Case Information
*2 To evaluate reliability, courts may assess an “expert’s reasoning or methodology, using as appropriate
criteria such as testability, publication in peer-reviewed literature, known or potential error rate, and general acceptance.” City of Pomona , 750 F.3d at 1044. However, “the inquiry is a flexible one”—the “list of factors are meant to be helpful, not definitive, and the trial court has discretion to decide how to test an expert’s reliability as well as whether the testimony is reliable, based on the particular circumstances of the particular case.” Primiano , 598 F.3d at 564; see also Kumho Tire , 526 U.S. at 141 (“ Daubert ’s list of specific factors neither necessarily nor exclusively applies to all experts or in every case”). The test “is not the correctness of the expert’s conclusions but the soundness of his methodology.” Daubert v. Merrell Dow Pharm., Inc ., 43 F.3d 1311, 1318 (9th Cir. 1995). “When an expert meets the threshold established by Rule 702 as explained in Daubert , the expert may testify and the jury decides how much weight to give that testimony.” Primiano , 598 F.3d at 565 (9th Cir. 2010); see also City of Pomona , 750 F.3d at 1044 (“Challenges that go to the weight of the evidence are within the province of a fact finder, not a trial court judge.”).
The parties do not dispute that Roffman is qualified as an expert in digital forensics. CREXi moves to exclude Roffman’s testimony under Federal Rule of Evidence 403 on the grounds that his conclusions are “minimally (if at all) probative, highly prejudicial, and will lead to a length mini-trial on irrelevant issues.” (Mot. at 14.) CREXi also moves to exclude the testimony under Rule 702 on the grounds that Roffman’s “use of IP addresses to attribute LoopNet access to CREXi is . . . unreliable.” ( Id . at 18.) CoStar argues that Roffman’s opinions should “be considered as a whole and in the context of CoStar’s theory in this litigation,” which is that CREXi infringed on CoStar’s images “on a massive scale by systematically accessing LoopNet to harvest CoStar’s copyrighted images, and then knowingly displayed those images on CREXi’s platform.” (Opp. at 16.) CoStar also argues that Roffman used a reliable methodology for his opinions. ( Id . at 21.) A. Roffman’s Report
Roffman was retained by counsel for CoStar to opine on CREXi’s access to the LoopNet website (owned by CoStar) by analyzing LoopNet logs and identifying what activity in those logs can be attributable to CREXi and its representatives (including CREXi employees and third-party vendors CREXi retained to complete certain work). (Roffman Report, ¶ 7.) His report details the methods he used to conduct this analysis. Roffman explains how IP addresses can be used to identify users and track their activity on a website, LoopNet’s tools for logging activity (referred to as “hits”) on its website during a “session,” and how users “anonymous online campaigns” attempt to mask user activity through virtual private networks (“VPNs”), outsourcing tasks to third-party business process outsourcing companies (“BPOs”), and rotating through multiple IP addresses in a single session. ( Id ., ¶¶ 28-66.) Roffman then reviewed LoopNet’s logs and identified five “buckets” of information that indicated activity was associated with CREXi (an “indicator of CREXi”). ( Id ., ¶¶ 73-100.) From there, Roffman reviewed LoopNet website activity generated by IP addresses containing an indicator of CREXi to reach the following conclusions:
(1) Indicators of CREXi are present in 31,651 LoopNet sessions. (2) CREXi Associated Sessions accounted for over 8 million hits to LoopNet. (3) Activity in the CREXi Associated Sessions supports a conclusion that CREXi conducted an anonymous
online campaign. (4) Activity observed in CREXi Associated Sessions can be correlated to specific allegations in the Complaint. (5) LoopNet activity observed outside of the CREXi Associated Sessions indicate that it is highly likely CREXi representatives have accessed LoopNet beyond the activity highlighted in [Roffman’s report]. ( Id ., ¶¶ 8-12, 103-133.) B. Rule 403
Rule 403 provides that the Court “may exclude relevant evidence if its probative value is substantially outweighed by a danger of one or more of the following: unfair prejudice, confusing the issues, misleading the jury, *3 undue delay, wasting time, or needlessly presenting cumulative evidence.” CREXi contends that now that CoStar has abandoned its misappropriation claim, the volume of CREXi’s access to LoopNet is irrelevant to CoStar’s remaining copyright infringement and DMCA claims, and that even if there were some probative value, it is substantially outweighed by potential prejudice. (Mot. at 15.) CREXi contends the jury will likely be confused and misled because the activity logs have “nothing to do with the asserted images,” since Roffman did not analyze what the LoopNet activity entailed ( e.g. , viewing a listing as opposed to copying the images in a listing). ( Id . at 16.)
CoStar’s theory of copyright infringement is that CREXi copied its images from LoopNet “on a massive scale,” including by having BPOs copy images from CoStar, crop out CoStar’s watermark logo, and upload them onto CREXi’s website, and that this activity was a company policy at CREXi. ( See TAC, ¶ 117, 122-124, 128, 160-170.) One of CREXi’s defenses to CoStar’s claims is that its users, not CREXi, provided the photos at issue and therefore any infringement is not attributable to CREXi. ( See Answer to TAC, ¶¶ 41-42, 56, 122.) Accordingly, Roffman’s analysis of the volume of activity on LoopNet’s website attributable to CREXi is relevant
1 Sidibe v. Sutter Health , to and probative of whether CREXi employees and BPOs copied images off of LoopNet. F 103 F.4th 675, 701 (9th Cir. 2024) (evidence “must be considered as a whole and in the context of Plaintiffs’ theory of this complex litigation,” and relevance “considers how each ‘brick’ of evidence can, together, form a wall, not whether each ‘witness can make a home run.’”).
Given that CoStar’s theory of the case is “mass infringement” by CREXi, the Court finds that Roffman’s opinions are not substantially more prejudicial than probative. See Positive Ions, Inc. v. Ion Media Networks, Inc. , 2007 WL 9701734, at *6 (C.D. Cal. Nov. 7, 2007) (finding volume of visitors to websites was relevant and not unfairly prejudicial because “low number of visitors” was “not the kind of evidence that is likely to provoke a jury to decide on an improper basis”). CREXi argues that allowing Roffman’s testimony will result in a “mini-trial” on technical concepts like VPNs and proxy servers. But those issues are relevant here because, as Roffman explains, the activity logs reflect that CREXi used VPNs, BPOs, and rotating IP addresses to avoid detection on LoopNet’s website. And while it is true that the activity logs do not reflect whether CREXi copied images from LoopNet (as opposed to copying other information or merely viewing listings), this goes to weight, not 2 admissibility. F C. Rule 702
CREXi contends that Roffman’s opinions 2, 3, 4, and 5 are also unreliable because IP addresses identify a computer network, not an individual user. (Mot. at 18.) But if an IP address is identified as a computer network associated with CREXi, including activity from users on that computer network would likely show activity by users associated with CREXi. Roffman’s testimony attempts to identify LoopNet activity by CREXi and its BPOs. To the extent CREXi contends this does not accurately capture activity attributable to CREXi, it is free to challenge 1 CREXi’s cases are inapposite—in those cases, the experts opined on issues that were not relevant to the claims or did not require expert knowledge. In re NFL’s “Sunday Ticket” Antitrust Litig. , 2024 WL 2076083, at *2 (C.D. Cal. May 2, 2024) (finding opinions based on other outcomes and conclusions from other lawsuits prejudicial); Brice v. Haynes Invs. , LLC., 548 F. Supp. 3d 882, 902 (N.D. Cal. 2021) (finding expert testimony regarding business arrangements between Native American tribes and defendants, and how tribes spent money received from these arrangements, were irrelevant to RICO and state liability for alleged scheme to charge illegally high interest rates for loans and likely to mislead the jury and wasting trial time); Opticurrent, LLC v. Power Integrations, Inc. , 2018 WL 6727826, at *11 (N.D. Cal. Dec. 21, 2018) (finding a vacated verdict had little probative value and cause unfair prejudice by confusing the jury about facts of the dispute at hand); In re Ford Motor Co. DPS6 Powershift Transmission Prods. Liab. Lit. , 2021 WL 3168224, at *3 (C.D. Cal. July 22, 2021) (finding jury may be “overly swayed” by expert testimony that offered only “commonsensical conclusions” that jurors could determine for themselves). 2 Moreover, CREXi’s own expert, Andrew Crain, testified that he “wouldn’t expect any digital trail” of forensic evidence showing copying or screenshotting activity to exist. (Crain Depo. Tr., 68:13-70:19, 71:2-20.) *4 Roffman’s approach on cross-examination. See Positive Ions , 2007 WL 9701734 at *6 (finding expert’s defining a “user” as one IP address when there could be multiple users behind an IP address went to weight, not admissibility).
Next, CREXi argues that Roffman’s opinions unreasonably assume that activity from certain IP addresses containing no indicators, which appeared in the same sessions as IP addresses with an indicator, is also attributable to CREXi. (Mot. at 19.) Roffman explains why he “swept in” activity from IP addresses with no indicator of CREXi—if a user accessed LoopNet from a “CREXi Referring Website” once, only the activity from that particular session would appear as associated with CREXi. Other activity from that IP address where a user did not access LoopNet through a CREXi Referring Website would be left off of the activity log results. ( See Roffman Report, ¶¶ 105-110.) Thus, Roffman included activity from these IP addresses to capture potential CREXi activity where a VPN or rotating IP addresses were employed. Unlike the cases CREXi cites where courts ruled an IP address alone cannot support an infringement claim, here Roffman relies on both IP addresses and sessions, which capture multiple IP addresses associated with one user in a session. ( See Roffman Report, ¶ 41.) Roffman acknowledges that his approach to this analysis “may still be over inclusive,” despite taking measures to remove some data that appeared associated with CoStar’s clients, CoStar, and private IP addresses. ( Id ., ¶¶106, 114.) Roffman also concedes that it was “not practical to analyze 8 million requests to confirm that each was in fact attributable to CREXi,” but that he “quantified the total number of possible hits” from CREXi “to make a good-faith effort at trying to quantify CREXi’s total.” ( Id ., ¶ 114 (emphasis added).)
Therefore, Roffman sufficiently explains his methodology, which is sound and based on his expertise in digital forensics. Again, to the extent CREXi disagrees with Roffman’s methodology because the results may be overinclusive, it is free to challenge his opinions on cross-examination—and indeed, CREXi’s expert Andrew Crain attacks Roffman’s approach in rendering these opinions. ( See Reply at 11.) But questions that go to weight, not admissibility, are for the jury to decide—not the Court.
Accordingly, CREXi’s Motion is DENIED .