Gonzales v. Google, Inc., 234 F.R.D. 674 (N.D. Cal. 2006)

Facts

  • The United States Attorney General sought to enforce a Federal Rule of Civil Procedure 45 subpoena against Google, a non-party, to obtain data for use in litigation challenging the Child Online Protection Act (COPA).
  • The underlying COPA case required a factual record on whether filtering software was a less restrictive alternative to COPA’s restrictions.
  • The government subpoenaed multiple search engines for (1) URLs from their indices and (2) text of user search queries with identifying information removed; other providers complied.
  • After negotiations, the government narrowed its request to (1) a random sample of 50,000 URLs from Google’s current index and (2) a random sample of 5,000 search queries from a specified time period.
  • The government asserted the URL sample would help test filtering effectiveness and estimate prevalence of harmful material, and the query sample would help assess how searches might lead to harmful material and how filters would perform in that context.
  • Google objected that the subpoena was overbroad and insufficiently relevant, imposed undue technical and operational burden, threatened disclosure of proprietary information, and raised significant user privacy and trust concerns, especially as to search queries.

Issues

  1. Whether the requested URL and search-query samples were relevant to the COPA litigation and reasonably calculated to lead to admissible evidence under Rule 45.
  2. Whether compelling production would impose an undue burden on a non-party, considering expense, technical burden, confidentiality, and business harm.
  3. Whether privacy risks associated with search-query disclosure outweighed the government’s asserted need.

Decision

  • The court granted the motion to compel in part and denied it in part.
  • Google was ordered to produce a random sample of 50,000 URLs, with production subject to a protective order.
  • The court declined to compel production of the 5,000 search queries, finding the government’s showing of need insufficient in light of privacy and burden concerns.
  • Under Rule 45, subpoenaed material must be relevant to the underlying action, and courts must protect non-parties from undue burden and significant expense.
  • Non-party status increases the court’s duty to limit discovery where the marginal evidentiary value is low relative to burden, confidentiality, or privacy harms.
  • Large-scale electronic discovery may be compelled when relevance is shown and the technical and financial burden is modest, particularly with cost-shifting and protective orders.
  • Search-query text may implicate user privacy even without traditional identifiers because query content can reveal sensitive or identifying information; courts may deny compelled disclosure when privacy and trust harms outweigh incremental utility.
  • Protective orders are a standard tool to reduce competitive and confidentiality risks when limited production is otherwise warranted.

Conclusion

The court applied Rule 45’s relevance and undue-burden limits to a non-party subpoena, compelling production of a limited URL sample under a protective order while rejecting compelled disclosure of search queries due to insufficient necessity and substantial privacy and business-risk concerns.