From Casetext: Smarter Legal Research

Oracle Am., Inc. v. Google Inc.

Sep 27, 2016
No. C 10-03561 WHA (N.D. Cal. Sep. 27, 2016)


No. C 10-03561 WHA


ORACLE AMERICA, INC., Plaintiff, v. GOOGLE INC., Defendant.



In this copyright infringement action, the jury found the accused infringement constituted fair use. The copyright owner now renews its motion for judgment as a matter of law and separately moves for a new trial. For the reasons stated below, both motions are DENIED.


The history of this case appears earlier (Dkt. No. 1988). In brief, Oracle America, Inc., formerly Sun Microsystems, Inc., has sued Google Inc. for copyright infringement with respect to Google's "reimplementation" of certain API packages in copyrighted Java 2 Standard Edition Versions 1.4 and 5. Following remand from the Federal Circuit, this action proceeded to a second jury trial on fair use, infringement otherwise having been established in the first trial as to certain uses. A pretrial order divided the second trial into phases. Phase one addressed defendant Google's fair use defense. Had the jury found for Oracle during phase one, the same jury would have determined willfulness and monetary remedies in phase two. A third phase, before the judge only, would have determined whether Oracle deserved equitable remedies, including whether Google had equitable defenses.

In phase one, the ten-person jury returned a unanimous verdict finding that Google had carried its burden on the defense of fair use. A comprehensive order denied both sides' motions for judgment as a matter of law, so judgment was entered in Google's favor (Dkt. No. 1988).

Oracle now repeats its motion for judgment, adding a further motion for a new trial under Rule 59. This order follows full briefing, oral argument, and supplemental declarations addressing discovery issues raised in support of a new trial request.



Oracle's new Rule 50 motion is denied for the same reasons as its old one (Dkt. No. 1988).

Oracle's argument that it is entitled to a new trial because the verdict was against the weight of the evidence, which incorporates by reference its brief on the motion for judgment as a matter of law, fails for the same reason.


Pursuant to Rule 59(a)(1)(A), a court may grant a new trial "for any reason for which a new trial has heretofore been granted in an action at law." Rule 61 provides that "no error in admitting or excluding evidence" constitutes a ground for granting a new trial "unless justice so requires." A district court has broad discretion in deciding whether to admit or exclude evidence. Ruvalcaba v. City of Los Angeles, 64 F.3d 1323, 1328 (9th Cir. 1995). A district court also has broad discretion in deciding whether to bifurcate a trial. See Danjaq LLC v. Sony Corp., 263 F.3d 942, 961-62 (9th Cir. 2001). To warrant a new trial on these grounds, the movant must show that the Court's rulings constituted an abuse of discretion plus caused it substantial prejudice.

Oracle's motion for a new trial challenges several discretionary decisions made at trial. Oracle's primary argument, however, is that Google perpetrated discovery-concealment misconduct. The charged misconduct, Oracle says, rates as a "game changer." For important context, however, this order first addresses Oracle's related contention that the Court abused its discretion in limiting the trial to Android as used in smartphones and tablets, postponing all other uses to later trials.

A. New Device Categories and Scope of Trial.

The original trial in 2010 covered Android versions called 1.0, 1.1, Cupcake, Donut, Eclair, and Froyo, as used in smartphones and tablets. The original jury found those versions infringed but deadlocked over fair use. On remand, the issue arose whether to retry that same case taking the infringement verdict as a given and postponing later developments to a future trial versus whether to expand the retrial to include post-2010 developments, a question that came into focus as follows.

After the remand, Oracle sought leave to file a supplemental complaint. Oracle's eventual motion for leave to file a supplemental complaint drew no opposition, and the motion was granted. The supplemental complaint identified six further versions of Android released since the original complaint. It further alleged that Google had implemented Android in various new device categories, including automobiles, wristwatches, televisions, and household appliances (Dkt. No. 1292).

Disagreement surfaced when the parties served their new expert reports. Oracle's expert reports evaluated Google's alleged use of new API packages from Java 2 Standard Edition Versions 6 and 7. But those versions had never been asserted in any operative pleading, including even the supplemental complaint. Only versions 1.4 and 5 had been asserted. Only versions 1.4 and 5 had been presented to the original jury and found to have been infringed. Google moved to strike the overreaching passages of Oracle's expert reports. This led to a hearing that featured the peril of the retrial spinning out of control via a piling on of ever-expanding "updating" issues. The Court expressed concern over the ever-mounting prolixity of this case and the need for a cutoff of new device implementations to be tried (without prejudice to trying the rest later). The Court observed (Dkt. No. 1470 at 9-10):

There's a much cleaner way to deal with this. We can roll back the clock to the moment that that [earlier] trial took place, and try it on that set of facts and the circumstances then. And then all these new products by [Oracle] and these new products by Google would not be in play.
And what that means is, over there on the Google side, that you're going to have to face another lawsuit downstream . . . .
In other words, the practical approach remained retrying the very trial revived by the Federal Circuit, complicated as it already was, preserving the infringement verdict, and saving for a later day all of the subsequent developments.

Nevertheless, the retrial expanded in two important ways. First, in light of Google's stipulation that the earlier jury's finding of infringement should apply to all later versions of Android up through Lollipop, a pretrial order eventually held that our retrial would cover those versions. A later stipulation included Marshmallow as well, adding a total of seven new major releases of Android to the original six. The second expansion was to include the post-2010 time period covered by these versions.

These expansions, by themselves, led to a vast inflation of Oracle's claimed recovery. At the first trial, Oracle's claim for monetary remedies clocked in at much less than a billion dollars, but now they rose to nine billion. The vast inflation flowed from the longer time period of sales of smartphones and tablets as well as the longer list of implicated versions of Android. The vast inflation resulted even though the uses on trial for the fair use defense remained, as before, smartphones and tablets.

The trial was not, however, expanded to include certain other more recent uses like Android TV, Android Auto, Android Wear, or Brillo. They presented a messier problem and were excluded from the scope of the upcoming trial (without prejudice to a later trial to cover them). Notably, the parties couldn't agree on whether the original verdict of infringement would have covered those uses (since they arose after the original verdict, and no evidence on them was presented at the original trial). Had those uses been included in the retrial, Oracle would have had the burden, Google urged, to prove that those uses infringed, rather than relying, as Oracle wished to do, solely on the original verdict of infringement and imposing on Google the burden to prove fair use. Oracle offered to move for summary judgment to establish that the original finding of infringement should be extended to these new implementations, but by the time of that offer, there wasn't sufficient time for the Court to pursue that alternative while sorting out the superabundancy of pretrial issues.

To repeat, all agree that under the pretrial orders, Oracle remained (and remains) free to pursue its claims for infringement arising from Google's implementations of Android in devices other than smartphones and tablets in a separate proceeding and trial.

The scope-of-trial issue surfaced in a second way. Oracle sought to introduce evidence of the excluded device categories at trial as part of its evidence of market harm under the fourth fair use factor. An order in limine, however, held that the only uses set for trial were smartphones and tablets (again without prejudice to a separate future trial as to other uses) (Dkt. No. 1781).

In its new trial motion, Oracle now argues that it was error to limit the device uses in play to smartphones and tablets. We should have had one mega-trial on all uses, it urges. This, however, ignores the fact that Oracle's earlier win on infringement in 2010 — the same win it wished to take as a given without relitigation — concerned only smartphones and tablets. And, it ignores the obvious — one use might be a fair use but another use might not, and the four statutory factors are to be applied on a use-by-use basis. Significantly, the language of Section 107(4) of Title 17 of the United States Code directs us to consider "the effect of the use upon the potential market for or value of the copyrighted work." Oracle cites no authority whatsoever for the proposition that all uses must stand or fall together under the fair use test of Section 107.

True, the fourth fair use factor must consider "whether unrestricted and widespread conduct of the sort engaged in by the defendant would result in a substantially adverse impact on the potential market for the original." Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569, 590 (1994). The concern with widespread use, however, is not whether uses distinct from the accused uses — each of which must be subject to distinct transformativeness analyses — might harm the market for the copyrighted works. Rather, the concern is whether a use of the same sort, if multiplied via use by others, would cause market harm, even though the actual use by the infringer caused only minimal harm. That is not our case. Again, our trial concerned two very important uses — smartphones and tablets — uses that implicated many billions of dollars. All other uses remained open for litigation in further trials.

Oracle relies on decisions from our court of appeals holding that supplementation of a complaint "is favored." E.g., Planned Parenthood of Southern Arizona v. Neely, 130 F.3d 400, 402 (9th Cir. 1997). It argues that postponing its claims relating to devices other than smartphones and tablets contravened the purpose of "promot[ing] as complete an adjudication of the dispute between the parties as is possible." LaSalvia v. United Dairymen, 804 F.2d 1113, 1119 (9th Cir. 1986). Oracle provides a five-page description of the various markets such as automobiles, healthcare devices, "Internet of Things," appliances, and machine-to-machine communication — all involving vastly different technology and functionality from smartphones and tablets — in which Oracle has allegedly suffered harm due to Google's Android-related offerings.

Allowing complaints to be supplemented is favored, but a district judge still has a separate responsibility to manage complex cases, including to decide which issues should be tried in which trial. Good reasons rooted in case and trial management favored the eventual scope of our trial.

Oracle itself, it must be said, successfully excluded at least one post-2010 development that would have helped Google. Specifically, a pretrial ruling obtained by Oracle excluded evidence tendered by Google with respect to Android Nougat. Significantly, this evidence would have shown that (back in 2008) all of the accused APIs could simply have been taken from OpenJDK, Sun's own open-source version of Java, apparently in full compliance with the open-source license. Put differently, Sun itself had given away Java (including all of the lines of code in suit) in 2008 via its open-source OpenJDK. In 2015, Google used OpenJDK to reimplement the Java APIs for the latest release of Android, which it called Nougat. Google wished to use this evidence under the fourth fair use factor to show that its infringement did no more market harm than Sun itself had already invited via its own OpenJDK release. Despite its importance, the Court excluded this development because it had not been presented by Google in time for effective rebuttal by Oracle. This exclusion was a major win for Oracle in the weeks leading up to trial.

Oracle also argues that the first trial was not expressly limited to smartphones and tablets, so it was inappropriate to impose that limitation for the retrial. This isn't correct. In 2012, at our first trial, Oracle presented no evidence of any uses beyond smartphones and tablets. The other alleged uses lay in the future and were not considered by our first jury. Google simply had not yet implemented any aspect of Android on any of the new devices at that time.

After considerable deliberation, the Court exercised its discretion to limit the scope of our trial to address the issue of whether the uses of the copyrighted materials considered at the first trial — smartphones and tablets — including all thirteen versions of Android enabling those uses were fair or not, saving for a future trial new and different uses. In this way, Oracle was allowed to take unquestioned advantage of the infringement verdict in the first trial while also taking full advantage of the subsequent revenue derived from those very device implementations — smartphones and tablets. That limitation also protected our second jury from needing to absorb ever greater complexity in technology and the business models of new and different uses. Oracle remains free to pursue those new and later uses in a future lawsuit, but it is not entitled to a new trial as to smartphones and tablets.

After the verdict, the Court invited counsel to propose scheduling for exactly such a trial on the alleged new and different uses, but both sides preferred to enter a final judgment and proceed to appeals with the understanding that the alleged new and different uses were still open for future lawsuits (Dkt. Nos. 2049-50).

B. The Charge of Discovery Misconduct and ARC++.

With the benefit of the foregoing history of the smartphones and tablets limitation, we turn to Oracle's charge of discovery misconduct. This charge is not anchored in any claimed error by the judge but is anchored in claimed misconduct by Google and its counsel.

At both trials, Google argued that Android's use of the copyrighted lines of code qualified as "transformative" (under the first fair use factor) because Java had been designed for desktops and laptops whereas Android transformed the code at issue to work in the then newly-emerging world of smartphones and tablets. Thus, Google drew a significant distinction between desktops and laptops (Java) and smartphones and tablets (Android). Oracle now accuses Google of withholding evidence in discovery that allegedly would have shown that Google was, by the close of our retrial, expecting soon to implement Android on desktops and laptops too. This argument will now be set out in detail.

Throughout the supplemental discovery period following the remand, Oracle sought discovery into all Google products that incorporated the copyrighted lines at issue. In response, Google identified its App Runtime for Chrome ("ARC"), which enabled laptops and desktops running Google's computer operating system, Chrome OS, to run certain Android applications. Chrome OS was and remains a different operating system from Android (Lin Dep. at 14-19, 107-09). ARC operated on top of Chrome OS and offered all of the Android APIs reimplemented from the Java code at issue. A related project, ARC Welder, enabled Android app developers to repackage the code in their apps for use on Chrome OS devices via ARC.

One of Oracle's own technical experts, Robert Zeidman, addressed ARC in detail in his opening report (Zeidman Rep. ¶¶ 126-43). Oracle's damages expert, James Malackowski, opined in his opening report that Google's release of ARC and ARC Welder and the availability of some Android functionality on Chrome OS devices "means Google is now using Android to occupy the original, traditional market of the Java Platform" (Malackowski Rep. ¶ 172). Oracle, however, never sought to introduce any of the evidence on which these comments were based (or to introduce the expert testimony). Oracle does not accuse anyone of misconduct as to ARC, but ARC supplies relevant background.

Now we come to the crux of the matter. In 2015, Google began a new project, which it internally called "ARC++." Among the goals of ARC++ was to "[p]rovide Chrome OS users with Play Android apps on Chrome OS without developer action" (Anderson Decl., Exh. 7 at *785). That is, Google intended for ARC++ to make the "entire Android app ecosystem" available on Chrome OS devices, so that Android apps would "appear alongside Chrome apps" in the Chrome OS program menu (id., Exh. 8 at *404, Exh. 10 at *396). With ARC++, Google planned to run "Android in an isolated container inside Chrome OS," and "[i]nside the container should be effectively another Linux environment, similar to on an actual device" (id., Exh. 9 at *417). That is, ARC++ would run an isolated instance of Android (with all of Android's public APIs, including those reimplemented from Java) in order to allow users to run all Android apps on Chrome OS devices. Google planned to include its "Play Store" — Google's app wherein users could purchase and download other Android apps — as part of ARC++ to facilitate access to those apps.

In 2015, Google produced to Oracle at least nine documents relating to ARC++ setting forth the information in the preceding paragraph (along with more extensive technical details) and tracking the development of the project (Anderson Decl. ¶¶ 16-20, Exhs. 6-14). This is a key fact in resolving the accusation at hand.

Our trial began on May 9, 2016. Our last day of evidence was May 19, which happened also to be the second day of Google's annual developer conference. On that day, Google announced via a blog post that it would make all Android apps available for use on Chrome OS devices via the Play Store (id., Exh. 15). Although the announcement did not refer to this new feature as ARC++ (no name was given), it reflected the same goals and technical details as the ARC++ project. The announcement stated the feature would first roll out on the experimental developer channel, though over time it would become generally available. The same day at the developer conference, Google demonstrated the use of the Play Store with several Android apps on Chrome OS devices. The presenters acknowledged the technical limitations of the earlier ARC, stating that Google was "building a whole new platform to run Android apps on Chromebooks," i.e., on laptops and desktops (Bush Decl., Exh. J at 3:30). One presenter explained that the new feature ran Android "directly on top of the Linux kernel [of Chrome OS]." Users could "run all of Android Marshmallow within Chrome OS. This includes the Google Play Store" (id. at 7:10).

In short, the announcement indicated that the full functionality of Android would soon be working on desktops and laptops, not just on smartphones and tablets.

Oracle now contends that Google's failure to supplement several responses to interrogatories, requests for admission, and requests for production of documents, as well as the deposition testimony of two witnesses to reflect developments in the ARC++ project constituted discovery misconduct warranting a new trial.

"The test to be applied when discovery misconduct is alleged in a Rule 59 motion must be borrowed from cases interpreting Rule 60(b)(3) . . . ." Jones v. Aero/Chem Corp., 921 F.2d 875 (9th Cir. 1990). Rule 60(b)(3) provides for relief from judgment for "fraud (whether previously called intrinsic or extrinsic), misrepresentation, or misconduct by an opposing party . . . ." To establish misconduct under Rule 60(b)(3), a moving party must:

(1) prove by clear and convincing evidence that the verdict was obtained through fraud, misrepresentation, or other misconduct.

(2) establish that the conduct complained of prevented the losing party from fully and fairly presenting his case or defense. Although when the case involves the withholding of information called for by discovery, the party need not establish that the result in the case would be altered.
Ibid. (quoting Bunch v. United States, 680 1271, 1283 (9th Cir. 1982)). A movant need not show that there would have been a different outcome without the alleged misconduct but need only demonstrate "'substantial interference' by showing 'the material's likely worth as trial evidence or by elucidating its value as a tool for obtaining meaningful discovery.'" Ibid. (quoting Anderson v. Cryovac, Inc., 862 F.2d 910 (1st Cir. 1988)).

Our court of appeals has recognized a "presumption of substantial interference if [the moving party] can demonstrate the misconduct was sufficiently knowing, deliberate or intentional." Ibid. Although Jones did not expressly lay out the framework for applying that presumption, it stated that Anderson, a decision from the First Circuit, "summarized the applicable standards and burdens of proof." Ibid. Anderson, 862 F.2d at 925, held that the presumption of substantial interference "may be refuted by clear and convincing evidence demonstrating that the withheld material was in fact inconsequential."

The oral argument on Oracle's motion for a new trial, which lasted two hours, focused almost exclusively on Oracle's "game changer" allegation of discovery misconduct. Following the hearing, counsel for both sides were ordered to file sworn declarations detailing Oracle's discovery requests on this point and Google's responses. After reviewing the parties' submissions, the Court called for sworn replies.

Throughout the briefing and argument on this motion, Oracle left the distinct impression — more accurately distinct misimpression — that Google had stonewalled and had completely concealed the ARC++ project. This was an unfair argument.

In fact, Google timely produced at least nine documents discussing the goals and technical details of ARC++ and did so back in 2015, at least five months before trial. Counsel for Oracle now acknowledges their legal team never reviewed those documents until the supplemental briefing on this motion (Hurst Reply Decl. ¶ 12). The Court is disappointed that Oracle fostered this impression that no discovery had been timely provided on the ARC++ project eventually announced on May 19.

Oracle contends that Google should have produced source code for the ARC++ project in response to a request for source code that "can be used to facilitate use of Android" on devices other than smartphones and tablets or that it should have identified ARC++ in an interrogatory seeking identification of "any software based on or derived from" Android that incorporated the 37 reimplemented Java API packages, among other similar requests. Google objected to vague language in those requests, and it was not clear to Google whether ARC++, which was in its early stages of development, would have been responsive to requests for information about "products," "software," or versions that were "developed or released," all of which are directed to completed projects. Indeed, the parties met and conferred about discovery responses and discussed Google's objections to Oracle's vague references to efforts to "port Android to desktop," but Oracle did not follow up on Google's objections (Anderson Decl. ¶¶ 30-39).

Rule 26(e) requires a party to supplement discovery responses in a timely manner only "if the additional or corrective information has not otherwise been made known to the other parties during the discovery process or in writing" (or if otherwise ordered by the Court). This creates a "'duty to supplement,' not a right." Luke v. Fam. Care and Urgent Med. Clinics, 323 Fed. Appx. 496, 500 (9th Cir. 2009). Nevertheless, Google had no duty to supplement responses with new information that had already been disclosed in the ARC++ documents already produced.

Oracle should have known that items produced in response to its own document requests potentially contained information that supplemented Google's earlier written discovery responses. Oracle's failure to review the ARC++ documents is its own fault.

It's important, most of all, to step back and remember the scope of our trial. Significantly, any evidence relating to implementations of Android on devices other than smartphones and tablets fell outside the scope of our trial, which was limited to uses on smartphones and tablets. Within the scope of our trial, therefore, Google fairly argued that Android was transformative because it took the declaring code in question, which had been designed for desktops and laptops, and reimplemented it for use in a new context, smartphones and tablets. It may well be true that the use of the copyrighted APIs in ARC++ (or any other later use) will not qualify as a fair use, but that will not and does not mean that Google's argument on transformative use as to the original uses on trial (smartphones and tablets) was improper. That Oracle failed to detect the ARC++ documents in its possession had no consequence within the defined scope of our trial.

Google committed a "fraud on the court," Oracle contends, by eliciting testimony that Android had not caused any harm to the market for the copyrighted works because it was not used on laptops and desktops. As stated, however, this remained a fair argument so long as the trial was focused, as it was, on the original uses — smartphones and tablets — and it remained a fair argument for the time period on trial (the blog announcement came later). The testimony and argument in question fell within the defined scope of our trial. Had Oracle brought up ARC or ARC++, the witnesses would plainly have clarified that their testimony related to the accused uses on trial.

Oracle further notes that the order denying its motion for judgment as a matter of law held that the jury could reasonably have found that "Android caused no harm to the market for the copyrighted works, which were for desktop and laptop computers" (Dkt. No. 1988 at 17). Again, "Android" in that context plainly referred to the accused original implementations of Android within the defined scope of our trial.

Google's launch of the full Android system on Chrome OS also remains, even now, in preliminary stages, available only to developers and on a limited set of devices. Oracle already had evidence of ARC++, but didn't realize it. Thus, to the extent Google's recent announcement had any value at our trial (or in discovery), Oracle already had evidence of the same project (and its predecessor), and it passed on any opportunity to introduce that evidence.

Nor would evidence of ARC++ have caused any interference relating to the Court's rulings limiting the scope of the trial. Indeed, in the briefing and argument on the scope of trial, Oracle never once mentioned ARC, ARC++, or any other use on laptop and desktop computers (neither did Google) (Dkt. Nos. 1559, 1612-3, 1643, 1682). This was so even though Oracle Expert Malackowski had already opined that the release of ARC "means Google is now using Android to occupy the original, traditional market of the Java Platform" (Malackowski Rep. ¶ 172). Instead, at oral argument, Attorney Lisa Simpson for Oracle identified "Android Auto" (not ARC or ARC++) as the most important implementation (to Oracle) that Oracle wished to add (Dkt. No. 1682, Tr. at 123). Oracle contends that the technical differences between ARC and ARC++ meant the latter presented a more compelling narrative both in pretrial motion practice and at trial, but both projects made the same 37 reimplemented Java API packages available for use on Chrome OS; any differences between ARC and ARC++ remained peripheral to Oracle's interest in the projects.

Oracle's purported "game changer" would not have changed anything at all, because the scope of the "game" was smartphones and tablets, postponing new and later uses to a later contest. ARC++ was not yet on trial. Thus, any failure to produce such evidence could not have substantially interfered with Oracle's preparation for our trial. On the contrary, it clearly and convincingly would have been inconsequential.

Out of caution, this order makes clear that the test under Rule 59 is "substantial interference," not "game changer." The phrase "game changer" is Oracle's phrase, even if it expresses a less favorable test than here applicable. This order applies the correct test, "substantial interference."

Oracle insists on taking depositions and document discovery into Google's failure to supplement all discovery responses to reflect the imminent release of a developer version of ARC++ and to present its findings at an evidentiary hearing. Oracle cites Jones v. Aero/Chem Corp., 921 F.2d 875 (9th Cir. 1990), for the proposition that failure to hold an evidentiary hearing on this issue would be reversible error. This type of fishing expedition will not be allowed, and Jones in no way requires such a course.

In Jones, two days after a jury found there had been no defect in the defendants' product, a third-party defendant produced a letter it received from one of the primary defendants nearly a decade earlier indicating that the primary defendant had known of the claimed defect and had explored remedial measures. The plaintiff moved for a new trial, claiming, inter alia, that the defendants had engaged in prejudicial discovery misconduct by withholding the correspondence. "At the hearing on the motion [for a new trial], the district court indicated it might later hold a hearing to determine whether [the] failure to produce the documents involved misconduct." Id. at 877.

Our court of appeals held that the district court improperly decided the motion based on whether the withheld evidence would have resulted in a "different outcome," rather than whether it caused "substantial interference," as required by decisions interpreting Rule 60(b)(3). The failure to hold a separate "hearing" — the court of appeals never referenced an "evidentiary hearing," contrary to Oracle — on the issue was a background circumstance. The actual error was in the standard applied, not the procedure for applying that standard. Notably, the court of appeals did not even require the district court to hold a subsequent hearing, but rather directed it to hold "appropriate proceedings to determine" whether discovery misconduct had occurred according to the proper standard.

In our case, the Court did hold "appropriate proceedings" and did hold a hearing at which the proper standard — Rule 60(b)(3) — was considered, and it further required sworn statements from counsel for both sides and then invited and considered sworn replies, all detailing the discovery conduct at issue. After reviewing many pages and exhibits, the Court finds that no misconduct has been shown (or would likely be shown even with the benefit of a fishing expedition). Nor could any omission of evidence relating to ARC++ have interfered with Oracle's case at all, much less substantially. Contrary to Oracle, ARC++ documents were in fact timely produced. They laid out the basic goals and technical details of the very product referenced on May 19. Since Oracle had that information, there was no need to supplement the written discovery to the extent evidence of ARC++ was responsive at all. Moreover, any further disclosure of ARC++ would have been of no consequence in Oracle's preparation for our trial or its presentation at trial, which later became limited in scope to smartphones and tablets. This ground for a new trial is rejected.

C. Stefano Mazzocchi.

Oracle next contends that a new trial is warranted due to the exclusion of minor evidence and testimony from Stefano Mazzocchi, a member of the board of directors of the Apache Software Foundation in 2008. Back then, Mazzocchi volunteered as a mentor overseeing the Apache Harmony Project and as a member of its Project Management Committee, which sought to create and offer an open-source reimplementation of the Java API. Google eventually used portions of the Harmony project in its reimplementation of 37 Java API packages in Android. Later on, Mazzocchi went to work for Google, but at the relevant time, he worked for neither side.

At our trial, Google presented evidence first (having the burden of proof), but it did not call Mazzocchi as a witness. Nevertheless, Google otherwise introduced evidence of Harmony to support its position that reimplementation of APIs without licenses flowered in the industry.

Oracle never properly designated Mazzocchi as a trial witness under Rule 26(a). Oracle wished to lay before the jury an email that Mazzocchi had sent in April 2008 during the development of Apache Harmony. (In fact, the exhibit was an email from the vice president of legal affairs at Apache and incorporated and responded to an email from Mazzocchi.) Despite Oracle's Rule 26 violation, the Court acquiesced in allowing Oracle to present almost everything it wished to present, including Mazzocchi and the email, save and except for two minor items.

Mazzocchi's email went to a mailing list of members of Apache (TX 5046). It expressed concern that Apache could not distribute Harmony without a license from Sun, even with new implementing code, because "the copyright on the API is real and hard to ignore." Mazzocchi added, "[s]o, we are, in fact, infringing on the spec lead copyright if we distribute something that has not passed the TCK and *we know that*." Our jury heard Mazzocchi's testimony regarding this email, and the entire email itself, including the quotations above, went into evidence, subject to one redaction.

That redaction is now the basis for Oracle's first assignment of error. Its second is that Oracle was precluded from eliciting testimony that Mazzocchi worked for Google at the time of the trial, though he had worked elsewhere when he sent the email.

(i) Redaction.

The Court held that Mazzocchi could testify and that his emails would be admitted, over Google's objection, subject to redaction of the following sentence in the email (TX 5046):

This makes us *already* doing illegal things (in fact, Android using Harmony is illegal as well).
An exchange regarding that redaction occurred (outside the presence of the jury) as follows (Tr. at 1588):
THE COURT: However, the one sentence that I think is too inflammatory and without foundation and should come out is the one sentence that says "This makes us *already* doing illegal things (in fact, Android using Harmony code is illegal as well)." That should not be used. But the two paragraphs that I think you're more interested in, they can be used.

So that one sentence about "This makes us *already* doing illegal things (in fact, Android using Harmony code is illegal as well)" that should be deleted or at least redacted.

MS. HURST (for Oracle): We'll redact that, Your Honor.

Although, as just shown, Oracle's counsel readily accepted that redaction and the email, as redacted, went before the jury, Oracle later — only after Mazzocchi had finished his testimony and had been excused — requested that the Court remove the redaction (Dkt. No. 1925). This was denied, a denial that forms a basis for the new trial motion.

Oracle now argues that sufficient foundation existed because Mazzocchi had "corresponded with the Apache Foundation's VP of Legal Affairs regarding legal issues related to use of copyrighted Java APIs in the Harmony Project" (Pl.'s Mtn. at 16) (citing Tr. at 1712-13).

The so-called "correspondence" with the lawyer, it turns out, went into evidence as the thread leading up to the "Mazzocchi email" (TX 5046; Tr. at 1715). So, whatever foundation existed for the redacted sentence made its way to the jury anyway. (Perhaps this hearsay from the lawyer shouldn't have been admissible at all, but no objection on that ground was made.)

Significantly, nowhere in any passage written by any lawyer did anything come close to what Mazzocchi said in the redacted sentence. So, the thread itself supplied inadequate foundation. Even if Mazzocchi had consulted a lawyer beyond the thread itself (and no such consultation was ever intimated), Mazzocchi himself was not a lawyer, so merely repeating what some lawyer might have told him would have been hearsay (within hearsay).

Indeed, Mazzocchi's testimony before the jury demonstrated that his legal conclusion was utterly without qualification (Tr. at 1727-28):

[MR. KWUN (for Google)]. So thinking back to April of 2008, what, if anything, did you know about fair use in copyright law?

A. I don't recall knowing anything about that.

Q. Did you know what the legal standard is for fair use?

A. I don't — didn't and still don't.

Q. After the email exchange with Mr. Ruby, did you resign as a member from the Apache Software Foundation?

A. No.

Q. And what, if anything, do you conclude from the fact that you did not resign your membership after that email?

A. I really cared about my involvement in Apache. I mean, this was all volunteer work, and I really wanted the foundation to do the right thing for protection of the membership and also for protection of the users.

I would have left slamming the door if I thought that what the foundation was doing was causing harm or doing any illegal things.

So since I wrote these email [sic], I must have changed my mind, something must have changed my mind whether that was the case. And I didn't leave.
Notwithstanding Mazzocchi's lack of training in the law, the Court allowed Oracle to make hay with "the copyright on the API is real and hard to ignore" and that releasing Harmony's reimplementation of the Java API code without passing the compatibility test would have constituted "infringing on the spec lead."

The Court similarly restricted Google from eliciting legal conclusions from former Sun CEO, Jonathan Schwartz, about whether Sun had any legal claim against Google. After his testimony veered too close to that conclusion, the Court issued a corrective instruction and allowed Oracle to question Schwartz about a document that Oracle had improperly clawed back as privileged (Tr. at 508-10, 526). (Schwartz could not recall the document, so it was not admitted into evidence.)

It is worth stressing that the email made no mention of "fair use." It had nothing to do with the fair use issue our jury had to decide. Mazzocchi admitted that he knew nothing about fair use. The Court had already told the jury that Android infringed the copyright subject only to the fair use defense, so a good case existed for excluding the entire email. Nevertheless, virtually all of it came in.

Nor did Mazzocchi's testimony, elicited by Google, that he "would have left slamming the door [at Apache] if [he] thought that what the foundation was doing was causing harm or doing any illegal things" open the door to using the redaction. Mazzocchi's testimony already responded to his understanding that Apache was infringing on Oracle's copyright, and by noting that something "changed [his] mind," he acknowledged that his email reflected initial concern about the legality of Apache's work anyway. Admission of the redaction would have been cumulative.

(ii) Mazzocchi's Employment.

Oracle also contends that it should have been permitted to cross-examine Mazzocchi based on his alleged bias as a current employee of Google. When the Court initially allowed Oracle, despite its inadequate Rule 26 disclosure and over Google's strenuous objection, to call Mazzocchi as a witness, the Court did so to allow presentation of his views when he worked for Apache in 2008 and ruled as follows (Tr. at 1589):

And don't bring up that he works at Google now unless bias becomes a problem. If it appears he's been coached to say things that may not be true, possibly then I would allow you to bring up that he works for Google and that Google — he has met with the lawyers and so forth. But for the time being, you should steer clear of that. And you may treat him as an adverse witness.

During direct examination before the jury, and without seeking leave to address the issue, counsel for Oracle asked Mazzocchi (after he denied recollection of the email containing the "illegal things statement") whether he had met with Google's trial lawyers, which he confirmed he had (Tr. at 1724). The Court allowed the questions over Google's objection.

On cross-examination by Google, as stated, Mazzocchi testified that following the email addressing the issue of Oracle's copyright in the Java APIs with regard to Harmony "something must have changed my mind whether that was the case" (Tr. at 1727). When Google passed the witness back for redirect, Oracle requested a sidebar to be allowed to elicit the fact that Mazzocchi became employed at Google the following year, in order to suggest it was his later employment with Google that had "changed his mind" about the legal status of the Apache Harmony project.

At the sidebar, the Court reviewed Mazzocchi's testimony and concluded that he testified that he would have left Apache sooner than 2009 if he had believed it had been doing something illegal, while he didn't begin his employment with Google until 2010. Contrary to Oracle, Mazzocchi's testimony suggested that something changed his mind before he began working at Google.

Even so, Oracle was able to offer evidence of Mazzocchi's purported bias by eliciting testimony that Mazzocchi spoke with Google's counsel before testifying (Tr. at 1724). Thus, the probative value of evidence of Mazzocchi's then-current employment was minimal, particularly in light of the substantial risk that the jury would mistakenly ascribe Mazzocchi's state of mind while at Apache to Google. (Indeed, Oracle sought to ascribe Mazzocchi's shift in his state of mind to Google, although it predated his employment with Google.)

In the larger picture, the jury heard evidence, pro and con, from both Sun (Oracle) and Google personnel concerning the extent to which reimplementation of APIs occurred in the industry. In view of this sea of evidence, the Mazzocchi email was cumulative. Nevertheless, virtually all of the email came into evidence, including his statement that reimplementing the Java API in particular constituted infringement of the copyright.

Thus, Oracle's contention that it is entitled to a new trial on the basis of the excluded evidence relating to Mazzocchi is rejected.

D. European Commission Response.

Oracle next contends that the Court improperly excluded a document containing responses to questions posed by the European Commission in connection with its 2009 review of Oracle's acquisition of Sun. The question called for an explanation of "the conflict between Sun and Google with regard to Google's Android" (TX 5295 at 39). Oracle sought to admit its response, which read, "Sun believes that the Dalvic [sic] virtual machine plus class libraries, which together constitute Android runtime environment, are an unauthorized derivative work of Java SE" (ibid.). Oracle wished to lay this response before the jury to meet testimony by Sun's former CEO, Jonathan Schwartz, that Sun had welcomed Google's then-recent announcement of Android as part of the Java community, and that industry reimplementations of the Java API had promoted rather than hindered Sun's business plan.

To avoid the self-serving hearsay problem, Oracle attempted to lay foundation for the response through the testimony of its CEO, Safra Catz, who oversaw the acquisition and testified that Sun (not Oracle) had supplied the answer. Out of the presence of the jury, the Court stated it would consider allowing Oracle to admit the response if it had originated with Sun rather than Oracle (Tr. at 1314).

The next morning, out of the presence of the jury, Oracle proffered several drafts of the response to the European Commission. These drafts purportedly traced earlier versions of the response. They originated from Sun's in-house intellectual property counsel. Google protested that these drafts had long been withheld from Google as privileged until the previous night, so that it had had no opportunity to vet Oracle's representations about the drafts. Counsel for Oracle responded that Oracle would waive the privilege. This after-the-deadline waiver, Google replied, failed to cure the prejudice. Temporizing, the Court warned Oracle that its disclosure of privileged documents would constitute an extraordinary waiver (Tr. at 1328).

Nevertheless, still out of the presence of the jury and using the privileged documents, counsel for Oracle traced the internal development of the response to the European Commission. One draft stated, colorfully, "[a] recidivist bank robber should not complain, at least to the authorities, that the bank's new owner might increase security measures around the bank" (Tr. at 1330). A subsequent email from Sun's in-house counsel noted that Oracle's corporate counsel had removed the colorful language and stated "Re Android, we liked our recidivist bank robber analogy" (Tr. at 1331). In light of its document tracing, Oracle proposed that Catz be permitted to testify that the response to the European Commission originated with Sun (how she would have known that on her own was never explained).

The Court rejected that proposal, a rejection that now serves as a ground for the Rule 59 motion.

It is true that Google presented evidence at trial that Sun had embraced a custom of reimplementation of APIs and that Sun's CEO had welcomed Android to the Java community. It is further true that Google argued to the jury that this welcoming attitude reversed only after Oracle took over Sun and brought this suit. Oracle was free to present counterevidence (and did) but the extraordinary after-the-deadline waiver of privilege was too timewise prejudicial to Google, should not have been allowed, and was not.

Counsel for Oracle contended they could offer an email from 2008 in which someone internal to Sun stated Google's conduct constituted copyright infringement, but no such document was ever shown to the Court or offered into evidence.

Oracle's gamesmanship deprived Google of a fair opportunity to vet the privileged documents and to verify the supposed chain of authorship. Anyway, the timing of the emails (at a time when Sun's employees had cause to curry favor with their new boss) suggested that any response "from Sun" was really "from Oracle." This ground for a new trial is rejected.

E. Self-Serving In-House Presentations.

Oracle was barred from placing in evidence certain self-serving in-house materials, offered supposedly to show how Android had hurt Oracle's markets for Java. Specifically, as part of its evidence on market harm under the fourth fair use factor, Oracle sought to admit Trial Exhibits 5961, 6431, and 6470, which were in-house slide show presentations at Oracle. They were used "as [Oracle's] way of planning for [the] next year. They're also used to educate [Oracle's executives] about what is going on in the business" (Tr. at 1356). The presentations included slides that discussed the purported impact of Android on Oracle's revenue.

Oracle invoked Rule 803(6) of the Federal Rules of Evidence, which provides an exception to the rule excluding hearsay evidence for records of a regularly conducted activity, as follows:

A record of an act, event, condition, opinion, or diagnosis if:

(A) the record was made at or near the time by — or from information transmitted by — someone with knowledge;

(B) the record was kept in the course of a regularly conducted activity of a business, organization, occupation, or calling, whether or not for profit;

(C) making the record was a regular practice of that activity;

(D) all these conditions are shown by the testimony of the custodian or another qualified witness, or by a certification that complies with Rule 902(11) or (12) or with a statute permitting certification; and

(E) the opponent does not show that the source of information or the method or circumstances of preparation indicate a lack of trustworthiness.

The Oracle-made documents contained slides with "highlights" and "lowlights" of certain fiscal years, identified "priorities and key messages," summarized revenue data, forecasts, and budgets, identified market challenges, and mapped out product strategies (Bush Decl., Exhs. 26, 27, 29). As to Trial Exhibit 5961, Oracle offered the testimony of its CEO, Safra Catz, to lay the foundation that the presentation had been prepared as part of Oracle's annual budget review (Tr. at 1357). When Oracle moved to admit that exhibit into evidence, Google objected, and the Court sustained the objection because it remained simply a slide show of internal self-serving propositions (even worse, created pending this lawsuit). The Court stated, "if it was just a financial statement, I would allow it, but there are too many slide shows in that document to qualify it as a business record" (Tr. at 1357). Counsel for Oracle sought to admit just page 21 of the exhibit, but that page, titled "FY11 Priorities and Key Messages — Java" suffered from the same self-serving problems. Indeed, that page addressed "integration- specific concerns" regarding the integration of Sun into Oracle — hardly a regularly-conducted activity.

In its brief, Oracle describes this page as "a spreadsheet of revenue and expenses for the first two quarters of fiscal year 2011 for Java embedded and forecasts for the third quarter" (Pl.'s Mtn. at 23). Page 21 does not meet that description. It is possible, it now appears, that counsel for Oracle intended to direct Catz and the Court to page 23, but that error by Oracle then would not now be a reason to grant a new trial.

Oracle sought to admit similar presentations, Trial Exhibits 6431 and 6470, through the testimony of its former vice president of worldwide original electronic manufacturer sales, Neal Civjan, but those presentations were excluded on similar grounds.

Rule 803(6) is not an open window through which any self-serving in-house internal hearsay sails into evidence at the author's behest:

The element of unusual reliability of business records is said variously to be supplied by systematic checking, by regularity and continuity which produce habits of precision, by actual experience of business in relying upon them, or by a duty to make an accurate record as part of a continuing job or occupation.
N.L.R.B. v. First Termite Control Co., Inc., 646 F.2d 424, 427 (9th Cir. 1981), opinion amended on reh'g sub nom. Natl. Lab. Rel. Bd. v. First Termite Control Co. Inc. (9th Cir. Aug. 5, 1981); see also Advisory Committee Notes, 1972 Proposed Rules, Note to Paragraph (6).

The Oracle presentations sought to be admitted were not the kinds of records that could be assured of their reliability due to systematic checking or habits of precision. On the contrary, the documents contained narrative, analysis, and commentary — i.e., self-serving argument. The only "regularity" of the self-serving presentations was that they arose as part of an annual budget review, but the statements themselves had not derived from such a systematic habit of precision. They otherwise lacked the indicia of trustworthiness sought by Rule 803(6). They were properly excluded as hearsay.

F. Bifurcation.

A pretrial order bifurcated the issues of fair use from willfulness and monetary remedies (Dkt. No. 1321 at 13). This prejudiced Oracle, it asserts, because "important market harm testimony never made it to the jury because it was relegated to the damages phase" and because "bifurcation provided a structural incentive for the jury to return a defense verdict" (Pl.'s Mtn. at 20).

Oracle's argument that bifurcation precluded it from presenting its market harm evidence is simply untrue. Nothing about the bifurcation precluded Oracle in phase one from presenting evidence of Oracle's lost revenue attributable to Android. Indeed, Oracle presented extensive evidence in phase one directed at the issue of market harm to the copyrighted works, the fourth fair use factor.

Although there was some overlap in the evidence relevant to market harm and Oracle's actual damages (and Oracle remained free to present it in phase one and did), the most complex evidence on Oracle's remedies — the disgorgement of Google's profits — had virtually no relevance to the market harm/fair use inquiry. Section 107(4) on fair use focuses on the "effect of the use upon the potential market for or value of the copyrighted work" (i.e., harm to Oracle). Section 504(b) on remedies allows a copyright owner to recover "the actual damages suffered . . . as a result of the infringement" (again, harm to Oracle) as well as "any profits of the infringer that are attributable to the infringement" (Google's profits from infringement) — to the extent the awards are not duplicative. Put differently, phase one focused on market harm to the copyrighted work whereas phase two focused on Oracle's damages from that market harm and possible disgorgement of Google's profits attributable to the infringement. Oracle's claim for disgorgement of Google's profits totaled more than ten times Oracle's claimed actual damages and thus would have dominated Oracle's case in phase two.

Google contended that the issue of disgorgement should not be presented to a jury. An order held that the jury would rule on disgorgement, but the Court would resolve Google's argument after the verdict, possibly treating the jury's verdict as advisory, if not conclusive (Dkt. No. 1769).

The disgorgement issue presented extraordinary complexity — complexity unrelated to market harm to the copyrighted works. For one, Google never directly sold Android. Instead, Google offered it free to all comers as open source. Google benefited indirectly. It used Android as a platform for its other services, which earned revenue from advertisements and sales of apps and media. But these other services (like its popular search engine) had already been operating and earning revenue well before Android. Oracle conceded this but contended that Android had multiplied that revenue. Thus, to isolate profits attributable to use of Oracle's copyrighted code, the jury would have been required to apportion, first of all, the revenue between the pre-existing technology already in place versus Android.

Next the jury would have had to further apportion between the accused lines of code versus the unaccused lines of code within Android. The infringing part of Android constituted only a small fraction of one percent of Android. Oracle conceded this but contended that this sliver held the key to the success of Android. These apportionment difficulties were just two examples of many posed by the disgorgement claim for our jury.

Thus, phase two was poised to present bone-crushing analytics on how to apportion any Android profits attributable to the infringement versus profits attributable to non-infringement. To meet this challenge, the parties presented dueling economic models yielding massively different answers. Again, unlike Oracle's lost profits segment, the apportionment/disgorgement problems had virtually no relevance to market harm and fair use.

In the Court's judgment and discretion, our trial was best managed by postponing that mind-bender to phase two, so that the jury could give its undivided attention in phase one to the critical issue of fair use. Dividing the trial further served the important purpose of saving the resources of the Court and the jury (and the parties) in the event that the jury decided against Oracle on fair use.

To repeat, Oracle was free to present its lost profits and other market harm evidence in phase one — and it did so at length. (In phase two, all previously admitted evidence would still have been deemed in evidence.)

Turning to Oracle's structural incentive argument, the Court instructed the jury not to allow any desire to conclude the trial sooner to influence its decision. We must presume the jurors followed the instruction, and there is nothing to indicate otherwise. Richardson v. Marsh, 481 U.S. 200, 206 (1987). Oracle's structural incentive argument, such as it is, would undermine every bifurcation of damages from liability. Yet the law plainly allows bifurcation.

The Court instructed the jury as follows (Dkt. No. 1950 ¶ 46):

Once you render a verdict on the fair use question, we may proceed to the shorter and final phase of the trial on damages issues, depending on your answer to the fair use question. This would still be within the June 10 end date stated earlier. Please do not allow any desire to complete trial sooner to influence your thinking. Once you render your verdict on the fair use issue, it will be final and may not be re-visited or modified during the second phase.

It deserves to be said, in favor of our jury, that the ten who served were as punctual, attentive, and diligent in note-taking as any jury this district judge has seen in seventeen years of service. They had all cleared their calendars. We were on target to meet or beat the time estimate given to the jury. Those with hardships had already been excused during jury selection. It is impossible to even suspect that bifurcation somehow steered the jury to rule as it did. The Court remains completely convinced that the verdict rested, after three days of deliberation, solely on the jury's sincere assessment of the evidence and the instructions of law.


For the reasons stated above, Oracle's motion for a new trial and its motion for judgment as a matter of law are DENIED.

IT IS SO ORDERED. Dated: September 27, 2016.




Summaries of

Oracle Am., Inc. v. Google Inc.

Sep 27, 2016
No. C 10-03561 WHA (N.D. Cal. Sep. 27, 2016)
Case details for

Oracle Am., Inc. v. Google Inc.

Case Details

Full title:ORACLE AMERICA, INC., Plaintiff, v. GOOGLE INC., Defendant.


Date published: Sep 27, 2016


No. C 10-03561 WHA (N.D. Cal. Sep. 27, 2016)