The California Supreme Court recently issued a ruling on the seemingly mundane issue of whether public agencies may charge Public Records Act (PRA) requesters for the costs of redacting exempt material from electronic records when responding to PRA requests. Although the decision clarifies that agencies must bear the cost of redacting exempt material from both paper and electronic records, it does not offer a bright line rule for a secondary, but crucial, issue in the case: What agency processes do, and do not, qualify as "data extraction," the costs of which agencies may permissibly shift to requesters? The opinion does provide some "guideposts," but does not offer entirely easy answers for public agencies trying to determine what—or whether-they may charge requesters when responding to electronic PRA requests.

|

The Public Records Act

The California Public Records Act requires requesters to pay agencies for the "direct" duplication costs for the production of paper records-namely "the cost of running a copy machine, and conceivably the cost of the person operating it." However, agencies must bear "ancillary" costs, including staff time for searching records and identifying and redacting exempt information.

In 2000, the Legislature amended the PRA, introducing rules specific to the production of electronic records. The amendments did not change the ordinary rule; requesters must still pay the direct duplication costs of producing electronic records (limited to the direct cost of producing the record in electronic format). However, the amendments shifted the cost for production of electronic records to the requester where the request requires "data compilation, extraction, or programming to produce the record."

In National Lawyers Guild v. City of Hayward, issued by the California Supreme Court on May 28, 2020, the court considered the narrow question of whether the term "extraction" includes the process of redacting exempt material from electronic records. The court concluded that "extraction" does not include redaction. Amici for the Petitioner included the ACLU of Northern California, Lawyers' Committee for Civil Rights, Reporters Committee for Freedom of the Press, and California News Publishers Association. Amici for the Respondent included the League of California Cities and California State Association of Counties.

|

The Facts

In 2014, the Hayward Police Department assisted the City of Berkeley in policing protests. Plaintiff National Lawyers Guild later submitted a PRA request to the Department, seeking records related to the Department's actions in policing the protests. The Department's Custodian of Records (COR) suspected certain videos from body cameras worn by officers policing the demonstrations were responsive. The City's IT department, using criteria provided by the COR, identified 141 videos, totaling approximately 90 hours. A cursory review of the videos revealed they contained exempt material. Recognizing that editing 90 hours of video would be extremely burdensome, the City asked NLG to narrow its request. In response, NLG requested six hours of video.

The COR then proceeded to edit out exempt material from the six hours of video. This process involved a number of steps, including separating the audio and visual components of the files, saving the audio material in a new file format, and saving the edited videos as new files. The whole editing process took 35.3 hours. The NLG later requested additional footage, for which the COR used the same editing procedure.

Between the two requests, the City charged NLG over $3000, which included the cost of editing time and time spent searching for responsive video. The City contended that the editing process constituted "data extraction," the costs of which could therefore appropriately be shifted to NLG.

|

The Court's Analysis

The City argued that "extraction" ordinarily means "taking something out," and therefore includes the process of redacting, or taking out, exempt material from a responsive record. By contrast, the NLG pushed for a narrow interpretation, arguing that "extraction" is limited, in this context, to "retrieving responsive information from a government repository in order to produce the responsive information in a newly constructed record." As the court recognized, common dictionary definitions of "extraction" supported both parties' interpretations. However, finding that the provision in question is "technical" in nature, the court found it more appropriate to consider technical definitions of "extraction," such as those found in computing dictionaries and other technical sources. NLG's interpretation was consistent with these more technical definitions, which generally refer to "a process of retrieving required or necessary data for a particular use, rather than omitting or deleting unwanted data."

Similarly, the court found that the legislative history of the 2000 amendments reflected the Legislature's concerns about "the difficulties associated with retrieving responsive data from massive, potentially intractable databases," but offered little support that the Legislature intended "extraction" to cover the cost of redacting electronic records. Relatedly, the court pointed out that the City's broad interpretation of "extraction" would create an unsupportable distinction between paper and electronic records, allowing agencies to charge for time spent redacting electronic records, but not paper ones.

Finally, the court considered article I, section 3, subdivision (b)(2) of the California Constitution, which requires a statute to be broadly construed if it furthers the people's right of access to information concerning the conduct of the people's business. Because redaction costs "are often nontrivial"—as evidenced here, where the cost of redaction exceeded $3000—allowing agencies to shift the costs of redacting electronic records to requesters could "erect[] … substantial financial barriers to access." Practical Implications

What is extraction, exactly? The court acknowledged that it could not "comprehensively catalog" the types of processes that will or will not qualify as "extraction." The "paradigmatic example" is when "the government agency is required to pull certain data from a large database in order to construct a record that can be disclosed to the requester," for example, "pulling demographic data for all state agency employees from a human resources database and producing the relevant data in a spreadsheet." However, "extraction" does not include "every process that might be colloquially described as 'taking information out,'" including time spent searching for responsive records in an e-mail inbox or computer's documents folder. In this case, the City implicitly conceded that the IT department did not perform data extraction when it performed the initial search for responsive videos. Indeed, the court likened this task to "searching a filing cabinet for responsive paper records." And in terms of the COR's video-editing process, despite its somewhat technologically complicated nature, the court held that "what [the COR] did was not substantively different from using an electronic tool to draw black boxes over exempt material contained in a document in electronic format."

Although the opinion does not draw particularly bright lines, the court's definition of data extraction does seem to require the construction of a new record, e.g., the new spreadsheet that is created after the requested human resources data is extracted from the larger database. This may be a clear distinction in some cases, and arguably more subtle in others. Here, the court found that the City's COR did not extract data in order to produce new videos, but instead merely deleted exempt data from existing videos. However, based on the court's description of the COR's process, the COR did not simply edit the existing videos, but had to splice material out of the existing videos, create new files, reupload those new files into the editing software, etc.

One takeaway, especially in an era of increased reliance on electronic files and databases, is that public agencies will need to continually consider whether the processes they are using to gather electronic records and data are extractive. Likewise, unless the agency's process clearly matches the paradigmatic example provided in National Lawyers Guild, the agency should expect requesters may challenge charges made for data extraction. Finally, public agencies should carefully consider-and consult with agency counsel about-potential PRA-related ramifications when evaluating whether and how to use a new technology that will generate public records.

Overly burdensome requests. The court acknowledged the concern that agencies might not have adequate funding to undertake redactions of overly burdensome PRA requests. However, it pointed to provisions of the PRA that already address this problem. For example, the PRA "requires agencies to disclose nonexempt portions of records only if they are 'reasonably segregable' from portions exempted by law." Likewise, agencies may withhold records where the public interest served by not disclosing outweighs the public interest of disclosure. Finally, public agencies may "[p]rovide suggestions [to requesters] for overcoming any practical basis for denying access to the records or information sought."

To the extent that responding to requests for body camera footage "present[s] unique concerns for government agencies with limited resources," the court held that only the Legislature can decide whether these unique burdens warrant special treatment.

Sarah M. Lucey is an environmental law fellow with Shute, Mihaly & Weinberger in San Francisco.