper story concerning a foreclosure auction held to satisfy his debts. Seventeen days later, Google complied with the decision by posting a dead-simple interface for capturing such so-called "right to be forgotten" requests.
Two months down the road, it's clear that building the Web site was the easy part, and not just because it was based on Google Forms. A new 13-page filing from Google to European data protection authorities makes plain that far trickier has been implementing a scalable, sustainable and equitable way of disappearing, per the court, "inadequate, irrelevant or no longer relevant, or excessive" search results on demand; writes the company, "in many cases we lack the larger factual context for a request, without which it is difficult to balance the competing interests."
In some ways, that ignorance might be Google's own doing: No matter how many URLs are being contested, the text box for explaining deletion requests maxes out at 1,000 characters.
But in other ways, suggests Google's response to European investigators, it is simply impossible for Google to ever understand the nuances about which it is being required to adjudicate. Google must, for example, figure out whether the "data subject" — the person requesting the take-down — has a legitimate tie to one of the 28 countries covered by the ruling. Moreover, it must take into consideration whether the publisher of the source information is "reputable," and do it in at least 25 different languages. It also must attempt to suss out whether the subject of the request is a public figure in a manner relevant to the subject at hand. And it must, in some cases, divine motivation.
"It can be difficult to draw the line," reads the company's report, "between significant political speech and simply political activity." If someone asks for the take-down of a link to a photo of them at a rally, is it because of the rally became controversial or because they simply don't like the photo? It can be challenging to know, reports Google, and each case must be assessed individually.
And yet, amid that information gap, Google has opted to approve the majority of requests for removals: more than half — 53 percent — of them are granted based on the first application, and an additional 15 percent are granted after more information has been requested and submitted.
Add into the mix the fact that Google has decided that it generally cannot let publishers know exactly why links to their content have been removed from search results because that in itself would be a breach of the data subject's privacy. Notice, says Jules Polonetsky , executive director of the D.C.-based The Future of Privacy Forum, is desirable as a means of giving publishers recourse.* "Maybe this," says Polonetsky of a data subject who successfully moves to get his search results sanitized, "is someone that a local Bavarian newspaper thinks is a future candidate for office," and thus coverage, even if critical, is warranted. It's left to Google, though, and not the Bavarians, to make that determination.
Is the answer to accept as the natural order that Google is going to act as adjudicator and simply figure out ways to provide the search engine with greater context? It isn't clear that that's scalable, either. Google says that as of July 18, it has responded to 91,000 removal requests involving 328,000 different Web pages. That's a considerable number of individual assessments to make, and it's not a done-one, done-forever situation, either. The court found that "even initially lawful processing of accurate data may, in the course of time, become incompatible with the directive," which means that a company like Google must be constantly aware of the ever-evolving meaning of its search results.
(Which, it's worth noting, puts any would-be next-generation search engines in a bit of a bind; getting a team in place to vet such requests would likely add considerably to start-up costs.)
"It's going to be awkward and messy," says Polonetsky, who adds that he believes that its the sort of decision-making best left to the courts.
Google seems to think so, too, but in the absence of that reality, the company has put together an "advisory council of experts" to help design the "right to be forgotten" moving forward. That group includes the editorial director at Le Monde, the head of a Poland-based privacy group, and Wikipedia's Jimmy Wales. It's a somewhat remarkable shift: Those of us who might have felt at the mercy of an uncaring Internet machine that applies its judgments en masse to the mass of online content are now, in much of Europe at least, at the mercy of a hyper-invested Google making one-off judgments about who is represented how online.
Update: These two sentence have been adjusted to clarify the particulars of Google's reporting to online publishers about search result removals and Mr. Polonetsky's comments in that context.