No Grounds for Google Objection to Search Request, DoJ Says

Govt. attorneys sounded clearly frustrated that their request, honored by all major search engines except Google, was being portrayed as an intrusion against personal privacy. The govt. has requested “the text only, without any additional identifying information,” of search queries entered into Google “only to perform a study, in the aggregate, of trends in the Internet.” The govt. said Google contradicted itself in its last filing -- saying the text of queries won’t reveal information about the source of those queries, but also that “proprietary demographic information” about its users would be revealed in a sample of search terms.

Google’s contention that search queries and resulting URLs have “no conceivable use” in the govt.’s case for the Child Online Protection Act (COPA) in U.S. Dist. Court, Philadelphia is irrelevant and misunderstands the case, DoJ said. “A court should be reluctant to accept a non-party’s view as to what evidence is relevant to an action, particularly when that action is pending before a different judge in a separate district,” DoJ said: “The Internet may be a fluid entity, and the content of the websites associated with particular URLs may be subject to change, but the drawing of a random sample of URLs at a particular moment in time would still produce a relevant sample for the study.”

Google’s trade secrets are in no danger of disclosure, the agency said. If the company had read the protective order in the Philadelphia court more closely, Google would have seen that “the order also affords it protections from the disclosure of confidential materials at trial,” a response to Google’s “open court” disclosure concern. It rebuffed Google’s suggestion that the govt.’s expert witness, Philip Stark, would use the confidential information for the benefit of Cogit.com, a Google rival he has connections with.

But the govt. doesn’t even need to show that evidence gleaned from Google must be admissible at trial, DoJ said, citing a San Jose court decision involving Compaq, since acquired by HP. “Even if Google had [shown its trade secrets were at risk], the government would still be entitled to disclosure” under the judicial presumption that the govt.’s evidentiary needs outweigh any private interest, DoJ said. As the largest search engine -- also considered by some one of the largest gateways “to access pornography” online -- “Google is of course a relevant population for the purpose of evaluating the character of the Internet.”

Regarding the burden to Google of complying with DoJ’s request, DoJ noted the company already has query logs, has produced index samples in the past, and regularly generates and posts query reports, as on the Google Zeitgeist page. Google’s estimate of 3-8 days for its engineers to comply is probably overstated, DoJ added. Any bickering over the definition of “random sample” is limited by “well established [standards] in the scientific field of statistics,” and the govt., not Google, would have to defend its methodology in court if questioned. The govt. will compensate Google for any “reasonable expenses” in complying, and the chance that drawing the random sample would interfere with Google’s search engine operations, as the firm claimed, is “vanishingly small,” DoJ said.

Google reveals information from searches “routinely” with no fuss, DoJ emphasized, and has no provision in its privacy policy for “aggregate non-personal information.” The search terms entered into Google will display in the user’s browser by default on the results page, and if a user clicks a link from the results, even the user’s IP address will be transmitted to the linked site’s owner because of the HTTP specification, DoJ said: “Google could easily prevent users’ search terms from leaking out,” but doesn’t.

What Does ECPA Have to Say?

The Electronic Communications Privacy Act (ECPA) has no bearing on search queries and URLs, DoJ said, noting Google didn’t cite case law or legislative history to support its argument that the request raises a “substantial question” about ECPA. “Websites are users of communications services, rather than providers” such as ISPs, which were later added to phone companies and e-mail providers as covered entities under ECPA. Amazon.com was excluded from the “provider” category in a 2001 Cal. case, DoJ said. Similarly Google can’t claim itself as a “remote computing service” (RCS), which “handles outsourced computer storage or processing”; by Google’s definition, any website would qualify for that ECPA category. Even if Google were an RCS, it doesn’t store communications for the customer’s sake, just its own business interest, which isn’t covered, DoJ added.

Treatment of search queries and results under ECPA has never been considered in federal court, the Stanford brief said, asking the court to accept additional briefs if it decides to rule on this claim: “Otherwise, there is a real danger [the finding] will set precedent on the statutory construction and scope of ECPA” absent a full legal airing. Other courts have said ECPA is “infamous… for its lack of clarity,” given the technology changes in the 20 years since its passage, the brief said.

Noting its president’s service on the COPA Commission created by Congress and its briefs filed in COPA litigation, CDT said ECPA “provides a clear and straightforward answer” on the govt.’s request to Google. Countering the govt.’s position, CDT said Google is indeed an RCS: “Internet users could, with considerable effort, maintain their own lists of URLs and send out their own robots or spiders to scour the Web looking for what they want.” But they outsource that task to Google for convenience -- “no different than a company outsourcing its payroll operations” to a firm like Electronic Data Systems, which testified in the hearings preceding ECPA’s enactment. But CDT distinguished Google from Amazon.com, saying the latter doesn’t “transform” and return customer-generated data “as new, different and useful data.”

RCS’s can’t be subpoenaed for customer communications, which includes Google search terms and results, CDT said. The govt. hasn’t used any of 5 methods authorized by ECPA to compel disclosure from RCS’s: Criminal search warrant; administrative subpoena; grand jury subpoena; trial subpoena; or a court finding that the information sought is “relevant and material to an ongoing criminal investigation.” CDT cited a San Jose judge’s previous rejection of the use of discovery subpoenas on RCS’s.

CDT also said the global nature of the Internet rendered DoJ’s discovery effort, and COPA itself, fatally flawed: “Analyzing a snapshot of search terms entered will not lead to any valid conclusions about how minors in the United States… gain access to sexual content.” Similarly, sexual content hosted outside the U.S. -- as most is, a 2002 National Academy of Sciences study found -- won’t be affected by DoJ’s request or COPA, CDT said.