NATIONAL LAWYERS GUILD, SAN FRANCISCO BAY AREA CHAPTER, Plaintiff and Respondent, v. CITY OF HAYWARD et al., Defendants and Appellants.
S252445
IN THE SUPREME COURT OF CALIFORNIA
May 28, 2020
First Appellate District, Division Three A149328; Alameda County Superior Court RG15785743
Justice Kruger authored the opinion of the Court, in which Chief Justice Cantil-Sakauye and Justices Chin, Corrigan, Liu, Cuellar, Groban concurred.
Justice Cuellar filed a concurring opinion.
NATIONAL LAWYERS GUILD v. CITY OF HAYWARD
S252445
This case concerns the costs provisions of the California Public Records Act (
I.
A.
The California Public Records Act (PRA or Act) establishes a right of public access to government records. “Modeled after the federal Freedom of Information Act (
The Legislature that enacted the PRA recognized that increased access to government information can have both intangible and tangible costs, and it crafted the PRA accordingly. First, and most important, the Legislature recognized that increased public access to government records can come at the expense of personal privacy and other important confidentiality interests. To mitigate these sorts of intangible costs, the Legislature crafted “numerous exceptions to the [PRA‘s] requirement of public disclosure.” (International Federation of Professional & Technical Engineers, Local 21, AFL-CIO v. Superior Court (2007) 42 Cal.4th 319, 329, citing
At the same time, the Legislature also recognized that increased public access to government information has costs of the more tangible, dollars-and-cents variety. Before providing access to requested records, public agencies need to locate and collect records, determine which records are responsive, determine whether any portions of responsive records are exempt from disclosure, convert the records into a reviewable format, and, if requested, create a copy of the record. To complete these tasks generally requires personnel time as well as the use of office equipment and supplies—all of which comes with a price tag. The PRA acknowledges as much and allocates certain costs to the requester, while others must be borne by the agency responding to the requests.
Precisely which costs may be allocated to the requester depends on the format of the requested record. Since 2000, the PRA has distinguished between nonelectronic records (sometimes referred to as “paper records,” though the record may be in another nonelectronic medium, such as audiotape) and electronic records. Paper records are governed by a general costs provision, enacted in its earliest form by the original statute in 1968. (
Before the statute was amended in 2000, there were no special rules for records kept in electronic format. Agencies had wide discretion to produce electronic records “in a form determined by the agency“—that is, in any form the agency saw fit. (
To account for differences in the costs of producing electronic versus paper records, the 2000 amendment introduced specific rules for the production of records held in electronic format. (Stats. 2000, ch. 982, § 2, p. 7142; see Sen. Com. on Judiciary, Analysis of Assem. Bill No. 2799 (1999–2000 Reg. Sess.) as amended June 22, 2000, pp. 3–4.) In newly added
After the 2000 amendments, the ordinary rule is the same for electronic records as paper records: Requesters must pay direct duplication costs (although the statute now specifies that in the case of electronic records, the “cost of duplication shall be limited to the direct cost of producing a copy of a record in an electronic format“). (
B.
In December 2014, demonstrations erupted in Berkeley, protesting grand jury decisions not to indict the police officers involved in the deaths of Eric Garner and Michael Brown, both unarmed African-American men. The Hayward Police Department provided mutual aid to the City of Berkeley in policing the demonstrations. After the demonstrations were over, plaintiff National Lawyers Guild, San Francisco Bay Area Chapter (NLG) submitted a public records request to the Department, seeking 11 categories of records relating to the Department‘s actions in policing the demonstrations. The requested records included relevant communications made during the demonstrations, operations and command center logs, and various reports, as well as records identifying supervisory and command officers who had approved certain police tactics used at the demonstrations and records relating to the use of those tactics. Soon after, NLG submitted a followup request for related records.
The Department‘s Records Administrator and Custodian of Records, Adam Perez, was responsible for identifying records responsive to the requests. For both requests, Perez first identified responsive text-based electronic records, such as written reports, logs, operational plans, and e-mails. He reviewed these documents for potential exemptions under the PRA and redacted them accordingly. He then converted the documents to portable document format (PDF), and they were e-mailed to NLG. NLG was never charged the costs to produce the copies of these text-based electronic records.
Perez next identified other types of electronic records potentially responsive to NLG‘s requests. Several Hayward officers policing the demonstrations were equipped with body-worn cameras. Though NLG had not explicitly requested videos from these cameras, Perez believed certain videos might be responsive. In the City of Hayward, police officers upload digital video from their body-worn cameras to an online digital evidence management system known as Evidence.com, which stores videos and other digital evidence on the Internet. From Evidence.com, videos can be downloaded in MP4 format to DVDs for production, storage, or other uses. On average the City collects more than 1,000 hours of body-worn camera video per month.
Because Perez did not have access to Evidence.com, he asked the City‘s Information Technology Manager of Public Safety, Nathaniel Roush, to search Evidence.com for videos responsive to NLG‘s requests. Perez provided Roush with 15 search criteria, and Roush searched Evidence.com using these
Perez reviewed the videos to determine whether they contained material exempt from disclosure under the PRA. After a cursory review, he concluded they contained exempt material, including personal medical information and law enforcement tactical security measures.2 (See
With the narrower set of videos in hand, Perez began the editing process. First, he identified the exact visual and audio segments that were exempt. Next, he used Windows Movie Maker to remove all exempt audio and visual material from the video files. Before he could remove the exempt audio segments, he had to separate the audio and visual material by taking out all of the audio material from each MP4, saving that audio material as an MP3, and reuploading the MP3 audio file into Windows Movie Maker. Last, he saved the edited videos as new MP4 files and downloaded them to a thumb drive storage device. This editing process took Perez 35.3 hours.
The City Attorney then informed NLG that the videos were available for pickup. But the City Attorney warned NLG that before anyone could pick up the videos NLG would need to pay the City‘s costs to produce the videos. The City invoiced NLG $2,938.583—$1 for the “DVD” (actually a thumb drive) containing the edited video copies and the remainder for 40.2 hours of staff time spent preparing the videos for production, consisting of 4.9 hours of Roush‘s time and 35.3 hours of Perez‘s time, as detailed above. NLG paid the invoiced amount under protest and received the videos.
After requesting the second set of videos, but before receiving them, NLG filed a petition for declaratory and injunctive relief and writ of mandate against the City and relevant City officials (collectively, Hayward). NLG sought a refund of the money it had paid to receive the first set of videos and a writ of mandate or injunction requiring immediate production of the second set of videos without costs beyond those necessary to copy the videos. Later, after paying for and receiving the second set of videos, NLG moved for a peremptory writ of mandate, arguing that Hayward‘s charges were excessive and seeking a refund of the money it had paid beyond the direct costs of duplicating the videos. Hayward argued in response that the invoiced costs were justified under the PRA because the City‘s staff had performed data extraction and compilation, as allowed under
The trial court disagreed with Hayward, holding that “the phrase ‘data compilation, extraction, or programming to produce the record’ ” does not include “making a redacted version of an existing public record.” Instead, this exception “applies only when a []PRA request requires a public agency to produce a record that does not exist without compiling data, extracting data or information from [an] existing record, or programing a computer or other electronic devise [sic] to retrieve the data.” The trial court thus found that Hayward‘s charges were unjustified and granted the petition for writ of mandate, directing Hayward to refund to NLG the charges for the City‘s staff time.
The Court of Appeal reversed, agreeing with Hayward that
II.
A.
The issue before us is one of statutory interpretation, so we begin by looking to the statutory language. If the language is clear in context, our work is at an end. If it is not clear, we may consider other aids, including the statute‘s legislative history. (Sierra Club v. Superior Court (2013) 57 Cal.4th 157, 165–166 (Sierra Club).)
The PRA provides that public agencies may recover the costs associated with producing a copy of an electronic record, “including the cost to construct a record, and the cost of programming and computer services necessary to produce a copy of the record” (
As the Court of Appeal in this case observed, both views find some support in common dictionary definitions of “extraction.” The verb “extract” is commonly defined to mean “to draw forth” or “to pull out (as something embedded or otherwise firmly fixed) forcibly or with great effort.” (Webster‘s 3d New Internat. Dict. (2002) p. 806 (Webster‘s Third).) This dictionary definition is capacious enough to encompass Hayward‘s broad interpretation as well as NLG‘s narrower one. (National Lawyers Guild, supra, 27 Cal.App.5th at pp. 947–948.)6
But general-purpose dictionary definitions are not always the most reliable guide to statutory meaning; sometimes context suggests that the Legislature may have been using a term in a more technical or specialized way. (See, e.g., Nelson v. Dean (1946) 27 Cal.2d 873, 879.)
between “data compilation” and “programming.” (
In the field of computing, the term “data extraction” does encompass a process of taking data out, but it is generally used to refer to a process of retrieving required or necessary data for a particular use, rather than omitting or deleting unwanted data. One computing dictionary, for example, defines the term “extract” as meaning “to remove required data or information from a database.” (Collin, Dict. of Computing (4th ed. 2002) p. 139, italics added; cf.
NLG‘s view aligns with this more technical usage of the term “extraction,” as well as with the particular context in which the term appears in
The process to which Hayward refers, by contrast, is not unique to the field of electronic records; redacting exempt material is a process common to the production of virtually every kind of public record, whether in paper or electronic format. The PRA has long had a term for this process: “deletion.” (
As a practical matter, reading
Responding to this concern at oral argument, counsel for Hayward emphasized that one general definition of “extraction” refers not just to “taking something out,” but to “taking out” with “special effort.” Counsel suggested we could therefore construe
cover redacting records in PDF, a task that is much simpler and requires less specialized technology and expertise. Moreover, courts could conclude that redactions that count as “extraction” today may not count as “extraction” tomorrow: Although the video redaction at issue here might have required special effort in 2015, advances in technology may one day make video redaction more routine and thus not chargeable as data extraction costs.
We doubt the Legislature intended us to read quite so much into the bare term “extraction.” A different provision of the PRA,
Whatever problems its own interpretation may have, Hayward argues that NLG‘s interpretation is unsupportable insofar as it would limit “extraction” to responses requiring the retrieval of data for purposes of constructing a record for public release. In Hayward‘s view, this should be a null set, because, as a general rule, the PRA (like the federal Freedom of Information Act, on which
Hayward‘s argument misunderstands the rule described in Sander. The PRA does sometimes require agencies to construct records for public release.
The rule to which Hayward refers is not a general prohibition on constructing records, as such, but rather a prohibition on requiring agencies to generate new substantive content to respond to a PRA request. The rule means that, for example, agencies need not draft summary or explanatory material, perform calculations on data, or create inventories of data in response to a records request. (See, e.g., Haynie v. Superior Court (2001) 26 Cal.4th 1061, 1075 [“Preparing an inventory of potentially responsive records is not mandated by the []PRA.“]; see also, e.g., NLRB v. Sears, Roebuck & Co. (1975) 421 U.S. 132, 161–162 [“The [Freedom of Information] Act does not compel agencies to write opinions in cases in which they would not otherwise be required to do so. It only requires disclosure of certain documents which the law requires the agency to prepare or which the agency has decided for its own reasons to create.“]; Students Against Genocide v. Department of State (D.C. Cir. 2001) 257 F.3d 828, 837 [rejecting argument that agencies must “produce new photographs at a different resolution in order to mask the capabilities of the reconnaissance systems that took them“].) But the rule does not mean that an agency may disregard a request for government information simply because the information must first be retrieved and then exported into a separate record before the information can be released.
Sander, supra, 26 Cal.App.5th 651, itself explained the distinction. Plaintiffs there requested records reflecting California Bar Examination applicants’ personally identifying characteristics, like race, law school, grade point average, bar exam score, and year of law school graduation. (Id. at p. 655.) To protect applicant privacy, the requester-plaintiffs proposed four different protocols the agency could use to “de-identify or ‘anonymize’ ” the data requested. (Id. at p. 658.) Each of these protocols ” ‘require[d] the State Bar to recode its original data into new values’ ” (id. at p. 667 [quoting trial court]), including through “recoding and binning”8 data (id. at p. 659), “[data] suppression (removing information from data that might be identifying),
extraction or computer programming if “necessary to produce a copy of the record.” (Ibid., quoting
In short, NLG‘s interpretation is more than supportable; it is the interpretation that more readily comports with the statutory text. Under that interpretation,
B.
We turn, then, to the legislative history. As explained above, before the Legislature enacted
Nothing in the legislative history explains precisely what the Legislature meant by its use of “extraction” in the special costs provision, but this omission is itself telling. The overarching motivation for
To the extent we can discern anything instructive from the legislative history, the lessons are generally consistent with NLG‘s view that the Legislature was primarily concerned with the costs of retrieving information from government stores, as opposed to time spent redacting exempt information. For example, in discussing
Hayward points to other portions of the legislative record in an effort to show the Legislature intended “extraction” to cover redaction costs. Hayward
Nothing in the record supports this inference. The opposition letters, of course, reflect only the opinions of their writers—all interested outside parties—and not those of the Legislature. (See Hassan v. Mercy American River Hospital (2003) 31 Cal.4th 709, 723 [“letters state the views of the writers, not the intent of the Legislature,” absent “support for [the proposed] interpretation from any source within the Legislature itself“]; Altaville Drug Store, Inc. v. Employment Development Department (1988) 44 Cal.3d 231, 238, fn. 6; cf. People v. Dennis (1998) 17 Cal.4th 468, 501, fn. 7 [declining to take judicial notice of letters in support of a bill in part because they “simply state[d] the views of two groups specially interested in supporting the bill‘s passage“].) Hayward does point to a pre-amendment “Question and Answers” sheet by the bill‘s author acknowledging the letter writers’ concerns. But nothing in that document, or any other document in the available legislative history, indicates the Legislature shared—much less acted on—the writers’ concerns about the costs of electronic redaction.
Nor is it fair to infer from the timing that
It is true, as Hayward notes, that many of the groups that had previously opposed the bill withdrew their opposition after
In sum, the legislative history offers little support for Hayward‘s proposed interpretation of
Neither the text of
To the extent any doubt remains, California‘s constitutional directive to “broadly construe[]” a statute “if it furthers the people‘s right of access” confirms our conclusion that redaction costs are not chargeable as costs of data extraction. (
Hayward counters that shifting costs to the requester would actually improve public access to electronic records. Hayward theorizes that allowing agencies to recoup redaction costs reduces the overall burden on the agency, which in turn allows the agency to (1) produce records more quickly; (2) redact records with greater fidelity to any claimed exemptions; and (3) rely less frequently on the catchall exemption in
While we do not doubt that greater funding for PRA compliance would yield many of the access benefits Hayward describes, we are not convinced that shifting redaction costs to requesters is the right way to secure those benefits under the statute. Redaction costs could well prove prohibitively expensive for some requesters, barring them from accessing records altogether. Even if higher costs to the agency mean slower disclosure rates or greater inconvenience to the requester, these burdens on access are insignificant if the alternative is no access at all.
To the extent Hayward is concerned about being made to respond to overly burdensome requests without adequate funding, the PRA does provide various solutions to ease those burdens. For example,
But no similar provisions protect requesters from costs that unduly burden their right of access to government information. Consideration of that right favors a rule that avoids shifting routine redaction costs as a condition of gaining the access the PRA promises.9
Hayward argues that requests for body camera footage present unique concerns for government agencies with limited resources. We do not doubt the point. Video footage has a unique potential to invade personal privacy, as well as to jeopardize other important public interests that the PRA‘s exemptions were designed to protect. Redacting exempt footage can be time-consuming and costly. But
III.
Applying this understanding here, we conclude the trial court was correct to disallow the City‘s charges for time its staff spent responding to NLG‘s requests.
The City charged for Nathaniel Roush‘s time spent searching Evidence.com for responsive videos, reviewing videos, downloading them to DVDs, and confirming their download. Roush never edited the videos; more specifically, he did not extract responsive data from any video. Hayward does not argue Roush performed data extraction with respect to the videos. We agree with this implicit concession. Roush‘s tasks of searching Evidence.com for
The City also charged for Adam Perez‘s time spent editing the videos. But to the extent Perez merely deleted exempt data from the videos (i.e., redacted them), he did not “extract[]” data in order to produce new videos within the meaning of
Hayward raises one final argument to justify at least some of its charged costs: It argues that Roush performed “data compilation,” as the term is used in
IV.
We reverse the judgment of the Court of Appeal and remand for further proceedings consistent with this opinion.
KRUGER, J.
We Concur:
CANTIL-SAKAUYE, C. J.
CHIN, J.
CORRIGAN, J.
LIU, J.
CUÉLLAR, J.
GROBAN, J.
NATIONAL LAWYERS GUILD v. CITY OF HAYWARD
S252445
Concurring Opinion by Justice Cuéllar
The majority opinion concludes that when City of Hayward employees spent hours editing out portions of digital body camera footage that were exempt from disclosure, those hours didn‘t fall within the ambit of data “extraction” encompassed by
The California Public Records Act (PRA;
But because such electronic data can be stored in nearly infinite ways, jurisdictions such as Hayward can respond to public records requests using technologies that continue to evolve. Imagine a not-so-distant future when government entities deploy more thoroughly automated, artificially intelligent systems for responding to PRA requests. Such systems would likely weave into a nearly seamless quilt –– either because of the software‘s design and functionality, or because of how the relevant data were classified –– the search of government databases for responsive records, their extraction from the databases, and the editing of portions of the data exempt from disclosure. Such technology could readily help agencies be more accurate, efficient, and thorough in responding to public records requests — and allow members of the public to receive quicker access to government records. (See Gomez, MuckRock Request Data Shows Big Difference in Backlogs Between States (Mar. 21, 2019) Muckrock <https://www.muckrock.com/news/archives/2019/mar/21/feature-state-data/> [as of May 26, 2020] [average response times for state public records requests filed through one organization range from 11 days in Vermont to 148 days in Oregon].)2
This technology will also merit nuanced application of statutory provisions such as the one at issue here. A “paradigmatic example of when
Someone eventually needs to pay for the development, refinement, and maintenance of such technologies — even in a world where people and firms extensively use open source software and loss leading products. Although certain now-familiar business models pivot on presenting the monetary costs of these systems to users as low enough to appear negligible or even nonexistent, such products may impose a host of subtle or unexpected costs in other forms. As we‘ve observed, products that “attract[] users with “free” and low-priced services” may in fact lock in dependence on expensive support services, or enable private companies “to mine, exploit, and market their users’ data to third parties.” (Day & Stemler, Infracompetitive Privacy (2019) 105 Iowa L.Rev. 61, 63, fn. omitted; see also Newman, The Myth of Free (2018) 86 Geo. Wash. L.Rev. 513, 563 [product users “systematically underestimate the amount of information costs they are willing to incur in exchange” for products that are advertised as “free“].) That software offered by such business models may be suitable for public agencies in some situations doesn‘t remotely mean it would make sense in every instance. (See, e.g., Paquette et al., Identifying the Security Risks Associated with Governmental Use of Cloud Computing (2010) 27 Gov. Inf. Q. 245, 251 [“prevent[ing] unauthorized access to both data and code” and the “[p]reservation of information and documents” are among the risks associated with the government‘s use of cloud services and third party software]; Schooner & Greenspahn, Too Dependent on Contractors? Minimum Standards for Responsible Governance (2008) 6 J. Cont. Mgmt. 9, 14 [among the challenges of privatizing government responsibilities is the dependence of agencies on contractors for service and support].) Click-wrapped gift horses are best looked in the mouth.
Government agencies willing to do so may often find that what‘s most consistent with their public mission is not to opt for the system with the cheapest sticker price. They may instead take best account of the full range of interests and concerns by selecting products that require subscriptions or otherwise involve greater up-front expenses but allow for greater certainty about long-term costs or otherwise evince fidelity to the civic values at stake. (Cf. Re & Solow-Niederman, Developing Artificially Intelligent Justice (2019) 22 Stan. Tech. L.Rev. 242, 285 [advocating for the use of technologies that are “more democratically legitimate” and advance goals other than profit
I don‘t construe the majority opinion‘s interpretation of the statutory scheme to foreclose that approach. Our interpretation and application of terms such as “extraction” should avoid, to the extent possible, making pivotal distinctions based on subtle technical details of the digital architecture used by government agencies. We should instead seek to advance the interplay of legislative purpose underlying the statutory scheme. (See Weatherford v. City of San Rafael (2017) 2 Cal.5th 1241, 1246–1247.) Our decision today is in that vein: It prudently recognizes that, in this particular context, Hayward may not shift its costs to records requesters for the time its employees spent redacting exempt material from digital body camera footage. Yet it continues to give leeway for government agencies to depend less on having employees cobble together edited reels of material, and more on making thoughtful choices about how best to navigate the full range of considerations relevant to making public records retrieval in the digital age as responsive and effective as possible.
CUÉLLAR, J.