Court File and Parties

COURT FILE NO.: CV-18-590521-00CP DATE: 20210114 SUPERIOR COURT OF JUSTICE - ONTARIO

RE: Glenn Emond and Graeme MacQueen, Plaintiffs – AND – Google LLC, Defendant

BEFORE: E.M. Morgan J.

COUNSEL: Douglas Lennox, for the Plaintiffs Careen Hannouche, for the Plaintiffs (Quebec) Sarah Armstrong, for the Defendant

HEARD: January 12, 2021

SETTLEMENT AND FEE APPROVAL

[1] This motion raises an important question of policy for class actions: what is a court to make of a settlement that results in payment to class counsel and to a cy-près recipient, but absolutely nothing in compensation to the supposedly injured members of the class?

I. Procedural background

[2] This class action was brought in three Canadian jurisdictions: Ontario, Quebec, and British Columbia. The British Columbia case has taken the lead and the parties were to argue the certification there while the other two jurisdictions took a wait-and-see approach.

[3] Certification was set to be argued in Vancouver in December 2018, and the parties worked during the months preceding that date to file all of the materials necessary and to prepare the record for the motion. There were, among other things, six expert witnesses whose reports were filed and full legal argument in the form of factums were exchanged. Counsel for the Plaintiffs has filed in this motion a copy of the entire record that was prepared for the B.C. certification motion. It was as voluminous as most certification records are in class action litigation.

[4] A few weeks prior to the December 2018 motion, a mediation session was held in Vancouver. At that meeting, a proposed national settlement of the claims at issue in all three actions was reached. That proposal was then reduced to writing (the “Settlement Agreement”), and the previously contentious B.C. certification motion was changed to a consent motion to certify the proceeding for settlement purposes. Similar consent motions were brought in Ontario and Quebec. The action was certified for settlement purposes in Ontario in November 2019: Emond v. Google LLC, 2019 ONSC 6584.

[5] For reasons unknown to me, the matter sat for over a year before a motion was brought to approve the settlement and the class counsel fees coming out of the settlement. That motion was heard in Vancouver on December 10, 2020, and the settlement and fees have now been approved by the B.C. Supreme Court: Chartand v. Google LLC, 2021 BCSC 7. The approval motion has not yet been heard in Quebec, and was heard in Ontario by me earlier this week.

[6] At the hearing I was advised by class counsel that notice of the settlement approval hearing had been properly advertised in newspapers and online as previously approved, and that only two individuals had contacted counsel to express their objection to the settlement. Although class counsel provided them with the coordinates, neither of the objecting class members attended at the hearing before me.

II. The allegations in the claim

[7] The allegations giving rise to this claim against Google are technical in nature. At its most fundamental level, the claim is one of privacy protection and non-consensual transmission or collection of phone data by Google. The data that was the target of Google’s now discontinued program tracks the identification numbers of cell phone towers (“Cell I.D. Data”) used to connect Android smart phones to a cell network. In Ontario, the cause of action is framed as intrusion upon seclusion, which is a tort that embraces the protection of interests in informational privacy: Jones v Tsige, 2012 ONCA 32, para 66.

[8] A more detailed description of the technological aspects of the claim is set out by Tucker J. in Chartrand, supra, at paras 9-18. It is a careful and thorough summary. I have repeated it here as I would be hard pressed to try to re-write or improve upon it:

In Cell ID Data, the smartphone is identifiable only as a product (i.e., by unit serial number). The user of the particular smartphone transmitting Cell ID Data cannot be deduced from Cell ID Data alone. Further, as tower service areas can be very large, and data does not provide any indication of signal strength, the geographic location of a smartphone cannot be determined with any specificity by reference to Cell ID Data.

Cell ID Data could, however, if considered alongside significant additional data obtained from other sources (e.g., sales records) assist someone in deducing the identity of a user or a more precise geographical location of the smartphone. (There is no assertion that such additional information is readily available to Google or to the general public.)

In or about 2013, Google concluded that Cell ID Data, considered in conjunction with two further forms of smartphone-generated data (‘Associated Data’), could be used to identify cell network congestion. As a forward-looking measure, Google began building a network congestion reporting feature (“Feature”) into Android O/S in 2014. If enabled, the Feature would transmit Cell ID Data and Associated Data from an Android O/S smartphone to Google when the smartphone was used.

Google ultimately abandoned the network congestion identification project, and the Feature was never enabled for that purpose.

In or around 2015, Google concluded that it could use Associated Data to improve the connections between its messaging system and Android O/S smartphones (“messaging project”). To that end, Google created a server-side code to collect transmitted Associated Data and, in January 2017, enabled the Feature.

From January through December 2017, Android O/S smartphones transmitted Cell ID Data and Associated Data to Google whenever a login request was made on the smartphone. Google’s server-side code collected the Associated Data for the messaging project, logging it in Google’s messaging server. Collected Associated Data was stored in the messaging server for 28 days, and then discarded in accordance with a Google policy.

Google never created a parallel server-side code to collect and log transmitted Cell ID Data. Subject to exception, transmitted Cell ID Data was “immediately discarded” by Google. The possible exception arose when the login request sent was defective. If so, the messaging server re-routed the defective request to a different server for entry in an error log (‘Error Log’). Any data linked to a defective login request is listed in the Error Log as part of the entry…

Thus, while a vast quantity of Cell ID Data was transmitted to Google over 2017, only a small fraction of it was ever collected by entry into the Error Log.

The Feature was disabled by Google at the end of 2017.

[9] The Settlement Agreement was concluded on behalf of an Ontario class defined in the same way as the other three classes, as follows:

All Ontario residents who used a smartphone running the Android operating system in Canada between January 1, 2017 and December 31, 2017, except included in the BC Class, the Quebec Class and any person who opts out of any of the Proceedings or are otherwise excluded under the definition of ‘Class Members or Class’.

[10] A class defined as all Android cell phone users is self-evidently large and unwieldly. Furthermore, from the description of the wrongful conduct alleged against Google, it is obvious that the harm done to each class member is rather small; indeed, as described it may not be greater than a notional harm since virtually no data was kept by Google and none was used even for Google’s technical purpose of addressing network congestion.

[11] Perhaps more to the point, it is obvious from Justice Tucker’s description that no individual cell phone users were identified in the Cell Phone ID Data that Google collected. As it turned out, not only were no individuals identified by Google, none could be identified in the Cell Phone ID Data that it collected. In response to a question posed at the hearing, I was advised that the impossibility of identifying the phone users effected by Google’s program only became known to class counsel when they read the Defendants’ expert reports.

[12] The technical impossibility of identifying individual cell phone users from the Cell Phone ID Data made it obvious to class counsel that the class action would be unmanageable. Not only would they be left with a class that numbered in the millions, but no one would be able to discern which individual claimants actually suffered a loss and so it would be impossible to know how to distribute any award to the class members. As class counsel explains it, it was this understanding that prompted them to pursue a settlement at a mediation rather than to forge ahead with the then upcoming contentious certification motion.

III. The settlement

[13] Technically complicated as the claim might be, the Settlement Agreement is not difficult to describe. Google pays $1,000,000 and admits no liability. One-third of the settlement amount goes to class counsel as fees, on top of that disbursements and HST get paid, and the balance is sent to the Law Foundation in the three provinces as a cy-près award. The Law Foundations are authorized to use the money to fund grants in their sole discretion, with a preference given to technology and privacy-related projects.

[14] In explaining the rationale for the settlement taking this form, class counsel emphasize the impossibility of identifying individual class members. In their factum on this motion they put the problem as follows:

The reason for cy-près donation is that the parties have realized from the evidence exchanged in the BC Action that it would not be possible to directly identify individuals who may be class members. Given this situation, any attempt by the parties to identify individual class members, so as to quantify their respective claims and distribute proportionate shares of any settlement amount, would not be possible.

[15] In addition, class counsel express a concern about the extreme size of the class even with the matter being certified by consent settlement purposes. Counsel explained to me in the hearing that there are several million class members as the class has been defined in the Settlement Agreement, and a claims administrator would not be able to administer and distribute a settlement fund for less than several dollars a person. That would make the settlement amount too small to be economically viable if class members were to get any actual recovery. They put the problem concisely in their factum:

Furthermore, the quantum of the recovery as against the class size does not make the option of direct distribution of damages to individual class members viable. Accordingly, there can be no individual distribution to the Class Members.

[16] Section 29 (2) of the Class Proceeding Act, 1992, SO 1992 c. 6 (now section 27.1) (“CPA”), provides that a class action “may be settled only with approval of the Court.” In weighing a settlement approval, the main question for the court is the reasonableness and fairness of the proposed settlement, which is analyzed with a view to ascertaining whether it is in the best interests of the class: Baxter v Canada (Attorney General) (2006), 2006 CanLII 41673 (ON SC), 83 OR (3d) 481, at para 10 (SCJ).

[17] That said, the Settlement Agreement does not have to embody a level of perfection. “All settlements are the product of compromise and a process of give and take”: Dabbs v Sun Life Assurance Company of Canada, 1998 CanLII 14855 (ON SC), [1998] OJ No 2811 (Gen Div), at para 30.

[18] As explained in Lavier v MyTravel Canada Holidays Inc., 2011 ONSC 1222, at para 21, the court may take into account:

(a) the likelihood of recovery or likelihood of success; (b) the amount and nature of discovery, evidence or investigation; (c) settlement terms and conditions; (d) recommendation and experience of counsel; (e) future expenses and likely duration of litigation and risk; (f) recommendation of neutral parties; (g) if any, the number of objectors and nature of objections; (h) the presence of good faith, arms-length bargaining and the absence of collusion; (i) the degree and nature of communications by counsel and the representative parties with Class Members during the litigation; and (i) information conveying to the court the dynamics of and the positions taken by the parties during the negotiation [citations omitted].

[19] Under the CPA, courts are in a position to approve a settlement entered into prior to certification (or, as here, certified for settlement purposes only), providing that the motion judge is convinced that class counsel “did everything reasonably possible to properly evaluate the claims”: Elliot v Joseph Brant Hospital, 2013 ONSC 124, at para 19. Under the circumstances, I have little doubt that class counsel was wise to seek a settlement. Having realized that the class was virtually impossible to define, their success in achieving certification was far from assured.

[20] This was not a case where the class members were identifiable but they might have trouble quantifying their respective individual losses: see Rosen v BMO Nesbitt Burns Inc., 2016 ONSC 4752, at para 20. Rather, the issue was whether the class could be properly defined since the current working definition – all Android cell phone users – encompassed millions of individuals, a large number of which would have been outside of the areas covered by the Cell Phone ID Data that Google collected. In any case, a mega class, comprised entirely of people who do not know whether they were or were not subject to the Defendant’s actions and who therefore cannot establish that they are properly class members, is really no class at all. It is likely that the case would have had difficulty passing the threshold requirement of section 5(1)(b) of the CPA that there be an identifiable class.

[21] Given the weakness of the case for certification, it is easy to understand why class counsel recommended that the matter settle. Google has offered to pay $1 million more than it likely would have had to pay had the matter been litigated to trial. Class counsel submits that this is in the best interests of the class as a whole. While each class member will have his or her own individual view of any settlement, the “best interests” analysis looks to the collective interest when reviewing a settlement: Parsons v. Canadian Red Cross Society (1999), 40 CPC (4th) 151 (SCJ).

[22] In making submissions about the merits of the settlement at the hearing before me, class counsel told me that although the settlement may not appear especially large or generous, the class derives benefit from it. He expressed the view that this was not what one would call a “nuisance settlement”, and that a million dollars is not a small amount of money for most class members.

[23] That view sounds accurate on one hand, but on the other hand it may be beside the point. After all, settlements for so-called “nuisance” amounts are measured with respect to the defendant – i.e. the party for whom the law suit is a nuisance – and not the plaintiffs.

[24] More importantly, one has to stretch one’s imagination and slightly suspend one’s disbelief to perceive the settlement as “serv[ing] the objectives of the particular case and the interests of the class members”: Sorenson v. easyhome Ltd., 2013 ONSC 4017, at para 30. None of the class members, including the representative plaintiffs for whom there is no honorarium, get any part of the settlement funds. And while the Law Foundation is encouraged in the settlement agreement to try to support research projects related in some way to digital privacy, it would take a rather magnanimous victim of wrongdoing to see that as recompense.

[25] This is not to say that the Law Foundation is not a worthy cause; it funds legal research and law reform projects as well as student internships and a host of other positive contributions to the legal community: see Cassano v. TD Bank (2009), 2009 CanLII 35732 (ON SC), 98 OR (3d) 543, at para 29 (SCJ). But the trickle-down impact on a given individual harmed by a specific tortfeasor is tenuous at best. The attenuated nature of this link to the objective of the claim makes it even more questionable whether there is any sense in which the class is realistically compensated by this award.

[26] The Settlement Agreement, to be clear, gives nothing to any class member but sends money to a public organization that is encouraged, but not required to fund research in some way related to the area of interest. That is what one of Canada’s leading class action scholars has called the “illusion of compensation”.

[27] In an article whose research assistants were funded, according to the acknowledgement footnote, by none other than the Law Foundation, Professor Jasminka Kalajdzic has demonstrated that cy-près awards as used in some contemporary class actions are in need of re-consideration. Specifically, she contends that where a cy-près award is not geared toward distributing unclaimed settlement funds but rather is the very (and, as in the present case, the only) mechanism by which aggregation of loss is effected, a radical transformation of the purpose of class actions has taken place: J. Kalajdzic, “‘The Illusion of Compensation’: Cy Près Distributions in Canadian Class Actions” (2013), 92 Can Bar Rev 173, 175.

[28] Class counsel have justified the use of cy-près here by pointing out that it does fulfill the deterrence/behaviour modification function of class actions: Western Canadian Shopping Centres Inc. v. Dutton, 2001 SCC 46, [2001] 2 SCR 534, at para 29. At least where the award forms a residual part of an otherwise compensatory remedy or is aimed only at a sub-class of claimants, courts have been prepared to view private litigants “as a regulator in the public interest for public policy objectives”: Alfresh Beverages Canada Corp. v. Hoescht AG, [2002] OJ No 79, at para 16 (SCJ). But where the fixed cy-près award is the sole award in the class action, Professor Kalajdzic asks, “have courts replaced the compensatory function of class action litigation with a civil fine model that is unavailable in any other litigation procedure?” Ibid, at 193.

[29] The Ontario Law Reform Commission report that was the catalyst for the CPA was of the view that deterrence is not an aim but an important by-product of class actions: OLRC, “Report on Class Actions” (Toronto, 1982), at 145. However, Canadian courts have been inclined to view deterrence as at least one of several goals of class proceedings legislation.

[30] Access to justice, judicial economy, and behaviour modification are often repeated by courts and lawyers as a trinity of class action objectives: see Hollick v. Toronto (City), 2001 SCC 68, [2001] 3 SCR 158, at para 15. More to the point, in Sun-Rype Products v. Archer Daniels Midland Co., 2013 SCC 58, [2013] 3 SCR 545, at para 97, Karakatsanis J. stated in a strongly worded dissent that in a case where “no individual seeks an individual remedy, it will not be necessary to prove individual loss. Such class actions permit the disgorgement of unlawful gains and serve…the broader purpose of behaviour modification.”

[31] All of which would make for an interesting argument if deterrence were realistic under the circumstances. As noted, however, while a $1 million settlement may seem large to any given class member, Google is a massive multi-jurisdictional enterprise that, according to the Supreme Court of Canada, “controls between 70-75 percent of the global searches on the Internet”: Google Inc. v. Equustek Solutions Inc., 2017 SCC 34, [2017] 1 SCR 824, at para 18. This is certainly a “nuisance” settlement for this particular defendant.

[32] In fact, one might surmise as much from the fact that Google settled the case instead of litigating the certification issue. Given the expert reports that cast doubt on the Plaintiff’s ability to establish an identifiable class, counsel for Google must have viewed their defense to the certification motion as having a high probability of success. Despite those odds, the settlement amount made the case not worth fighting.

[33] By analogy to punitive damages where deterrence is also the goal, if the award has no “sting” for the defendant then behaviour modification is unlikely to be achieved: Whiten v Pilot Insurance Co., 2002 SCC 18, [2002] 1 SCR 595, at para 32. As indicated, one can ponder whether the present settlement is really anything more than a pin prick for Google. But in any case, Google had already put an end to the Cell ID Data collection. Deterrence or behaviour modification may press the right rhetorical button, but frankly it rings a bit hollow.

[34] The fact that the class members cannot be identified from the data collected by Google was not confirmed for class counsel until the expert reports came in describing the technical unfeasibility of making such individual identification. As already indicated, class counsel advised me of that at the hearing and his response was entirely credible. I have no doubt that without seeing reports by experts with technological knowledge beyond that of most lawyers, it would have been hard to know that with any certainty.

[35] That said, it must have been a question mark hanging over this claim all along. The very fact that no one knew until the experts reported that the individual cell phone users could not be identified means that the case was started without knowing whether it was really a viable class action with an identifiable class of persons suffering a loss.

[36] Starting civil litigation generally entails weighing risk against potential reward. – the evidence may not be strong or the arguments may not be accepted, and the recovery must be potentially large enough to counterbalance these possibilities. But starting a class action without knowing whether there is really a class that suffered the alleged harm takes the calculus to a unique level.

[37] There have been cases where settlement was approved as being in the class’ best interest where class counsel got its fees and class members got nothing; typically, this has occurred where it was only in the discovery or evidence exchange process that class counsel was able to evaluate the case accurately and realized how weak it was: K. Field Resources v. Bell Canada International, [2005] OJ No 3935 (SCJ.). On the other hand, there have been cases where courts have refused to approve a settlement providing for class counsel fees but no recovery for class members where the case appeared to be a “strike suit” with little merit brought in hopes of extracting an early settlement: Epstein v. First Marathon Inc., (2000), 2000 CanLII 22797 (ON SC), 41 CPC (4th) 159 (SCJ).

[38] Refusal to approve a settlement because its terms left the class members out of all recovery is relatively rare. Such a result would generally turn on a finding of bad faith in commencing the action: Ibid., at para 76. That is not the case here. The evidence shows that the claim was started in good faith, but was always calculated to be a risky one. When the risk turned too great during the runup to the certification motion, class counsel covered the risk by negotiating a cy-près award in the Settlement Agreement. In that way, the award could arguably be characterized as “materially beneficial” to the class: Elliott v. Boliden Ltd (2006), 34 CPC (6th) 339, at para 35 (SCJ). And even if the material benefit was unrealistic, at least the illusion of compensation was created for the class.

IV. Counsel fees

[39] As already indicated, class counsel fees are 1/3 of the settlement amount, which comes to $333,333. This does not include disbursements or taxes, which are to be paid out of the settlement funds over and above the legal fees. I am unsure how much those will come to, but they will obviously reduce the amount going to the Law Foundations by way of cy-près.

[40] Section 32(1) of the CPA requires court approval of class counsel’s fees. This must be done on the basis of the fees being fair and reasonable, the first measure of which is their proportion in relation to the result of the case: Eidoo v. Infineon Technologies AG, 2014 ONSC 6082 at para 83.

[41] Contingency fees are a well-established feature of class action litigation. This approach has been approved by the courts on the theory that it increases access to justice and gives the lawyer the necessary economic incentive to take the case in the first place and to do it well.: Gagne v. Silcorp Ltd., 1998 CanLII 1584 (ON CA), [1998] OJ No 4182, at para 14 (Ont CA). In general, a contingency fee of 33.33% of the recovery in the case has been held to be presumptively valid: Cannon v. Funds for Canada Foundation, 2013 ONSC 7686, at para 3.

[42] Professor Kalajdzic writes that a primary concern about cy-près awards that send the defendant’s money to class counsel and then to somewhere other than to the class members, is that the self-interest of class counsel may dominate the negotiation process. That is, achieving a settlement may become paramount over achieving an adequate remedy. “Recognizing that class members do not benefit significantly from a cy près settlement and the risk that lawyers may not vigorously protect class members’ interests, one [U.S.] circuit has noted that district courts are permitted to ‘decrease attorneys’ fees where a portion of a fund w[ould] be distributed cy près’”: Kalajdzic, at 185, citing In re Baby Prods. Antitrust Litig., 708 F.3d 163 (3d Cir, 2013). The idea is not to prohibit what can on occasion be a helpful approach to settlement, but to disincentivize its overuse.

[43] I do not know how the contingency amount compares to the hours that class counsel put into the case. At the hearing I asked class counsel if he has time dockets he can produce for me, but he indicated that since this has always been a contingency fee case he and his colleagues did not keep track of their time.

[44] I do understand this, especially since the case was litigated primarily in British Columbia which is a no-costs jurisdiction for class actions and there would be no reason to produce dockets to the court. However, it leaves me with little to go on in assessing whether the contingency fee is reasonable and fair. If proportionality is an important criterion, it is certainly disproportionate to the nil recovery that the class members are receiving.

[45] As previously stated, the quantity of work invested in the aborted certification motion was substantial. However, $333,333 in fees alone would put counsel’s recovery at the high end. There is no evidence as to how many billable hours actually went in the case, but some reduction is called for. Even a relatively modest reduction from class counsel’s fee request would have the benefit of reducing the incentive to bring risky claims against large defendants in order to extract a settlement that is nuisance value to the defendant but large enough to support significant legal fees.

[46] A trimming of class counsel’s fees might also underscore the rather diluted way in which cy-près awards really “compensate” class members. Under the Settlement Agreement, the balance of the $1 million payment that does not go to class counsel will go to the Law Foundation. Those funds will therefore be used to support policy research by academics and students and other legal initiatives. The value of that work will, one supposes, ultimately trickle down in some way to all members of the bar. As a result, class counsel may be doing what class members are doing – i.e. contemplating the idea that the reduction in their fees is actually a form of benefit to themselves.

[47] In my view, it is preferable to take out the frailties of the Settlement Agreement – the fact that it gives no money to the class members and the fact that the Law Foundation is only tenuously connected with the subject matter of the law suit – in class counsel’s fees. While these issues raise a concern about the settlement itself, I am mindful that $1 million is the most that one could expect to extract from Google in a case like this. I would not want to disapprove that settlement and send the parties back to the drawing board. It is more effective to express the concern through a modest reduction in fees.

V. Disposition

[48] The Settlement Agreement is approved.

[49] Class counsel’s fees are approved at $300,000, not including disbursements and applicable taxes.

Date: January 14, 2021
Morgan J.