Oracle America, Inc. v. Google Inc.

Filing 1573

ORDER RE INTERNET AND SOCIAL MEDIA SEARCHES OF JURORS. Signed by Judge Alsup on 3/25/16. (whalc1, COURT STAFF) (Filed on 3/25/2016)

1
2
3
4
5
6

IN THE UNITED STATES DISTRICT COURT

7
FOR THE NORTHERN DISTRICT OF CALIFORNIA
8
9
ORACLE AMERICA, INC.,

11
For the Northern District of California

United States District Court

10

12
13
14
15

No. C 10-03561 WHA

Plaintiff,
v.
GOOGLE INC.,

ORDER RE INTERNET
AND SOCIAL MEDIA
SEARCHES OF JURORS

Defendant.
/

16

Trial judges have such respect for juries — reverential respect would not be too strong

17

to say — that it must pain them to contemplate that, in addition to the sacrifice jurors make for

18

our country, they must suffer trial lawyers and jury consultants scouring over their Facebook

19

and other profiles to dissect their politics, religion, relationships, preferences, friends,

20

photographs, and other personal information.

21

In this high-profile copyright action, both sides requested that the Court require the

22

venire to complete a two-page jury questionnaire. One side then wanted a full extra day to

23

digest the answers, and the other side wanted two full extra days, all before beginning voir dire.

24

Wondering about the delay allocated to reviewing two pages, the judge eventually realized that

25

counsel wanted the names and residences from the questionnaire so that, during the delay, their

26

teams could scrub Facebook, Twitter, LinkedIn, and other Internet sites to extract personal data

27

on the venire. Upon inquiry, counsel admitted this. The questionnaire idea cratered, and the

28

discussion moved to whether Internet investigation by counsel about the venire should be

1

allowed at all. That led to a series of memoranda by counsel on the subject and now to this

2

order.

3

An ordinary Google search on the venire would fetch hits, including links to many

4

Facebook profiles, which, in turn, would at least display profile information classified as

5

“public.” Counsel could uncover even more personal information by logging onto their own

6

Facebook accounts and researching the specific venire persons. In this way, counsel could mine

7

not only the “public” data but the details classified as “for friends only” or “friends of friends,”

8

depending on the fortuity of friend listings. The same is true, more or less, for other social

9

media sites.
The Court, of course, realizes that social media and Internet searches on the venire

11
For the Northern District of California

United States District Court

10

would turn up information useful to the lawyers in exercising their three peremptory challenges,

12

and, might even, in a very rare case, turn up information concealed during voir dire that could

13

lead to a for-cause removal. While the trial is underway, ongoing searches might conceivably

14

reveal commentary about the case to or from a juror.

15
16
17

Nevertheless, in this case there are good reasons to restrict, if not forbid, such searches
by counsel, their jury consultants, investigators, and clients.
The first reason is anchored in the danger that upon learning of counsel’s own searches

18

directed at them, our jurors would stray from the Court’s admonition to refrain from conducting

19

Internet searches on the lawyers and the case. This is a high-profile lawsuit, as stated, and dates

20

back to 2010. Nearly one million hits (including tens of thousands of news results) appear in a

21

Google search for “Oracle v. Google.” These include strong opinions on both sides and at least

22

the usual amount of inaccurate information. In this very case, we earlier learned that both sides

23

hired online commentators who have promoted their respective litigation viewpoints on blogs

24

and other web sites. Numerous of the top results for searches on counsel for either party (or for

25

the undersigned judge) include discussions of this controversy and its national policy

26

implications. As a result, we have an unusually strong need to prevent any member of our jury

27
28
2

1

from yielding to the impulse to conduct Internet searches on our lawyers or our case or its

2

history.

3

Our jury will, of course, be admonished to refrain from any Internet research about the

4

lawsuit, the parties, the lawyers, or the judge — an admonition that will be regularly repeated.

5

See Judicial Conference Committee on Court Administration and Case Management,

6

Proposed Model Jury Instructions: The Use of Electronic Technology to Conduct Research on

7

or Communicate about a Case (June 2012). Indeed, the entire venire will be told to refrain

8

from any such research soon after it reaches the courtroom and well before the final jury is

9

even selected.
Google is willing to accept an outright ban on Internet research about the venire and our

11
For the Northern District of California

United States District Court

10

jury, provided the ban applies equally to both sides. Oracle, however, will not. Oracle initially

12

took a broad position on the scope of Internet research it intended to conduct, but it has

13

since purported to scale back its plan. On numerous occasions, Oracle has supplied confusing

14

answers to the Court’s inquiries about its plan, and its responses make little sense in light of

15

how the Court understands the most prominent social media sites to operate. For purposes of

16

this order, it is unnecessary to pin Oracle down in its intentions, but it will be necessary for

17

Oracle to pin down its specific search intentions by the time jury selection begins, as outlined

18

below.

19

To return to the first concern, the apparent unfairness in allowing the lawyers to do to

20

the venire what the venire cannot do to the lawyers will likely have a corrosive effect on fidelity

21

to the no-research admonition. Once our venire learns of the lawyers’ Internet searches (as the

22

venire would via the instruction Oracle requests), a very serious risk will be presented that they

23

will feel justified in doing to the lawyers (and to the case itself) what the lawyers are doing to

24

them, namely, conducting Internet searches — despite the no-research admonition. The

25

one-sidedness of Oracle’s approach will be hard to accept and therein lies the danger. Given

26

the massive volume of Internet commentary on the lawyers and the case, this presents a

27

significant risk.

28
3

1

A second danger posed by allowing counsel to conduct research about the venire and the

2

jury is that it will facilitate improper personal appeals to particular jurors via jury arguments

3

and witness examinations patterned after preferences of jurors found through such Internet

4

searches. For example, if a search found that a juror’s favorite book is To Kill A Mockingbird,

5

it wouldn’t be hard for counsel to construct a copyright jury argument (or a line of expert

6

questions) based on an analogy to that work and to play upon the recent death of Harper Lee, all

7

in an effort to ingratiate himself or herself into the heartstrings of that juror. The same could be

8

done with a favorite quote or with any number of other juror attitudes on free trade, innovation,

9

politics, or history. Jury arguments may, of course, employ analogies and quotations, but it
would be out of bounds to play up to a juror through such a calculated personal appeal, all the

11
For the Northern District of California

United States District Court

10

moreso since the judge, having no access to the dossiers, couldn’t see what was really in play.

12

See United States v. Nobari, 574 F.3d 1065, 1077 (9th Cir. 2009).

13

A third reason is to protect the privacy of the venire. They are not celebrities or public

14

figures. The jury is not a fantasy team composed by consultants, but good citizens commuting

15

from all over our district, willing to serve our country, and willing to bear the burden of

16

deciding a commercial dispute the parties themselves cannot resolve. Their privacy matters.

17

Their privacy should yield only as necessary to reveal bias or a reluctance to follow the Court’s

18

instructions. It is a weak answer that venire persons, through their social media privacy

19

settings, have chosen to expose their profiles to scrutiny, for navigating privacy settings and

20

fully understanding default settings is more a matter of blind faith than conscious choice.

21

(Otherwise, there would be no need for websites explaining the intricacies of privacy settings.)

22

For all these reasons, the Court has considered exercising its discretion to impose an

23

outright ban preventing counsel and the parties from conducting social media and Internet

24

searches on venire persons as well as on the final empaneled jury. Such a ban would be within

25

the sound exercise of discretion to protect the integrity of our process and to curb unnecessary

26

intrusions into juror privacy. A main problem in doing so, however, is that the lawyers would

27

then be precluded from learning information readily available to the press and every member of

28
4

1

the public in the gallery. That is, with an outright ban, everyone in the gallery could have more

2

information about the venire persons and the empaneled jurors than the lawyers themselves.

3

Of course, the Court cannot control those in the gallery, but it can control the trial teams. And,

4

lawyer surveillance is what leads to the problems above, so such a ban on the trial teams would

5

be logical. Still, the Court respects the excellent trial lawyers in this case and their burden in

6

this trial and is reluctant to order them. Rather, the Court calls upon them to voluntarily consent

7

to a ban against Internet research on the venire or our jury until the trial is over. If they will so

8

agree, we will so advise the venire at the start of jury selection and this will surely have a

9

positive effect on fidelity to the no-research admonition. If all counsel so agree, counsel will be

11
For the Northern District of California

United States District Court

10

given an enlargement of time to conduct extra voir dire themselves.
In the absence of complete agreement on a ban, the following procedure will be used.

12

At the outset of jury selection, each side shall inform the venire of the specific extent to which it

13

(including jury consultants, clients, and other agents) will use Internet searches to investigate

14

and to monitor jurors, including specifically searches on Facebook, LinkedIn, Twitter, and so

15

on, including the extent to which they will log onto their own social media accounts to conduct

16

searches and the extent to which they will perform ongoing searches while the trial is underway.

17

Counsel shall not explain away their searches on the ground that the other side will do it, so

18

they have to do it too. Nor may counsel intimate to the venire that the Court has allowed such

19

searches and thereby leave the false impression that the judge approves of the intrusion.

20

Counsel may simply explain that they feel obliged to their clients to consider all information

21

available to the public about candidates to serve as jurors. Otherwise, counsel must stick to

22

disclosing the full extent to which they will conduct searches on jurors. By this disclosure, the

23

venire will be informed that the trial teams will soon learn their names and places of residence

24

and will soon discover and review their social media profiles and postings, depending on the

25

social media privacy settings in place. The venire persons will then be given a few minutes to

26

use their mobile devices to adjust their privacy settings, if they wish. The venire persons will

27

also be given the normal admonition that they cannot do any research about the case, the

28
5

1

parties, or the lawyers and that they cannot speak to anyone about the case, including by making

2

any social media postings about it. Only the names and places of residence of those called

3

forward to the box shall be provided to counsel (so the identities of venire persons still in the

4

gallery will remain private).

5

In this case, there is a further special point that both sides may wish to address with the

their own, prospective jurors are likely to wonder whether Google will be mining the histories

8

of Internet searches by the venire persons to determine their interests in politics, careers,

9

hobbies, dating, shopping, travel, or other intimate facts. Although Google has assured the

10

Court that it has no intention to review such search histories, our venire will not know this

11
For the Northern District of California

venire. The very name of the defendant — Google — brings to mind Internet searches. On

7

United States District Court

6

(unless told) and may speculate. It would, therefore, be prudent to explain to our venire that

12

neither party will resort to examining search histories on any search engine.

13

Thereafter, until the trial is over, each side will be permitted to view online whatever it

14

told the venire it would review — but nothing more. If, as we proceed forward, either side

15

detects any apparent misconduct by a juror, counsel must immediately report it to the Court and

16

the other side regardless of whether the juror appears favorable to their side or not. Each side

17

must preserve an exact record of every search and all information viewed so that if a motion is

18

later made alleging misconduct by a juror, we will all be able to see when the objecting side

19

first learned of the problem.

20

During voir dire, each prospective juror that has been called forward will be asked if he

21

or she can still follow the usual prohibitions on research and public statements about the case

22

despite the one-sidedness (that is, despite the fact that one or both trial teams will be trying to

23

monitor each juror’s social media postings). If the prospective juror cannot do so, he or she will

24

be excused on that basis.

25

With regard to the problem of personal appeals, this order rules now that no personal

26

appeals to any juror may be made. This prohibition bars witness examinations or jury

27

arguments (or opening statements) exploiting information learned about a juror via searches,

28
6

1

such as, without limitation, favorite books, texts, verses, songs, or analogies to likes or dislikes

2

expressed on the Internet. This prohibition, of course, applies whether or not the name of the

3

particular juror is called out.

4

In addition to the Court’s own thorough voir dire, each side shall have twenty minutes of

5

venire examination directed to bias or other for-cause challenges, subject to enlargement for

6

good cause (but this may not be used for “conditioning the jury” or “extracting promises” or

7

drawing out argumentative material from jurors to argue the case).

8
9

The Court would much prefer to fully protect the privacy of all venire persons from
Internet searches and only reluctantly allows the foregoing.

11
For the Northern District of California

United States District Court

10

*

*

*

This is an emerging and developing concern. To that end, this order will now step

12

back and offer a snapshot of how some of today’s prominent social media sites protect (or don’t

13

protect) personal information, which information may be useful for other judges, lawyers, and

14

academics in working through this concern:

15

!

With regard to Facebook, typical profiles contain the following

16

information: lists of personal connections (i.e., “friends”), pictures,

17

videos, check-ins at real-world locations, scheduled events, posts

18

dating back to the creation of the user’s profile (including

19

commentary on news stories and discussions with other “friends”),

20

relationship status, work experience, educational background,

21

current city, home town, contact information, and certain personal

22

interests such as favorite quotes, membership in certain groups,

23

hobbies, political preferences, religious views, sexual orientation,

24

and preferred media. Access to segments of this information can be

25

regulated by the account holder according to various privacy

26

settings. The highest level of privacy keeps information accessible

27

only to the user. This is rarely used, inasmuch as a key purpose of

28
7

1

Facebook is to share information with other people using the

2

platform. The next level of privacy allows a user to reveal

3

information only to his or her “friends” (or to a more limited list).

4

The next level allows access to “friends of friends,” which includes

5

not only the user’s friends but all of the friends of the user’s friends,

6

which can vastly multiply access. (For example, if someone has

7

three hundred friends and each of them has three hundred different

8

friends, access would expand to 90,000 viewers.) The most

9

expansive setting is “public,” meaning at least everyone with a
Facebook account. Certain “public” information, including a user’s

11
For the Northern District of California

United States District Court

10

primary picture, list of likes, hometown, current city, education and

12

work history, and favorite quote, is even available to Internet users

13

without Facebook accounts or who have not logged onto their

14

accounts. Generally, a Facebook user’s post history (i.e., the stream

15

of text, articles, discussions, pictures, and videos the user shares) is

16

not available to searchers without a Facebook account. By default

17

today, most information is accessible only to friends, although a

18

default setting of “friends of friends” has been employed by

19

Facebook in the past. This means that someone may log onto

20

Facebook and automatically access all Facebook information

21

(including post history) on a juror classified as “public,” as “friends

22

of friends” (if there is a “friend” in common) or as “friends only” (if

23

the investigator happens to be a Facebook “friend” of the juror).

24

“Public” information (excluding post history) is accessible as well

25

via ordinary Google searches without ever logging onto Facebook.

26

Finally, Facebook does not notify its users when their profiles have

27

been visited.

28
8

1

!

With regard to Twitter, a typical user’s profile includes that user’s

2

“tweet” history, which includes all posts made since the user

3

created the account, the list of other Twitter accounts that user

4

“follows” and the list of other users that “follow” that user’s

5

account. When one Twitter user “follows” another Twitter user, the

6

latter’s posts appear in the former’s default real-time feed of tweets.

7

Access to this information is regulated by a choice between two

8

privacy settings. The default privacy setting keeps all of a user’s

9

tweets public, in which case anyone may view those tweets, even
without ever logging onto a Twitter account. Additionally, anyone

11
For the Northern District of California

United States District Court

10

who is logged onto Twitter can view the list of a user’s “followers”

12

and the accounts that user “follows” if the user’s tweets are made

13

public. Alternatively, a user may elect to keep tweets “protected,”

14

in which case the above information is available only to other

15

Twitter users whom the user has affirmatively approved to view

16

that information. Twitter does not notify its users when their

17

profiles have been visited.

18

!

With regard to LinkedIn, a typical user’s profile includes a picture,

19

the user’s employment and education history, contact information, a

20

list of skills, publications, awards and interests, among other line

21

items that might appear on a résumé. The profile also includes

22

information such as a list of the “connections” the user has

23

established on the site. (Two users “connect” once one user sends a

24

request to connect and the other accepts the request, which opens up

25

additional personal data and avenues of communication.) A user’s

26

profile also displays a list of groups established on the site that the

27

user has joined. Users can post content, such as links to news

28
9

1

articles or original writing. For each item on a user’s profile

2

(except for the list of “connections”), the user may elect to publish

3

the item to his or her “public profile,” making it visible to everyone,

4

including individuals who are not even logged onto LinkedIn, or to

5

make it visible only to the user’s “connections.” A user may further

6

select whether to reveal his list of personal connections to other

7

users. Unlike Facebook and Twitter, LinkedIn’s default setting is to

8

display a notification informing a user each time his or her profile is

9

visited by another LinkedIn user and identifying that visitor by
name. A visitor may change that setting so that a visitee receives

11
For the Northern District of California

United States District Court

10

no notification, or so the notification only displays the visitor’s

12

place of employment without any further identifying information.

13

With regard to the case law, most of it discusses the problems of jurors using of social

14

media and conducting Internet searches, thereby exposing jurors to commentary about the case.

15

E.g., United States v. Feng Li, No. 14-3294-CR, 2015 WL 7005595, at *3 (2d Cir. 2015);

16

United States v. Fumo, 665 F.3d 288 (3d. Cir. 2011); Dimas-Martinez v. State, 385 S.W.3d 238

17

(Ark. 2011). Several other decisions address whether a party waives arguments seeking to set

18

aside a verdict based on a juror’s apparent bias if that apparent bias could have been discovered

19

through an Internet search of that juror before the verdict (where counsel were allowed and able

20

to conduct such searches). E.g., Johnson v. McCullough, 306 S.W.3d 551, 559 (Mo. 2010);

21

Burden v. CSX Transp., Inc., No. 08-04, 2011 WL 3793664, at *9 (S.D. Ill. Aug. 24, 2011)

22

(Chief Judge David R. Herndon).

23

There are precious few decisions addressing our immediate problem, namely, whether

24

counsel should be allowed to conduct Internet and social media research about prospective and

25

empaneled jurors.

26

The American Bar Association issued an opinion that, within limits, it is ethical for

27

counsel to conduct Internet searches on prospective jurors. Specifically, in Formal Opinion

28
10

1

No. 466, the ABA considered the extent to which an attorney may conduct Internet searches of

2

jurors and prospective jurors without running afoul of ABA Model Rule 3.5(b), which prohibits

3

ex parte communication with jurors. That opinion determined that “passive review” of a juror’s

4

website or social media that is available without making an “access request” and of which the

5

juror is unaware is permissible within ABA Model Rule 3.5(b). The ABA likened such review

6

to “driving down the street where the prospective juror lives to observe the environs in order to

7

glean publicly available information that could inform the lawyer’s jury-selection decisions.”

8

Access requests, such as friend requests on Facebook, “following” users on Twitter, or seeking

9

to “connect” on LinkedIn, whether on one’s own or through a jury consultant or other agent,
constitute forbidden ex parte communications within the rule. That such searches are not

11
For the Northern District of California

United States District Court

10

unethical does not translate into an inalienable right to conduct them.

12

Although the ABA determined that a range of activities is permitted without violating a

13

professional duty, it cautioned that judges may limit the scope of the searches that counsel could

14

perform regarding the juror’s social media “[i]f a judge believes it to be necessary, under the

15

circumstances of a particular matter . . . .”

16

California has not promulgated a rule regarding the ethical scope of Internet research

17

on jurors or prospective jurors, nor has the California State Bar issued an opinion on that

18

subject. The California State Bar website provides a link to the ABA opinion discussed above

19

as well as links to opinions from the Association of the Bar of the City of New York and the

20

Oregon State Bar. Formal Opinion No. 2012-2, issued by the ABCNY, largely mirrors the

21

ABA opinion, except that the ABCNY would prohibit any Internet research that notified the

22

juror such research occurred (such as using LinkedIn in a manner that sent an automatic

23

notification informing the user that another user had visited his or her profile). Formal Opinion

24

No. 2013-189, issued by the Oregon State Bar provides, unlike the ABA or the ABCNY, that a

25

lawyer may affirmatively request access to private aspects of a juror’s social media profiles,

26

provided the lawyer accurately represents his or her role in a case when asked by the juror.

27
28
11

1

Neither the New York City opinion nor the Oregon opinion addressed the judge’s discretion in

2

prohibiting such searches or access requests.

3

In United States v. Norwood, No. 12-CR-20287, 2014 WL 1796644 (E.D. Mich. May 6,

4

2014) (Judge Mark A. Goldsmith), the defendants opposed a plan to empanel an anonymous

5

jury (which would protect the jury from intimidation in a case involving a violent criminal

6

enterprise). Defendants argued that their counsel needed access to the jurors’ identifying

7

information to monitor their social media accounts during the trial to ensure compliance with

8

the no-discussion admonition. Judge Goldsmith rejected the defendants’ argument because the

9

proposed monitoring would “unnecessarily chill the willingness of jurors summoned from [the]
community to serve as participants in our democratic system of justice.” Id. at *4 (quoting

11
For the Northern District of California

United States District Court

10

United States v. Kilpatrick, No. 10-20403, 2012 WL 3237147, at *1 (E.D. Mich. Aug. 7, 2012)

12

(Judge Nancy G. Edmunds)). A jury questionnaire had also specifically asked prospective

13

jurors whether they could abide by the judge’s admonitions regarding Internet and social media.

14

Thus, the defendants’ interest in monitoring the jurors’ social media postings could not

15

overcome the interest in protecting the jurors from intimidation and violence, which interest

16

was served by empaneling an anonymous jury.

17

In Carino v. Muenzen, No. A-5491-08T1, 2010 WL 3448071, at *10 (N.J. Super. App.

18

Div. Aug. 30, 2010), a trial court in New Jersey prohibited the plaintiff’s counsel from using the

19

Internet to investigate the jurors because they had failed to notify opposing counsel that they

20

would be conducting such searches, although the trial court cited no basis for requiring such

21

notification. The New Jersey Appellate Division determined that the trial court abused its

22

discretion by imposing such a prohibition “in the name of ‘fairness’ or maintaining a ‘level

23

playing field.’” Ibid.

24

Oracle cites Sluss v. Commonwealth, 381 S.W.3d 215, 226-227 (Ky. 2012), for the

25

contention that counsel’s lack of access to social media “effectively precluded” full voir dire,

26

but that decision did not address a prohibition on social media and Internet searches. Rather, it

27

addressed two jurors’ false statements made during oral voir dire regarding their respective

28
12

1

relationships to the victim’s mother. A review of the jurors’ respective Facebook profiles later

2

revealed that both jurors were Facebook “friends” with the victim’s mother. One of the jurors

3

in question unequivocally denied having a Facebook account (which later proved false). Thus,

4

the case was remanded for a hearing regarding the jurors’ honesty and whether the true facts

5

warranted for-cause removal.

on denial of rehearing (Nov. 26, 2013), the California Court of Appeal recognized that although

8

certain tools are available to ensure that jurors do not conduct research about the attorneys or

9

the case, a judge lacks the authority “to impose, as a prophylactic measure, an order requiring”

10

defense counsel to remove pages from their website to ensure they could not be viewed by the

11
For the Northern District of California

Finally, in Steiner v. Superior Court, 220 Cal. App. 4th 1479, 1493 (2013), as modified

7

United States District Court

6

jurors, which “constituted an unlawful prior restraint on [counsel’s] constitutional right to free

12

speech.”

13
14
15

*

*

*

Both sides shall inform the Court By MARCH 31 AT NOON, whether they will consent to
a ban against Internet research on the venire or the empaneled jury until the trial is over.

16
17

IT IS SO ORDERED.

18
19

Dated: March 25, 2016.

WILLIAM ALSUP
UNITED STATES DISTRICT JUDGE

20
21
22
23
24
25
26
27
28

13