Court File and Parties

Court File No.: CV-18-599580-00CP Date: 2025-10-24 Superior Court of Justice - Ontario

Re: Douglas Donegani, Matthew Howat and Lyne Brassard v. Facebook, Inc.

Before: J.T. Akbarali J.

Counsel: Theodore Charney, Caleb Edwards and Elie Waitzer, for the plaintiffs Mark Gelowitz, Robert Carson and Lauren Harper, for the defendant

Heard: September 25 and 26, 2025

Endorsement

Overview

[1] In this class action, the plaintiffs claim that the defendant, Facebook Inc., has misused their data by making it available to third parties, including apps, device integration partners, and messaging apps, without their consent.

[2] I heard the certification motion on July 30 and 31, 2024 and released reasons on December 19, 2024. In those reasons, I addressed the criteria under 5(1)(a) and (c) of the Class Proceedings Act. I directed the plaintiffs to provide a new class definition and ordered that the matter return for further submissions on the question of the workability of the class definition, preferable procedure, and the litigation plan: Donegani v. Facebook, Inc., 2024 ONSC 7153.

[3] The parties have argued the issues that remained outstanding after my first set of reasons. In addition, the plaintiffs have delivered an amended notice of motion, seeking to certify a common issue regarding nominal damages that they argue was missed in my earlier reasons, and which the defendants argue was not missed, but rejected as a common issue.

[4] The issues before me on this continued certification motion are:

a. Is there an identifiable class of two or more persons that would be represented by the representative plaintiffs?

b. Did I already determine that a common issue regarding nominal damages for the breach of the duty of honesty, good faith and fair dealing could not be certified? If I did not make that determination, is that a common issue raised by the plaintiff's claim?

c. Is a class proceeding the preferable procedure?

d. Is there a workable litigation plan to advance the proceeding?

[5] I only repeat law or facts from my first set of certification reasons in these reasons if it is necessary to understand my analysis. These reasons should thus be read together with my earlier reasons on this certification motion.

Section 5(1)(b) – Is there an identifiable class?

[6] As I said at paras. 126-127 of my earlier reasons:

For [the identifiable class] criterion to be satisfied, there must be a rational relationship between the class, the cause of action, and the common issues, and the class must not be unnecessarily broad or over-inclusive: Pearson v. Inco Ltd., et al., 78 O.R. (3d) 641, at para. 57 (ON CA).

In determining whether there is an identifiable class, the court asks whether the plaintiff has defined the class by reference to objective criteria, such that a person can be identified to be a class member without reference to the merits of the action. The class must be bounded, and not of unlimited membership, or unnecessarily broad, and have some rational relationship with the common issues: Hollick v. Toronto (City), 2001 SCC 68, [2001] 3 S.C.R. 158, at para. 17, Cloud v. Canada (Attorney General) (2004), 73 O.R. (3d) 401 (C.A.), at para. 45. The class definition needs to identify all those who may have a claim, will be bound by the result of the litigation, and are entitled to notice: Bywater Toronto Transit Commission, (1998), 43 O.R. (3d) 367 (Gen. Div.). Defining the class is a technical, rather than substantive, challenge: Waldman v. Thomson Reuters Corp., 2012 ONSC 1138, 99 C.P.R. (4th) 303, at para. 122.

[7] The plaintiffs' factum for the July 2024 certification hearing proposed an inadequate class definition. They have proposed a new definition at this hearing. The defendant agrees that the new proposed definition addresses the technical and semantic challenges it identified in 2024. It disagrees that the new proposed definition is workable.

[8] The class definition now proposed is as follows:

Class Definition:

The following are defined terms for the purposes of this class definition:

A. "Third Party Programs" means:

i. integrations created to integrate Facebook with products developed by Amazon, Microsoft, and Apple;

ii. Messaging partnerships entered into between Facebook and Netflix or Facebook and RBC;

iii. Applications developed for use with the Facebook platform by AirBNB, Lyft, Cambridge Analytica/Aleksander Kogan (the thisisyourdigitallife app created by Aleksander Kogan for Cambridge Analytica), and Yahoo.

B. "Installing User" means a Facebook user who connected their own Facebook account to one of the Third Party Programs, either by downloading, using, or otherwise installing or signing up for it.

C. "Affected Friends" means any Facebook user who was friends on Facebook with an Installing User.

The class is defined as all Affected Friends in Canada during the period from 2009 to present.

Excluded from the class are residents of the Province of Quebec, the defendant or their subsidiaries, affiliates, officers, directors, senior employees, legal representatives, heirs, predecessors, successors, and assigns.

[9] Of note, the plaintiffs no longer seek to have "friends of Affected Friends" included in the class definition.

[10] The plaintiffs argue that the definition is objective, because the question of whether an Installing User signed up for a Third-Party Program can be determined without reference to the merits of the claim.

[11] They argue that the definition is rationally connected to the common complaint at the heart of the litigation: that Facebook shared data of Affected Friends without their consent with third parties because the Affected Friends were Facebook friends with an Installing User. The plaintiffs argue that the class definition will identify all those who may have a claim, will be bound by the result of the litigation and are entitled to notice.

[12] The defendant argues that the class definition proposed is not the end of the story. The plaintiffs' proposed litigation plan states:

The plaintiffs will seek certification of a class definition, setting out the boundaries of the class. "Class Members" will be those who fall within the class definition and appear on the Master Class List (as further defined below).

[13] The litigation plan goes on to propose that the court set out a process for creating a "Master Class List." In brief, the proposal is that: (i) Facebook provides a list of its Canadian users whose accounts consented to installing one of the nine identified third party programs (thus creating nine lists); (ii) Facebook provides lists of its Canadian users who were Facebook friends with anyone on the lists in (i) above; (iii) The two sets of nine lists will be combined and de-duplicated by class counsel, leading to one list of Affected Friends for each of the third party programs; and (iv) the nine merged lists will be combined and de-duplicated, while retaining the information as to which third party program an Affected Friend's data may have been shared with, leading to a single list: the Master Class List.

[14] This proposal by the plaintiffs thus creates a class definition that does not match the class membership. Rather, by creating the Master Class List, the scope of the class would shrink. The actual class would be narrower than its definition. A Facebook user could not determine objectively whether they are a class member based on the definition alone; they must also appear on the Master Class List, assuming one can be created from Facebook's records.

[15] The evidence indicates that there are limits to the information in Facebook's records. For example, when an Installing User has deleted their Facebook account, Facebook no longer has access to information about who their Facebook friends were. The process for creating a Master Class List would exclude Affected Friends of such Installing Users even though they fall within the class definition. In this sense the class membership is under-inclusive.

[16] At the same time, the process proposed would include Facebook friends of Installing Users even if those friends were not friends with the Installing User when the Installing User installed the third-party program, or only became friends with the Installing User after Facebook disabled the ability to share friend information with the app. The class definition is thus also overbroad, because it includes people who cannot have a claim.

[17] The defendants argue that the gap between the proposed class definition and the subset of that definition who may have a claim (those people on the Master Class List) could only be filled with inquiries into the individual circumstances of each potential class member. Discovery into individual issues is generally not permitted at the common issues trial stage: Curtis v. Medcan Health Management Inc., 2025 ONSC 2902 at para. 4.

[18] Thereafter there would be another level of enquiry: individual inquiries would be required of those on the Master Class List to determine whether their data had been shared (or, possibly, made available for sharing, which is the language the plaintiffs have begun using to describe the wrong of which they complain) without their consent.

[19] In my view, the class definition proposed cannot be understood in isolation from the Master Class List process the plaintiffs also propose. The Master Class List process the plaintiffs have planned is, in fact, a way to refine the class definition to home in on the subset of those who fall within the class definition and who may actually have a claim against Facebook. But the only way to identify those class members is through individual inquiries: first, to determine if they should appear on the Master Class List, and thereafter, to determine if their data was actually shared (or perhaps was made available for sharing) without their consent. This last step would have to occur at individual issues trials, although that conclusion may create an order of operations problem that will become clear when I turn to the common issue analysis below.

[20] Those people who may have a claim but whose Installing User Facebook friends have deleted their accounts will not appear on the Master Class List.

[21] Those people who fall within the class definition but do not appear on the Master Class List will be bound by the results of the litigation, even though they were never actually intended to be class members.

[22] The people who appear on the Master Class List may or may not have a claim, depending on individual issues such as when they were friends with the Installing User.

[23] The plaintiffs argue that a proposed class is not overbroad because it may include persons who ultimately will not have a claim against the defendants: Robinson v. Medtronic Inc., at para. 125.

[24] In Robinson, at para. 137, Perell J. explained the court's concern about over-inclusive or under-inclusive class definitions:

The Court's concern in all this, however, is to ensure that the class definition accords with the design and purposes of the Class Proceedings Act, 1992. The court's concern about a class definition is that it binds the persons who ought to be bound (a concern about under-inclusiveness) and that it does not bind persons who ought not to be bound (a concern about over-inclusiveness). Viewed from the court's perspective, an over-inclusive class definition will bind persons who ought not to be bound by judgment or by settlement, be that judgment or settlement favourable or unfavourable. It is to be remembered that until individual trials, if necessary, class members are not direct participants in the action, but they will be bound by the common issues trial and the settlement if approved by the court. The court is concerned about a proper class definition both as a matter of procedural and substantive justice.

[25] The plaintiffs are in fact proposing an evolving class definition, that requires discovery relating to individual class members' circumstances to define its limits, that is under-inclusive in some respects, and that is overbroad in others. It will necessarily bind some people who should not be bound, and omit some people who ought to be bound. It is not an appropriate class definition.

[26] Additionally, as I have alluded to, the record does not give me confidence that the data required to develop the Master Class List exists. While it is true that Facebook estimated the number of potentially affected Canadian users from the Cambridge Analytica scandal, the evidence indicates that it did so adopting as expansive a methodology as possible, which was subject to limitations rendering it both under-inclusive and over-inclusive.

[27] I note that the plaintiffs could have, but did not, cross-examine Facebook's affiants to explore their evidence about Facebook's data retention. Similarly, in the litigation plan, the plaintiffs refer to making requests under PIPEDA of Facebook and the third party apps to determine what, if any, data was shared. They argue that Facebook and the third party apps are required by law to preserve the information they would seek through the PIPEDA requests. However, they did not attempt even a test case to find out what information could be obtained about the plaintiffs.

[28] In submissions, counsel for the plaintiff agreed that their class would consist of "many millions of people." The notion that sufficient data dating from over a decade ago could be found about many millions of people from Facebook and the third party apps requires more than aspiration. It requires some basis in fact, which is not present on this record.

[29] In the result, I would not certify this action based on the failure to articulate a workable class definition. Although that is sufficient to dispose of the motion, in the event I am wrong about the class definition, I address the other issues.

Section 5(1)(c) - Are there issues in common?

[30] In my earlier reasons, I proceeded to consider the commonality of the issues that survived the s. 5(1)(a) analysis on the assumption that a workable class definition (that could get at those people whose data was shared with, or was made available to, the third party apps, without their consent because they were Facebook friends with an Installing User) would be found. Nothing in the definition proposed by the plaintiffs in this continued motion changes my analysis about the common issues.

[31] However, I must also consider the issues arising with respect to the common issue the plaintiffs ask me to address in these reasons:

Is the defendant liable to the class for nominal damages for breach of the duty of honesty, good faith and fair dealing? If so, in what amount?

[32] The threshold question is whether I have already addressed whether a common issue regarding nominal damages is capable of certification. The parties take differing views of my earlier reasons.

[33] My earlier reasons on common issues tracked the proposed list of common issues that the plaintiffs prepared. Among those issues were proposed common issues 14 and 15 relating to damages:

14. Is the defendant liable to the class for damages for:

a. breach of contract

b. breach of the duties of honesty, and good faith and fair dealing?

f. disgorgement of revenues/profits?

g. punitive damages?

15. If the defendant is liable to the class for damages, can the court assess damages in the aggregate, in whole or in part, for the class? If so, what is the amount of the aggregate damages assessment?

[34] With respect to damages for breach of the duties of honesty, good faith, and fair dealing, I held, at para. 163:

The question of damages for the breach of the duties of honesty, good faith and fair dealing, cannot be determined in common. Whether, and to what extent, each class member suffered damages is an individual issue. It engages questions such as whether their data was shared, the nature of the data shared, and whether they had authorized the sharing of the data with the third party directly. Even if a breach of the above duties is proven, it is not a given that each class member suffered damages.

[35] My analysis centered around compensatory damages. Although there was mention in the plaintiffs' factum about nominal damages, I did not engage with that question. While it could have been better highlighted in argument to ensure it was considered separately from the question of compensatory damages, the issue of nominal damages was raised by the plaintiffs before the first hearing. I thus find that it is appropriate for me to consider the plaintiffs' argument that a common issue regarding nominal damages is capable of certification.

[36] At paras. 134-137 in my earlier reasons, I described the law relating to common issues as follows:

Common issues are defined in the CPA as "common but not necessarily identical issues of fact, or common but not necessarily identical issues of law that arise from common but not necessarily identical facts."

To satisfy this requirement of the certification test, the plaintiffs must establish that there is some basis in fact to conclude that: (i) the proposed common issues actually exist; and (ii) the proposed common issues can be answered in common across the entire class and will significantly advance the claims of the entire class: Simpson v. Facebook, 2021 ONSC 968, 469 D.L.R. (4th) 699, at para. 43, Pioneer Corp. v. Godfrey, 2019 SCC 42, [2019] 3 S.C.R. 295, at para. 105, Batten v. Boehringer Ingelheim (Canada) Ltd., 2017 ONSC 53, at para. 162, aff'd 2017 ONSC 6098 (Div. Ct.), leave to appeal refused, M48535 (28 February, 2018) (Ont. C.A.).

When considering whether a claim raises a common issue, the court asks whether it is necessary to resolve the issue in order to resolve each class member's claim, and whether the issue is a substantial ingredient of each of the class members' claims. The issue is a substantial ingredient of each claim if its resolution will advance the case or move the litigation forward, and if it is capable of extrapolation to all class members: Vivendi Canada Inc. v. Dell'Aniello, 2014 SCC 1, [2014] 1 S.C.R. 3, at para. 46.

To be certified as a common issue, an issue cannot be common only when stated in overly broad terms. "Inevitably such an action would ultimately break down into individual proceedings. That the suit had initially been certified as a class action could only make the proceeding less fair and less efficient": Rumley v. British Columbia, 2001 SCC 69, at para. 29; Broutzas v. Rouge Valley Health System, 2018 ONSC 6315, at para. 270.

[37] The plaintiffs have taken the position that, based on my earlier reasons, the only claim that will be going forward (subject to this motion) is the breach of the duty of good faith and fair dealing, which is a form of breach of contract. The plaintiffs argue that damages would normally be compensatory for such a claim. However, the plaintiffs submit, at para. 32 of their factum, that "this is not a situation where significant compensatory damages have accrued." They note that Facebook did not charge the class for its services, nor does it appear that the alleged improper data sharing caused any pecuniary damages. Moreover, claims for severe mental distress are likely to be limited, especially as users were not aware of the data sharing.

[38] The plaintiffs thus indicated that if the nominal damages question is certified, they will disclaim compensatory damages on behalf of the class. They rely on Sharp v. Royal Mutual Funds Inc., 2021 BCCA 307, at para. 164, where the British Columbia Court of Appeal found that it would have been open to the plaintiffs in that case to disclaim compensatory damages on behalf of the class in favour of pursuing nominal damages.

[39] On the other hand, if no common issue regarding nominal damages is certified, the plaintiffs would pursue their compensatory damage claims. I have some difficulty with this position, given that they have acknowledged there is no indication that any compensatory damages have been suffered, and there is nothing in the record to suggest anyone has suffered any compensatory loss.

[40] Nominal damages are available for causes of action that do not require proof of loss, like breach of contract. As the British Columbia Court of Appeal held in Insurance Corporation of British Columbia v. Ari, 2025 BCCA 131, at para. 35, nominal damages are awarded where the plaintiff establishes a breach of a right but fails to establish a loss caused by the wrong. They are not properly damages at all, but are awarded to affirm an infraction of a legal right which, while not giving rise to damages, gives the plaintiff a right to the verdict or judgment because their legal right has been infringed.

[41] The plaintiffs argue that determining nominal damages would not require the court to determine the nature of the data shared and whether class members had separately authorized the sharing directly, because they would operate as symbolic vindication for Facebook's alleged breach of its contractual duty of good faith and fair dealing.

[42] The plaintiffs suggest nominal damages could be determined in the aggregate, without the need for expert evidence. All one would have to do is count the number of class members whose data was improperly shared and multiply that number by a nominal amount per class member.

[43] However, the plaintiffs acknowledge that to determine liability at trial, and who is entitled to nominal damages, it will be necessary to identify the class members whose data was shared, and that understanding the number of class members will permit the common issues trial judge to determine the appropriate quantum of nominal damages: see plaintiffs' factum, para. 41.

[44] Here arises the order of operations problem that I mentioned earlier. Before nominal damages could be determined in common, individual issues would have to be discovered upon in order to develop the Master Class List, and then individual issues trials would have to be held to determine whether class members' data were shared (or made available for sharing) before a common issues trial judge could determine the number of class members who are entitled to nominal damages.

[45] The individual issues are woven into the nominal damages issue. The nominal damages issue could not be determined in common at the time of the common issues trial.

[46] For these reasons, I would not certify the nominal damages question.

Section 5(1)(d) – Preferable Procedure

[47] In my earlier reasons, at paras. 169-171, I described the law relating to preferable procedure as follows:

In order to determine whether a class proceeding is the preferable procedure, the court must consider the importance of the common issues in relation to the claims as a whole: Hollick, at para. 30.

In Bennett v. Lenovo (Canada) Inc., 2017 ONSC 5853, at paras. 84, 86, Perell J. summarized the criteria relevant to a preferable procedure analysis:

a. whether a class proceeding would be better than other methods, such as joinder, test cases, or other means of resolving the dispute;

b. whether a class proceeding represents a fair, efficient, and manageable procedure that is preferable to any alternative method of resolving the claims;

c. whether a class proceeding is the preferable procedure is judged by reference to the purposes of access to justice, behaviour modification, and judicial economy, and by taking into account the importance of the common issues to the claims as a whole, including the individual issues.

This proceeding pre-dates the recent amendments to the CPA such that the criteria in s. 5(1.1) of the CPA need not be considered.

[48] In my view, this proceeding fails the preferable procedure branch of the test for the following reasons.

[49] First, the plaintiffs have indicated that they would disclaim seeking compensatory damages in favour of nominal damages if I found that question could be certified. I have declined to certify the nominal damages issue, but what is of note here is the plaintiffs' admission that it does not appear that the alleged unlawful sharing of user data caused pecuniary damages, and that "this is an ideal case for nominal damages, as there are unlikely to be pecuniary damages."

[50] There is no evidence of compensatory damages in the record at all.

[51] In Atlantic Lottery Corp Inc. v. Babstock, 2020 SCC 19 at para. 68, the majority of the Supreme Court of Canada held:

Even were the breach of contract claim to survive, the application judge's certification decision would have to be revisited. Given my conclusion that each of the plaintiffs' claims should be struck, it is unnecessary to address certification in detail. I respectfully disagree with my colleague, however, that the plaintiff's breach of contract claim, standing alone, would satisfy the preferability requirement in s. 5(1) (d) of the Class Actions Act (Karakatsanis J. Reasons, at paras. 165-70). As I have explained, punitive damages and disgorgement are unavailable to the plaintiffs. Without those remedies, the plaintiffs would be pursuing a breach of contract action wherein each plaintiff effectively elects to pursue nominal damages in lieu of the actual damages they have suffered. Such an action would not further the principal goals of class actions, namely judicial economy, behavior modification, and access to justice ( Hollick v. Toronto (City), 2001 SCC 68, [2001] 3 S.C.R. 158, at paras. 27‑28 )

[52] This statement in Atlantic Lottery has been repeatedly relied upon for the principle that the need to show some evidence of compensable loss is a fundamental prerequisite for the certification of a class proceeding: Maginnis and Magnaye v. FCA Canada et al, 2020 ONSC 5462, at para. 41, aff'd 2021 ONSC 3897 (Div. Ct.), at paras. 27-30; Palmer v. Teva Canada Ltd., 2022 ONSC 4690, at para. 140, aff'd 2024 ONCA 220.

[53] In the s. 5(1)(a) analysis in my earlier reasons, I found that the question of whether nominal damages amount to relief that can vindicate the nature of the class members' interest is not one that ought to be determined at the certification stage: at para. 122. That does not mean, however, that the question about compensable loss does not otherwise arise in the certification analysis.

[54] The plaintiffs point to other cases that have certified questions regarding nominal damages. In Lam v. Flo, 2025 BCSC 993, the court certified a question regarding nominal damages, but questions regarding compensatory damages and disgorgement were also certified in that case.

[55] The plaintiffs argue that in Tucci v. Peoples Trust Company, 2020 BCCA 246, the British Columbia Court of Appeal certified an issue regarding nominal damages. I disagree. There, the court noted that the certification judge certified a question asking whether class members' nominal damages could be assessed in the aggregate. However, the Court of Appeal noted that what the certification judge described as "nominal damages" were actually damages to compensate plaintiffs for damage occasioned by the data breach, that is, lost time and inconvenience. The court noted that such damages are compensatory in nature, requiring proof of loss. The court found that, because it was apparent that the motion judge contemplated an actual assessment of proven damages, the certified damages question had to be revised to remove the reference to nominal damages.

[56] There is no certified case of which I am aware where there is only a claim for nominal damages.

[57] While the plaintiffs here indicated they would pursue compensable damages if the nominal damages question was not certified, I cannot ignore their admission that there are unlikely to be pecuniary damages, nor the fact that the record discloses no evidence of compensable loss.

[58] Without compensable loss, certifying the action would not meet the goal of promoting access to justice. As Belobaba J. held in Maginnis, at para. 39, "absent compensable harm, there are no access to justice concerns."

[59] In Singer v. Schering-Plough Canada Inc., 2010 ONSC 42, at para. 206, after expressing doubt that the plaintiff had suffered any damages, Strathy J., as he then was, found that a plaintiff who wished to make a point of principle could appropriately do so in the Small Claims Court or as a test case. In other words, where there are no compensable losses, a class action is not the preferable way to make a point.

[60] Nor am I satisfied that this proceeding would advance the goals of behaviour modification. While allowing that there are times where nominal damages ordered on behalf of a large enough class could inflict enough pain on a defendant that the goal of behaviour modification could be advanced, I do not find that that possibility is sufficient to make a proposed class proceeding the preferable procedure in this case.

[61] Facebook has been subject to regulatory proceedings relating to the issue of data sharing with third parties. In Canada, regulatory proceedings addressing some of the issues raised in this case are ongoing. Regulatory proceedings were also commenced in the United States and United Kingdom. Some of these have included significant fines. Regulatory proceedings are a preferable way of encouraging behaviour modification in this case, where there is no evidence of compensable loss.

[62] Finally, I am certain that this proceeding would not advance the goal of judicial economy. To the contrary, the proceeding would be a waste of judicial resources. The proceeding would be large, unwieldy, and factually complex. It would require significant judicial resources. Without any evidence of compensable loss, and in the face of an underfunded justice system, judicial resources are better spent elsewhere.

[63] The class proceeding proposed would not be fair, efficient, and manageable, in view of the number of individual issues that even the plaintiffs' own litigation plan requires to be canvassed during the discovery phase of the common issues trial, with individual trials to follow.

[64] Other options, like a test case or regulatory proceedings, are preferable to a class proceeding in this case.

[65] I recall again plaintiffs' counsel's statement that the class would be "many millions of people." It is a significant exercise to dig through each person's Facebook account and the accounts of all of their Facebook friends to figure out who was friends with whom at what time when which apps were engaged with, and to determine whose data was actually shared or made available for sharing without their permission. The challenge is made especially stark in the face of an absence of evidence on the record that the data would even exist. To do all of this without a shred of evidence of compensable loss would be contrary to the goals of class proceedings.

[66] I would not certify this proceeding because I find that it is not the preferable procedure.

Section 5(1)(e) – Litigation Plan

[67] I have already outlined the steps in the litigation plan that are problematic, including the numerous individual inquiries required to develop a Master Class List, the lack of any evidence that the PIPEDA inquiries the plaintiffs suggest they would make would yield results, and the sheer size of the class which creates significant manageability concerns. The litigation plan makes clear that a class proceeding is not the preferable procedure in this case. Nothing more need be said about the litigation plan.

Conclusion

[68] For the reasons given above, I decline to certify this proceeding.

Costs

[69] The parties have agreed on costs of both, the original hearing and the subsequent hearing. I thank them for their efforts to do so.

[70] The defendant is the successful party on this motion. Based on the parties' agreement, the defendant is entitled to costs as follows:

a. With respect to the original hearing, costs fixed at $375,000 all-inclusive.

b. With respect to the second hearing, costs fixed at $125,000 all-inclusive.

[71] Facebook is thus entitled to its costs of $500,000 all-inclusive.

J.T. Akbarali J.

Date: October 24, 2025