No. C09–03043 JF (HRL)
April 06, 2011
Alfredo Torrijos, Brian Stephen Kabateck, Richard L. Kellner, Kabateck Brown Kellner, LLP, Mickel Montalban Arias, Arias Ozzello & Gignac, Los Angeles, CA, David R. Buchanan, Seeger Weiss LLP, Harvey Kesner, New York, NY, Eric David Freed, Jeffrey Leon, Freed & Weiss LLC, Chicago, IL, Jonathan Shub, Terrianne Benedetto, Seeger Weiss LLP, Philadelphia, PA, Michael Vincent Storti, Carlsbad, CA, Rosemary M. Rivas, Finkelstein Thompson LLP, San Francisco, CA, Stan M. Doerrer, Finkelstein Thompson LLP, Steven N. Berk, Berk Law LLC, Washington, DC, J. Paul Gignac, Arias, Ozzello & Gignac LLP, Santa Barbara, CA, Melissa Meeker Harnett, Wasserman Comden Casselman & Esensten, L.L.P., Tarzana, CA, for Plaintiffs.
Whitty Somvichian, Cooley Godward Kronish LLP, San Francisco, CA, for Defendant.
ORDER GRANTING PLAINTIFFS' MOTION TO COMPEL PRODUCTION
*1 Plaintiffs RootZoo, Inc., Fox Test Prep, and Steven Price (collectively, “Plaintiffs”), on behalf of themselves and all others similarly situated, bring this action against defendant Facebook, Inc. (“Facebook”) for breach of contract and violation of California's Unfair Competition Law, Cal. Bus. & Prof.Code § 17200 et seq. Docket No. 101 (“Second Amended Complaint” or “SAC”). Plaintiffs, who advertise on Facebook, allege that Facebook misrepresented the quality of its click filters, which are systems that Facebook applies to screen out certain clicks that do not meet certain requirements (so-called “invalid” clicks) so those clicks are not billed to advertisers. As a result, Plaintiffs allege that Facebook charged the advertisers for invalid and fraudulent clicks contrary to Facebook's contractual obligations.
In August 2010, the parties entered into a two-tiered stipulated protective order. Docket No. 95. After several amendments and motions to dismiss, the pleadings were resolved in January 2011.
Discovery disputes have since arisen, and Plaintiffs filed a motion to compel in relation to three of them. Docket No. 150 (“MTC”). First, Plaintiffs say that Facebook has refused to agree to an Electronically Stored Information (“ESI”) Protocol. Second, Plaintiffs say that, rather than actually producing certain documents to them, Facebook has impermissibly uploaded those documents to an internet website. Third, Plaintiffs say that Facebook has refused to respond to several of their Requests for Production of Documents. Facebook opposed Plaintiffs' motion (Docket No. 154 (“Opp'n”), and oral argument was heard on March 22, 2011.
Plaintiffs first urge the Court to direct Facebook to meet and confer with them to agree to an ESI Protocol that would “set forth the manner and form of electronic production, including an agreement on search words or phrases, custodians, time frame and/or terms that Facebook will employ in producing ESI ....” MTC at 1. Because the discovery process to date has been difficult (discussed in more detail below), Plaintiffs believe an ESI Protocol is necessary. See id.
Facebook argues that an ESI Protocol is not necessary, and, even if it were, there is no basis for the Court to impose “the rigid[,] up-front requirements that [P]laintiffs are demanding.” Opp'n at 9–10. Facebook's stated concern that “forcing the parties to try to anticipate and address all potential issues on the form of electronic production would likely have the result of frustrating and slowing down the discovery process,” however, is speculative. The argument that an ESI Protocol cannot address every single issue that may arise is not an argument to have no ESI Protocol at all.
Moreover, the clear thrust of the discovery-related rules, case law, and commentary suggests that “communication among counsel is crucial to a successful electronic discovery process.” Romero v. Allstate Ins. Co., 271 F.R.D. 96, 109 (E.D.Pa.2010).
Indeed, as one court has explained:
*2 The Federal Rules of Civil Procedure, case law, and the Sedona Principles
all further emphasize that electronic discovery should be a party-driven process. Indeed, Rule 26(f) requires that the parties meet and confer to develop a discovery plan. That discovery plan must discuss “any issues about disclosure or discovery of [ESI], including the form or forms in which it should be produced.”
Fed.R.Civ.P. 26(f)(3)(C) (emphasis added).
Aguilar v. Immigration and Customs Enforcement Div. of U.S. Dept. of Homeland Sec., 255 F.R.D. 350, 358 (S.D.N.Y.2008).
Given this cooperative charge and the parties' disagreements so far, the Court believes than an ESI Protocol is needed. Thus, within 30 days from the date of this order, the parties shall meet and confer and agree to an ESI Protocol that addresses the formats in which the various forms of ESI will be produced. Further, Facebook shall meet and confer with Plaintiffs about the search terms that Facebook has used, and will use, in its search of electronic documents.
Plaintiffs next request that the Court direct Facebook to re-produce “all ESI it has thus far produced in native format irrespective of the manner in which it has been previously produced.” MTC at 1. Plaintiffs' motion contains a litany of complaints about Facebook's document productions so far. First, they state that it has produced a significant number of documents in nonnative formats and/or in non-searchable or otherwise unusable formats. Id.
at 9–12. For example, Plaintiffs state that Facebook produced an 18,000–page customer complaint database in PDF format even though Facebook does not maintain the database in that format. Id.
at 11. Plaintiffs claim that this PDF file is “unusable” since it is “a static copy of a dynamic document” and because Facebook redacted the names of the complaining customers without including any other identifying information. Id.
Second, Plaintiffs state that, rather than producing documents directly, Facebook instead has uploaded certain documents to a website, Watchdox.com, that allows Facebook to restrict Plaintiffs' ability to review the documents in certain ways. For one, it requires Plaintiffs to review these documents on a computer that is connected to the internet. In addition, “to prevent copies [of these sensitive, confidential documents] from being inadvertently disclosed to third parties” (Opp'n at 1), documents uploaded on Watchdox.com cannot be printed by Plaintiffs. Other restrictions and options, which previously were being used but now are not, include setting document expiration dates, tracking which documents have been reviewed and by whom, and rendering documents non-searchable or non-annotatable.
Facebook's methods of production, at least with respect to the documents mentioned in Plaintiffs' motion, are either unduly burdensome, not contemplated by the stipulated protective order entered in this case, or both. For example, Facebook's use of Watchdox.com is unduly burdensome upon Plaintiffs. Even though Facebook says it has addressed each of Plaintiffs' concerns as it is notified of them (e.g.,
upon being notified, it gave new reviewers access to the secure website, corrected the orientation of sideways or upside-down documents, disabled the tracking feature, etc.), each of these steps make the discovery process less efficient without providing any real benefit. Facebook's claim that Watchdox.com is needed because the documents uploaded to it are extremely sensitive and confidential is belied by the fact that a two-tiered protective order—which Facebook agreed to be bound by—exists in this case.
Indeed, if the documents are as sensitive and confidential as Facebook asserts, they can produce them as “Highly Confidential–Attorney's Eyes Only.” Facebook's fear of inadvertent disclosure—the stated reason in support of its “no printing” restriction—is purely speculative.
*3 As such, Facebook is prohibited from further use of Watchdox.com, and it shall produce to Plaintiffs (in accordance with the stipulated protective order) any documents that have been uploaded to Watchdox.com. In addition, to the extent possible, Facebook shall re-produce to Plaintiffs, in a searchable format, any documents that were produced in an unsearchable format.
Lastly, Plaintiffs request that the Court order Facebook to respond to several of their RFPs. MTC at 1.
Broadly speaking, RFP Nos. 13–20 seek documents and source code relating to Facebook's systems to identify invalid clicks and filter them from valid clicks. Docket No. 150–2 (“Shub Decl.”), Ex. B at 9–10.
According to the parties' representations during oral argument, Facebook has produced documents responsive to these RFPs, but it has not produced responsive source code.
With respect to source code, Plaintiffs bear the burden of establishing that “the trade secret sought is relevant and necessary to the prosecution or defense of the case before a court is justified in ordering disclosure.” Hartley Pen Co. v. U.S. Dist. Court, 287 F.2d 324, 331 (9th Cir.1961); see also
In re Apple and AT & TM Antitrust Litigation, No. C07–05152 JW (PVT), 2010 WL 1240295, at *2 (N.D.Cal. Mar.26, 2010); Urbina v. Goodyear Tire & Rubber Co., No. CV07–3705 CAS (CTX), 2009 WL 481655, at *3 (C.D.Cal. Feb.23, 2009).
In that regard, Plaintiffs contend that the source code “will enable [their] experts to determine the vitality of ... Facebook's click systems to ensure they were working as ... Facebook so intended.” MTC at 17. In support of this claim, Plaintiffs' expert submitted a declaration that specifically set forth why Facebook's source code is relevant and necessary to Plaintiffs' claims, and Facebook has not persuasively refuted his reasoning.
The source code in this case implemented Facebook's desired filtering, and whether that filtering lived up to Facebook's claims and contractual obligations is the issue here.
Accordingly, Plaintiffs' motion to compel Facebook to produce source code in response to these RFPs is granted. Facebook shall produce responsive source code to Plaintiffs in accordance with the applicable protections described in Section 9 of this District's “Stipulated Protective Order for Litigation Involving Patents, Highly Sensitive Confidential Information and/or Trade Secrets.” Facebook shall provide such access within 30 days from the date of this order.
RFP Nos. 1, 22, and 23 seek monthly data that identifies the number of Facebook's advertisers and the number of valid clicks and invalid clicks on those advertisers' ads. Shub Decl., Ex. B. at 7, 10–11.
Plaintiffs say that these documents are fundamental to establishing the size of the class and number of clicks that Facebook filtered during the class period, and the Court agrees.
*4 At oral argument, the Court learned that Plaintiffs served interrogatories that seek the same information. Facebook explained that, because there are no existing documents responsive to RFP Nos. 1, 22, and 23, it will provide a narrative description in response to Plaintiffs' interrogatories, and Plaintiffs will receive the requested information that way. Upon hearing Facebook's representation, Plaintiffs withdrew their motion to compel with respect to these RFPs.
RFP No. 24 seeks documents related to customer complaints about invalid clicks. Shub Decl., Ex. B at 11.
In response, Facebook produced an 18,000–page PDF file containing the text of a database. Plaintiffs say that it is completely unusable since it is not formatted and does not include any fields that describe the text. In sum, they say that the PDF is a poor substitute for the database and it has not been produced as it appears at Facebook. Facebook says that the database is stored in a proprietary format within Facebook and there is no feasible way to produce it as Plaintiffs request.
At oral argument, at least two alternatives were suggested: (1) Plaintiffs' counsel could go to Facebook and review the database there; or (2) Facebook could provide Plaintiffs' counsel with a computer onto which the proprietary software and the database are loaded. Accordingly, the parties are ordered to meet and confer and to select one of these possible alternatives—or any other that they may devise—to resolve this dispute. The parties shall do so within 30 days from the date of this order.
Based on the foregoing, the Court GRANTS Plaintiffs' motion to compel and ORDERS as follows:
1. Within 30 days from the date of this order, the parties shall meet and confer and agree to an ESI Protocol that addresses the formats in which the various forms of ESI will be produced. Further, Facebook shall meet and confer with Plaintiffs about the search terms that Facebook has used, and will use, with respect to its search of electronic documents.
2. Facebook is prohibited from further use of Watchdox.com, and it shall produce to Plaintiffs (in accordance with the stipulated protective order) any documents that have been uploaded to Watchdox.com. In addition, to the extent possible, Facebook shall re-produce to Plaintiffs, in a searchable format, any documents that were produced in an unsearchable format.
3. Facebook shall produce responsive source code to Plaintiffs in accordance with the applicable protections described in Section 9 of this District's “Stipulated Protective Order for Litigation Involving Patents, Highly Sensitive Confidential Information and/or Trade Secrets.” Facebook shall provide such access within 30 days from the date of this order.
4. The parties shall meet and confer to select an alternative means for Facebook's response to RFP No. 24. Within 30 days from the date of this order, the parties shall select one of these possible alternatives: (1) Plaintiffs' counsel could go to Facebook and review the database there; (2) Facebook could provide Plaintiffs' counsel with a computer onto which the proprietary software and the database are loaded; or (3) any other alternative that the parties devise.
The Sedona Conference, a nonprofit legal policy research and education organization comprised of judges, attorneys, and electronic discovery experts dedicated to resolving electronic document production issues, has advised that:
Cooperation ... requires ... that counsel adequately prepare prior to conferring with opposing counsel to identify custodians and likely sources of relevant ESI, and the steps and costs required to access that information. It requires disclosure and dialogue on the parameters of preservation. It also requires forgoing the short term tactical advantages afforded one party to information asymmetry so that, rather than evading their production obligations, parties communicate candidly enough to identify the appropriate boundaries of discovery. Last, it requires that opposing parties evaluate discovery demands relative to the amount in controversy. In short, it forbids making overbroad discovery requests for purely oppressive, tactical reasons, discovery objections for evasive rather than legitimate reasons, and “document dumps” for obstructionist reasons. In place of gamesmanship, cooperation substitutes transparency and communication about the nature and reasons for discovery requests and objections and the means of resolving disputes about them.
“The Case for Cooperation,” 10 SEDONA CONF. J. 339, 344–45 (2009) (footnote omitted) (emphasis omitted). See also id.
at 342–45, 354–62; Allman, “Conducting E-discovery After the Amendments: The Second Wave,” 10 SEDONA CONF. J. 215, 216 (2009) (“cooperation among parties in [e-]discovery has emerged as a decisive mandate ....”). “Among the items about which the court expects counsel to ‘reach practical agreement’ without the court having to micro-manage e-discovery
are ‘search terms, date ranges, key players and the like.’ ” Trusz v. UBS Realty Investors LLC, ––– F.R.D. ––––, No. 3:09 CV 268(JBA), 2010 WL 3583064, at *4–5 (D.Conn. Sep. 7, 2010) (quoting 10 SEDONA CONF. J. at 217 (footnote omitted)) (emphasis added). “Thus, counsel are generally directed to meet and confer to work in a cooperative, rather than an adversarial manner, to resolve discovery issues.” Id.
at *5 (citing SEC v. Collins & Aikman Corp., 256 F.R.D. 403, 415 (S.D.N.Y.2009).
As Plaintiffs point out, if Facebook wanted to place documents on a secure website such as Watchdox.com, it could have raised that issue before it agreed to the stipulated protective order that clearly contemplates producing
documents to the other side, not merely providing access to them.
Request No. 13
states: “Please produce all documents that relate or refer to your systems that are designed to detect invalid clicks.” Request No. 14
states: “Please produce all documents that relate or refer to your systems designed to pre-filter invalid clicks, i.e.,
to remove invalid clicks before they are even seen by the filters, e.g.,
test clicks generated by your personnel for testing purposes and clicks improperly recorded in the log files whose records have some technical problems resulting in the clicks being unreadable or meaningless.” Request No. 15
states: “Please produce all documents that relate or refer to your systems designed to online filter invalid clicks, including, but not limited to those which are anomaly based (e.g., if normal average clicking on a given advertisement or by a given visitor is 10 times per week and then, in a given week there are 100 clicks on that advertisement or 100 clicks by that visitor), and those which are rule-based (e.g., if a double-click on a given advertisement, with the second click being within x seconds of the first click, then, the second click is deemed invalid).” Request No. 16
states: “Please produce all documents that relate or refer to your systems designed to post-filter invalid clicks offline, including both automated monitoring and manual inspection stages.” Request No. 17
states: “Please produce all documents that relate or refer to revisions to filter parameters, introductions of new filters or conditions to filters and the removal of old underperforming filters .” Request No. 18
states: “Please produce all documents that relate or refer to the performance of the filters, i.e.,
the number or percentage of clicks that the filters categorize as invalid and/or non-billable to customers.” Request No. 19
states: “Please produce all documents that relate or refer to your systems used to track clicks for purposes of charging your CPC customers.” Request No. 20
states: “Please produce all documents that relate or refer to your systems used to track the number of clicks from a single source on a given customer's advertisement.”
As Plaintiffs' expert explained:
[T]he source code performs the actual filtering process and since the questions being raised in this case revolve around the efficiency and accuracy of the filtering process, there is no substitute for source code review. Because all source code changes are dat[e]/time-stamped, source code review in tandem with documentation review also allows for the identification of any time lags between the determination of need for a particular filter and the date that a filter went “live.”
Docket No. 159–2 (“Supp. Cockerham Decl.”) 3. He continued:
Facebook also states that the questions I posed in my earlier Declaration [and for which the source code “would be an invaluable tool”] “have nothing to do with the contents of the computer source code, which is simply the method by [which] Facebook has programmed its system.” I disagree with that statement for the following reasons:
a. Despite significant external documentation of source code, the code itself is the critical component in question.
In [a Facebook engineer's] declaration he states: “The only thing the source code may show (assuming someone not familiar with Facebook's systems and code could interpret it) is simply how the click filtering software is coded. The decisions and reasons why a certain filter may apply to a certain click are explained in the documentation discussed in paragraph 4 above, not in the code.” To use an analogy if the manufacturing systems for a Ford Bronco truck are resulting in defects in the final Ford Broncos, looking at the documentation of the manufacturing process alone is likely insufficient to verify that the manufacturing process is in fact working as designed and documented. The manufacturing process including all moving parts would need to be examined. So, in the case of Facebook click filtering software code, there is no replacement to being able to review the code in tandem with all documentation surrounding the code and its functionality.
Coders are human and humans make mistakes. In computing circles, we call those [ ]mistakes [“]bugs.” Documentation may show in some detail what a developer believes they coded. Only the code itself can verify that the execution of the objective is in fact aligned with the documentation.
c. Source code is nearly always surrounded with “comments.” These comments surrounding specific portions of the click filtering code may shed light on the objectives of that specific filter as well as issues and factors that anyone analyzing the efficacy of the code would need to know.
Request No. 1
states: “Please produce documents sufficient to identify the number of CPC advertising customers that contracted with you per month for CPC advertising placement on your website.” Request No. 22
states: “Please produce all documents sufficient to identify on a monthly aggregate basis the number of billable clicks on advertisers' ads.” Request No. 23
states: “Please produce all documents sufficient to identify on a monthly aggregate basis the number of non-billable clicks on advertisers' ads.”
Request No. 24 states: “Please produce all documents that refer or relate to any communications, or notification of any kind, that you received from CPC advertising customers, potential customers and/or visitors regarding charges for invalid clicks of any kind.”
End of Document.