Meta & Children's Online SafetyDeposition of George Volichenko
Raw text
Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 Page 1 of 43 AMENDED Exhibit 12 PLAINTIFFS' OMNIBUS OPPOSITION TO DEFENDANTS' MOTIONS FOR SUMMARY JUDGMENT Case No.: 4:22-md-03047-YGR MDL No. 3047 In Re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation 1 Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA Page 2 of 43 2 IN RE: SOCIAL MEDIA ) 3 ADOLESCENT ADDICTION/PERSONAL ) INJURY PRODUCTS LIABILITY ) MDL No. 3047 4 LITIGATION ) ) 5 THIS DOCUMENT RELATES TO: ) ALL ACTIONS ) 6 SUPERIOR COURT OF THE STATE OF CALIFORNIA 7 FOR THE COUNTY OF LOS ANGELES SPRING STREET COURTHOUSE 8 COORDINATION PROCEEDING ) 9 SPECIAL TITLE [RULE 3.400] ) Lead Case No. ) for Filing 10 SOCIAL MEDIA CASES ) Purposes ) 22STCV21355 11 MONDAY, DECEMBER 16, 2024 12 CONFIDENTIAL 13 ATTORNEYS' EYES ONLY PURSUANT TO PROTECTIVE ORDER 14 15 16 17 18 19 20 Videotaped deposition of George Volichenko, held at the offices of Lieff Cabraser Heimann & Bernstein, LLP, 275 Battery Street, 29th Floor, San Francisco, California, commencing at 9:05 a.m. Pacific Time, on the above date, before Carrie A. Campbell, Registered Diplomate Reporter, Certified Realtime Reporter, Illinois, California & Texas Certified Shorthand Reporter, Missouri, Kansas, Louisiana & New Jersey Certified Court Reporter. 21 22 23 24 25 Job No. 7053449 877-370-3377 Page 1 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER APPEARANCES : MOTLEY RICE LLC 1 2 3 BY: 4 5 6 PREVIN WARREN pwarren@motleyrice.com ABIGAIL BURMAN aburman@motleyrice.com 401 9th Street NW, Suite 630 Washington, DC 20004 Page 3 of 43 Page 2 (202) 386-9610 7 8 and 9 MOTLEY RICE LLC 10 BY: ANNIE E. KOUBA (VIA ZOOM) akouba@motleyrice.com 11 TOPE O. LEYIMU (VIA ZOOM) tleyimu@motleyrice.com 12 JODI WESTBROOK FLOWERS (VIA ZOOM) jflowers@motleyrice.com 13 SARA O. COUCH (VIA ZOOM) scouch@motleyrice.com 14 P. GRAHAM MAIDEN (VIA ZOOM gmaiden@motleyrice.com 15 28 Bridgeside Boulevard 16 17 Mount Pleasant, South Carolina 29464 (843) 216-9000 and 18 19 MOTLEY RICE LLC BY: JESSICA COLOMBO (VIA ZOOM) 20 jcolombo@motleyrice.com 20 Church Street 21 Hartford, Connecticut 06103 (860) 746-1248 22 23 and 24 25 Golkow Technologies, 877-370-3377 A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 4 of 43 Page 24 1 introduce yourself to the jury. 2 Where did you grow up? 3 A. Hi, everyone. My name is 4 5 6 7 8 Georgy Volichenko. I go by George, but my legal first name is Georgy with a Y. I grew up in Moscow, Russia. I college when I was 18, so about 16 years ago. And, yeah. came to the US for 9 Ω. Okay. Can you tell us a little 10 bit about your education? 11 A. I went to school at Carnegie 12 Mellon University in Pennsylvania. I studied 13 statistics, did a bachelor's and master's, 14 15 and then went on to data science analytics work. Yeah. 16 Ω. Okay. What was your first job 17 after receiving your master's? 18 A. My first job was at a company 19 20 called Comscore, which is an advertising analytics company, and it was in Reston, 21 Virginia. 22 2. Okay. At some point did you 23 leave Comscore? 24 A. I did. I did. About three 25 years after I started working there. 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 5 of 43 Page 25 1 Ω. Okay. And where did you work 2 next? 3 A. I worked at Facebook before it 4 was called Meta. So it was 2015 2016 5 through 2018, for two years. 6 Ω. Okay. And what sort of work 7 did you do during that stint? 8 A. At Facebook? 9 Ω. Yes. 10 A. I worked for the marketing 11 department, consumer marketing, doing 12 analytics work, helping with measurement of 13 marketing campaigns. 14 Ω. 15 Okay. And how was your experience at Facebook during that first 16 stint? 17 A. It was pretty great, actually. 18 I went there specifically. I was following 19 20 my manager from Comscore, who got a job there and was, you know, kind of hiring for her 21 team. So it's one of the main reasons I kind 22 of made that jump, is because I wanted to 23 work with her. 24 We had, you know, a good team, 25 good kind of like office culture. It was, 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 6 of 43 Page 26 1 you know, before any of the remote work 2 transition, SO it was 3 Ω. 4 A. 5 6 it was a good time. Okay. Why did you leave? I left because I wanted to explore other career options, kind of spend some time on introspection and doing some 7 8 traveling. Ω. Yeah. Okay. Did you have other work 9 10 A. 11 12 13 14 15 16 17 before you subsequently returned to Facebook? Oh, yes. Yes. I worked at several startups, yeah, like larger startups, like Strava and Calm, the meditation app. And I even worked at Zillow briefly, the real estate app I'm sure everyone knows. And, yeah, I didn't get back to working at Meta until much later. Like 2022. I didn't 18 Ω. Okay. 19 A. It was a year. 20 Ω. And you referred to Meta. When 21 you returned to the company, had it been 22 rebranded as Meta from Facebook? 23 24 A. Ω. That's correct. Okay. Did you interview to 25 work at Meta or for a specific product within 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 7 of 43 Page 27 1 Meta? 2 A. Good question. So the - for 3 4 data science, the interview process is usually general. Sometimes you get a chance 5 to kind of talk to a specific team. 6 2. Uh-huh. 7 A. But that happens pretty much at 8 9 the end of the interview process when there's some sort of indication that you might be a 10 good candidate. So my interview process was 11 general. 12 Ω. Okay. And where did you wind 13 up working within Meta when you rejoined? 14 A. Yeah. So I had a specific kind 15 16 of preference to work on an integrity-focused team, which is, you know, a very general 17 term, but so my recruiter at the time 18 linked me with the team that I ended up 19 working at was teen mental health team within 20 Instagram. 21 Ω. 22 Okay. Is there any particular reason you were interested in working on teen 23 mental health issues? 24 MR. CHAPUT: Object to the 25 form. 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 8 of 43 Page 28 1 You can go ahead and answer. 2 THE WITNESS: Yeah. So 3 QUESTIONS BY MR. WARREN: 4 Ω. Well, let me rephrase the 5 question then. 6 Were you interested in working 7 on 8 A. teen mental health issues in particular? Not in particular. I think it 9 was a matter of kind of vacancy. 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 I But I had some kind of surface-level experience in my first stint at Facebook on a team called protect and care, believe it was called, and they were focused on, you know, kind of protecting users from malicious behavior. Specifically it was something around some bad actors and, like, ways that you could identify that and basically create safeguards for users. And I thought it was really interesting. And SO when I came back for my second stint and did the interviews, I kind of described that experience and said that something along those lines would be really cool. 25 And they linked me up with teen 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 9 of 43 Page 29 1 mental health. I think could you repeat 2 your question, just to make sure I'm not 3 Ω. I can just ask another one. I 4 think 5 A. Okay. 6 Q. you answered that one. 7 Were you excited to work on 8 teen mental health issues? 9 A. Yes. Yes, I was. I was. 10 Ω. And why was that? 11 A. Because I think, first of all, 12 I think it's a it's a nontrivial problem. 13 It involves some really novel challenges that 14 I hadn't previously encountered in my career, 15 16 17 SO that was exciting. But also for reasons of just, you know, making a positive difference, I guess, is another reason. 18 2. 19 Do you have any personal experience in your life with teen mental 20 health issues and social media? 21 MR. CHAPUT: Object to form. 22 THE WITNESS: No, not really. 23 QUESTIONS BY MR. WARREN: 24 Ω. Okay. During your time at 25 Instagram and, sorry, let me withdraw the 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 10 of 43 Page 30 1 question. 2 Did you work specifically on 3 Instagram or on other Meta products? 4 A. Only Instagram. 5 2. Okay. During your time at 6 7 Instagram, would you say you developed an understanding of its business model? 8 MR. CHAPUT: Object to the 9 form. 10 11 THE WITNESS: I would say MR. WARREN: Hold on 12 a Counsel, what's the basis for second. 13 the objection so I can rephrase it if 14 necessary? 15 MR. CHAPUT: It's vague. 16 17 Ω. QUESTIONS BY MR. WARREN: Okay. Mr. Volichenko, let me 18 reask you the question. I just want to know 19 if you think it's vague. 20 21 During your time at Instagram, would you say you developed an understanding 22 of its business model? 23 A. I would say in general terms, 24 25 yes. I probably developed a better understanding in my first time at Facebook. 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 11 of 43 Page 31 1 Ω. Okay. Could you explain the 2 business model of Instagram then? 3 MR. CHAPUT: Object to the 4 form. Foundation. 5 THE WITNESS: I can continue? 6 QUESTIONS BY MR. WARREN: 7 Ω. You can. 8 A. The business model is well, 9 it's based on advertising, right? So they 10 sell ads. 11 There's several things that I 12 13 14 15 think are unique about their business model. Well, probably not unique anymore with other kind of companies doing similar stuff, but I think they were the first. And first 16 offering really good targeting, right? 17 18 So you can, as an advertiser, you can kind of go after users - as an 19 that have specific interests or live in a 20 specific geography and basically just have a 21 lot more -- a lot higher likelihood that 22 people who see your ads will be interested in 23 your product because of targeting. 24 25 But also I think that the business model of Instagram and, you know, 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 12 of 43 Page 32 1 2 3 4 5 other social networks very much SO is kind of maximizing the amount of time and attention that people kind of dedicate to the platform, thus allowing for just, like, more space and time to present the advertising. 6 7 So I think that's, like, really big part of the business model that a 8 maybe we just don't - I don't know. It's 9 not a top of mind for people, but it's very 10 important. 11 Ω. 12 Okay. Does Meta make more money if its users use Instagram more 13 frequently? 14 MR. CHAPUT: Object to form. 15 Foundation. 16 THE WITNESS: Yes, absolutely. 17 QUESTIONS BY MR. WARREN: 18 Ω. 19 Does Meta make more money if its users use Instagram for longer periods of 20 time? 21 22 23 MR. CHAPUT: Object to form. Foundation. THE WITNESS: Yes. 24 QUESTIONS BY MR. WARREN: 25 Ω. 877-370-3377 And all things being equal, Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 13 of 43 Page 40 1 2 3 the data scientist that was before me was also involved in that. don't remember his name. Unfortunately, I 4 And then, you know, the rest of 5 the team is definitely there, SO software 6 7 engineers, designer. There's was it content designer, which is like a content a 8 copyrighter. There's a person who does kind 9 of qualitative research, like a UX 10 researcher. 11 12 13 For all those people, I can't say for sure who was working on a project because it was before my time. 14 Ω. Are you familiar with the term 15 "p-NAC"? 16 A. Yes. 17 Ω. What is p-NAC? 18 A. Yeah. p-NAC is SO NAC 19 20 21 22 23 24 stands for negative appearance comparison. p-NAC is basically probability. So Pis lowercase, which is often used in statistics as, like, probability. You know, you've heard of p-values also related to probability. 25 So P-NAC was a model they 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 14 of 43 Page 41 1 2 3 4 5 6 7 8 developed on my team, and I think this guy Justin, Justin Cheng, was, like, the main kind of the - the brain behind it. It was model that was scoring content, right, on how likely it is to be associated with negative appearance comparison. So you could imagine, like, for example, like, a picture of a a Christmas tree, 9 low p-NAC. A picture of, like, 10 some beautiful girl in, like, a bikini, posing 11 would be potentially high p-NAC. 12 Ω. 13 14 Okay. And was the p-NAC scoring model part of the solution that the team proposed to address negative appearance 15 16 17 comparison? A. Yes. MR. CHAPUT: Object to form. 18 19 Foundation. QUESTIONS BY MR. WARREN: 20 Ω. Okay. Was that project subject 21 to a review process? 22 A. Yeah, absolutely, like every 23 other project. 24 Ω. Okay. And what's your 25 understanding of that review process? 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 15 of 43 Page 44 1 MS. KRAMER: Object to form. 2 Foundation. 3 QUESTIONS BY MR. WARREN: 4 Ω. Let me let me withdraw the 5 question. 6 Do you know if Adam Mosseri 7 approved the negative appearance comparison 8 project? 9 A. I know for a fact that he did 10 not approve it. 11 2. What happened? 12 A. So the well, SO the model is 13 only part of the part of the solution that 14 was proposed, right? 15 Kind of the key component is 16 that content gets essentially either 17 down-ranked or completely removed if it has 18 high p-NAC, right? 19 So - but only for teens, 20 21 22 23 right? And all of the stuff that we did was specifically for users who stated that their age is, you know, within our definition of teen, which is 13 to 17. 24 And so the solution that was 25 proposed is, like, hey, we have this model, 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 16 of 43 Page 45 1 can we test kind of down-ranking it for teen 2 users. 3 And Adam had, like, a very 4 clear very clearly kind of denied us that 5 opportunity because -- I believe the reason 6 7 8 9 10 11 was that we, as a platform, should not pass judgment on the content because, like, when we start doing that, it might, like, affect content creators. They will be kind of unsure why their content was potentially down-ranked. And it's just, like, a can of 12 worms that we don't want to go there. 13 And yeah. So that was 14 that was the reason that it was, like, shut 15 down at that point. 16 Ω. 17 Do you know how Adam Mosseri communicated his decision concerning the 18 negative appearance comparison project? 19 A. I know that there were - 20 21 22 it was 23 was an e-mail thread about it with people weighing in, including my PM. everyone in between, which is He was on it and, like, there a lot of well, mainly a lot of 24 people. 25 877-370-3377 And I'm sure there were other Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 17 of 43 Page 46 1 methods of communication that I'm not aware 2 of. 3 Ω. Are you aware of whether 4 5 6 Mr. Mosseri expressed any view in that e-mail chain about negative appearance comparison content relative to other kinds of content on 7 Instagram, how prevalent it was? 8 MR. CHAPUT: Object to form. 9 QUESTIONS BY MR. WARREN: 10 Ω. 11 Let me reask the question. Are you aware of whether 12 13 14 15 Mr. Mosseri expressed any view about how prevalent negative appearance content is on the Instagram platform relative to other kinds of content? 16 MR. CHAPUT: Object to form. 17 THE WITNESS: I'm not. No, I'm 18 not aware of that. 19 QUESTIONS BY MR. WARREN: 20 Ω. Did you come to understand 21 whether Mr. Mosseri's decision not to move 22 23 24 forward with the p-NAC project had any impact on the morale of the Instagram mental well-being team? 25 A. Well, my impression I can't 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 18 of 43 Page 47 1 2 know for sure, but my impression was that they were pretty disappointed because they 3 spent a lot of time, and they thought - they 4 thought very they believed very strongly 5 in the need for this, right? And I think 6 there was even, like, some surveys and 7 research that this is a big problem, and they 8 wanted to work on it. 9 They felt like they had a 10 potential solution, but without without 11 being able to, quote/unquote, pass judgment 12 on content, they felt like they had their 13 hands fully tied. 14 15 And in combination with the lack of clarity in what's coming next, this 16 was all kind of compounding in a, you know, a 17 low morale. That was my impression. 18 Ω. Okay. And you testified 19 earlier that you came to develop a view as to 20 why the data scientist before you and the 21 data engineer had left. 22 A. Yeah. 23 Ω. What was that view? 24 MR. CHAPUT: Object to form. 25 Foundation. Golkow Technologies, 877-370-3377 A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 19 of 43 Page 48 1 2 3 4 THE WITNESS: My view is that they were likely just didn't see the point in being there. I mean, there could be other reasons. And they 5 wanted to -- kind of similar to me, 6 they wanted to make 7 8 difference, and they felt like they just didn't have an opportunity to do that. a 9 QUESTIONS BY MR. WARREN: 10 Ω. 11 12 Okay. Now, you talked about the review process that the negative appearance comparison project went through. a bit 13 14 15 16 In your personal experience, is that typical of the review processes that proposed safety features had to undergo at Instagram? 17 MR. CHAPUT: Object to form. 18 19 20 21 22 23 24 25 THE WITNESS: Yes, pretty typical. I think there's on the kind of level of how much of the app experience the feature involves, it can go higher up, like in that case to Adam, stop at, like, well-being leadership something that does not depending or it can kind of is 877-370-3377 if it's Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 20 of 43 Page 68 1 2 2. 3 move 4 QUESTIONS BY MR. WARREN: Did the well-being team want to forward with the intervention to reduce negative appearance comparison? 5 MR. CHAPUT: Object to form. 6 Foundation. 7 THE WITNESS: Yes. 8 QUESTIONS BY MR. WARREN: 9 Ω. Okay. But they were not 10 allowed to do so by Instagram leadership? 11 MR. CHAPUT: Object to form. 12 Foundation. 13 THE WITNESS: Yes. 14 QUESTIONS BY MR. WARREN: 15 Ω. Okay. So I'd like to ask you 16 some questions about what your team pivoted 17 to after work on negative appearance 18 comparison came to an end. 19 Did the team begin work on 20 something called problematic use? 21 A. Not immediately after, but 22 shortly, shortly after, yes. 23 Ω. Okay. And was problematic use 24 another way of referring to social media 25 addiction? 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 21 of 43 Page 96 1 the phrase "screenshot risk"? 2 A. Screenshot risk? 3 Ω. Yeah. 4 A. No. 5 Ω. Okay. George, were you 6 involved in evaluating to what extent the 7 Take a Break and alternative topics features 8 were actually used by Instagram users? 9 A. Yes. 10 (Volichenko Exhibit 3 marked 11 for identification.) 12 QUESTIONS BY MR. WARREN: 13 Ω. Okay. So I'm going to direct 14 your attention to another document ending in 15 Bates number 399807. This will be Volichenko 16 Exhibit 3. 17 And go ahead and take a minute 18 to review this, and just let me know when 19 you're ready. 20 A. Okay. 21 Ω. Okay. So why don't we start on 22 page 4 of this document. 23 A. Okay. 24 Ω. Does this indicate that you 25 opened this conversation on June 9, 2022? 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 22 of 43 Page 97 1 A. 2 this page. Wait. Page 4. Is that Okay. oh, 3 Ω. You can also look on your 4 screen. 5 A. Yes, it does indicate that. 6 Ω. Okay. And would you have 7 opened and reviewed this document in the 8 ordinary course of your work at Meta? 9 10 A. Ω. Correct, yes. Okay. So flipping back to 11 page 1, just looking at the title, do you see 12 the words "IGWB e-mail review"? 13 A. Yes. 14 2. What does IGWB stand for? 15 A. Instagram well-being. 16 Ω. And then it says "mental 17 well-being." 18 Right? 19 A. Yes. 20 Ω. Is that the name of the 21 specific team you worked for? 22 A. Yes. 23 Ω. And the next part of the title 24 is "H1 2022." 25 What does H1 mean? 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 23 of 43 Page 98 1 A. First half of 2022. 2 Ω. Okay. And then it says, "Goal 3 methodology update V2." 4 Right? 5 A. Yes. 6 Q. Okay. At some point while you 7 were working at the Instagram mental 8 well-being team, did the team reduce its goal 9 for the number of userS that would adopt Take 10 a Break? 11 12 MR. CHAPUT: Object to form. THE WITNESS: I think so. I 13 mean, I may be misremembering. 14 QUESTIONS BY MR. WARREN: 15 Ω. Okay. Can you read the first 16 paragraph to the jury, please? 17 A. "Thanks for the feedback. Team 18 19 20 21 22 23 will keep meaningful adoption and leave as percentage but does not believe we have path to hitting that goal, SO would like to request we reduce .6 percent to .25 percent given the tactics we have aligned and prioritized for this half. I have reflected a a 24 these changes in V2 proposal below." 25 2. Okay. Does that refresh your 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 24 of 43 Page 99 1 recollection as to whether the team reduced 2 its goal? 3 A. 4 that it was done early on in my Yes. Just wanted to clarify in my time 5 on 6 the team, and potentially the the decisions were -- some of you know, it was it 7 might have been initiated prior to my time as 8 well. 9 Ω. Okay. So let's take a look at 10 the table on the first page. 11 A. Okay. 12 Ω. Looking at the first row, what 13 14 was the team's revised goal in terms of the percentage of weekly active teen Instagram 15 userS that adopted Take a Break? 16 A. Sorry, are you asking about the 17 goal or the actual amount that 18 Ω. The goal. The goal. 19 A. It's unclear for Take a Break 20 specifically. So what they're saying here is 21 that the goal for adoption, the combined 22 adoption of all the features, was 23 .25 percent. 24 Ω. Okay. And if you just look 25 three rows up, is there a row specific to 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 25 of 43 Page 104 1 2 3 Do you know what the author of this statement is referring to when they refer to 4 5 6 a one-way door principle? A. Ω. I believe I do, yes. And what is that? A. So one-way door refers to 7 8 features that once we launch them, we basically cannot disable them, right? 9 With some -- with some features 10 that I would say have, like that are less 11 sensitive, right, you can - you can in 12 theory, like, enable it. You know, say it 13 14 doesn't work or it, like, affects something negatively, you can shut it down. 15 With features with some 16 other features, especially features in this 17 18 kind of well-being space, once you put it out there, you know, you can't turn it off 19 because - 20 reasons, well, I believe there are multiple but the most obvious one I can think 21 22 23 24 25 userS of is that, like, it just looks really bad, right? Like the company is, like, doing something to protect the users, but then they're, like, eh, don't want to protect We're going to turn it off. we 877-370-3377 anymore. Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 26 of 43 Page 105 1 So that is at least a component 2 of why these one-way door features exist. 3 And from what I remember, the 4 guidance that we received from our 5 product manager, 6 7 8 9 was that a vast majority of things that we work on would fall into that category, right? So for Take a Break, for example, if we make it opt-out, when we -- 10 11 when we test it, right, random sample of users. we enable it for some And let's say it 12 doesn't work. It tanks a bunch of metrics 13 that we can't we can't, you know, effect. 14 But for those users, we already have it's 15 out there. We can't we can't turn it off 16 for them. Which is not the end of the world, 17 but it creates a very fragmented user base, 18 right? 19 It's, like, some people have it, others don't. We, like, lost track of who 20 has it. It gets really messy. 21 22 And I don't know if that is really the extent of the reasoning there, 23 but, yeah. 24 Ω. Okay. Thank you. 25 So now that you've explained 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 Page 27 of 43 PURSUANT TO PROTECTIVE ORDER Page 122 1 2 3 4 the time is that, look, TikTok is doing it, opt-out. They're just kind of enabling it for everyone. Should we consider doing that. And, yeah, we just felt like we 5 weren't 6 7 we weren't, again, like, aggressive enough, and we weren't necessarily taking a stance on this matter. 8 (Volichenko Exhibit 6 marked 9 for identification.) 10 QUESTIONS BY MR. WARREN: 11 Ω. Uh-huh. Okay. 12 Let me show you another 13 document, and this will be a document ending 14 in Bates Stamp 5017, and we'll mark it as 15 Volichenko 6. 16 A. Thank you. 17 Ω. Do you recall this 18 presentation? 19 A. Yes. 20 Ω. At some point while you were 21 working at Meta, did you have accesS to this 22 presentation? 23 A. I did. 24 Ω. Okay. So just starting on the 25 first page, what's the title of this 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 28 of 43 Page 123 1 presentation? 2 A. 3 "Late Night Use, Instagram Mental Well-Being, Problematic Use, Strategy 4 and Design." 5 Ω. And do you see some text at the 6 bottom in gray? 7 A. Yes. 8 2. Can you read that to the jury? 9 A. "A/C Priv, Communities Review, 10 August 4, 2022." 11 Ω. What does A/C Priv mean? 12 A. Attorney-client privilege, I 13 believe. 14 Ω. Do you recall receiving any 15 kind of instruction regarding when to place 16 the words "A/C Priv" on documents? 17 MR. CHAPUT: Object to the 18 form. 19 QUESTIONS BY MR. WARREN: 20 Ω. You can answer. 21 A. Okay. I thought there was more 22 coming. 23 I do recall at some point in my 24 employment at Meta, or maybe Facebook - 25 prior to my first time that I was working at 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 29 of 43 Page 124 1 Facebook, we had some training on when and 2 where to use the A/C Priv label. 3 Ω. And what was your understanding 4 as to when to use that label? 5 A. The gist of it was, like, if 6 7 8 9 a you don't feel comfortable it being public, your conversation kind of, like, leaking to the public, and if you feel like it would be, like, sensitive topic or this conversation 10 11 12 might be misconstrued, you should label it A/C Priv, and then it's, like, safe from from everyone's eyes. 13 2. Do you know what the letters 14 A/C stand for? 15 A. I'm assuming it's 16 attorney-client. 17 Ω. Okay. 18 A. 19 Ω. I mean, I'm pretty sure it is. Okay. Let's 20 A. Alternating current? I don't 21 know. 22 2. Okay. Let's start on slide 10. 23 24 25 MR. CHAPUT: Counsel, sorry. Before you proceed, I don't see the confidentiality designation on the 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 30 of 43 Page 130 Objection. 1 QUESTIONS BY MR. WARREN: 2 Ω. And what does it indicate? 3 4 5 MR. CHAPUT: Foundation. THE WITNESS: 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Yeah, I think this is based on kind of user research. So this is kind of the more qualitative research that would be doing. And SO some of the findings is that teens feel pressure to be available at all times, there's social norms around kind of being available and being online, and that Instagram is used to kind of wind down before like, when you're in bed before going to sleep, and that the existing tools within the operating systems as well as Instagram itself are not sufficient currently. QUESTIONS BY MR. WARREN: Ω. Okay. Let's go back to the executive summary on slide 3, please. A. Which slide? Ω. Slide 3. Golkow Technologies, 877-370-3377 A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 31 of 43 Page 131 1 2 3 4 A. 5 6 7 8 Could you read to the jury the very last paragraph in dark gray that begins with "Quiet Mode"? "Quiet Mode: A controls-based bundle that helps teens manage their relationship with Instagram at night engagement triggers to pull them into the app, and less stimulating experience when a fewer 9 10 browsing the app." Ω. What are engagement triggers? 11 12 13 MR. CHAPUT: Foundation. THE WITNESS: Object to form. 14 15 16 the app, or 17 Engagement triggers would be anything that has the goal of, like, bringing you into if you're into the app, bringing you into other surfaces, 18 right? 19 20 So most of the time about badges, which is, like, the we think 21 22 little red dot over the app icon on your home screen. Especially it has 23 some, like usually it has, like, a 24 number on it. That is massively 25 effective as an engagement trigger, 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 32 of 43 Page 132 1 2 3 4 5 to -- especially for people who kind of have this level of kind of OCD-like behavior that they should be all cleared. So you kind of see what I'm going what's going on there, 6 right? 7 8 9 10 11 12 And then when you're inside the app, you have the same kind of badge on your inbox icon in the top right corner. Those kinds of things are the engagement triggers. And part of the functionality 13 of Quiet Mode that we definitely 14 considered I don't know if it was, 15 like, launched as part of the main 16 feature - is suppressing those. So 17 basically we're not we're not 18 trying to bring you into the app 19 during Quiet Mode hours with the 20 feature on. 21 22 Ω. 23 24 25 QUESTIONS BY MR. WARREN: Okay. So is it correct that Meta has the ability to vary the amount of engagement triggers that pull users into Instagram? 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 1 A. PURSUANT TO PROTECTIVE ORDER Yeah. Page 33 of 43 Page 133 2 3 MR. CHAPUT: Object to form. Foundation. 4 THE WITNESS: Yes. 5 QUESTIONS BY MR. WARREN: 6 Ω. Okay. So it can increase those 7 engagement triggers to pull in more users? 8 MR. CHAPUT: Object. Object to 9 the form. Foundation. 10 11 12 13 14 15 16 17 18 THE WITNESS: Well, they have the ability to, you know, kind of fire them whenever, but usually they are caused by some activity. You know, you're getting a message "like" or whatever, right? So I don't think they can necessarily increase them because they're tied to events, but they can definitely suppress them. or a 19 QUESTIONS BY MR. WARREN: 20 Ω. Okay. 21 A. Turn them off. 22 Ω. So I guess let me reask the 23 question. 24 25 Can Meta vary when, to what extent and to whom it deploys engagement 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 34 of 43 Page 134 1 triggers? 2 A. Yes. 3 MR. CHAPUT: Object to the 4 form. Foundation. 5 QUESTIONS BY MR. WARREN: 6 Ω. Okay. Going back to this 7 slide, it refers to a, quote, "Less 8 stimulating experience." 9 Do you see that? 10 A. Yes. 11 Ω. Does Meta have the ability to 12 13 control whether a browsing experience on Instagram is more or less stimulating? 14 15 16 MR. CHAPUT: Object to the form. Foundation. THE WITNESS: Yes. 17 QUESTIONS BY MR. WARREN: 18 Ω. How does it do that? 19 MR. CHAPUT: Object to the 20 form. Foundation. 21 THE WITNESS: Well, there are 22 many possible ways to do that. I 23 think I guess it depends on what 24 how you define "stimulating." 25 877-370-3377 I think in this case I would Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 35 of 43 Page 135 1 say, like, you can 2 3 that you are 4 you can, like, limit the number of pieces of content -- you're, like, presenting, putting in front of the 5 user, right? 6 7 is a 8 9 10 11 So, like, I think currently it it's like an infinite scroll news feed, right? You just keep scrolling and scrolling, and it just keeps showing you more stuff, right, generating content that algorithm 12 thinks you would like you are 13 likely to engage with or whatever it 14 is, right? 15 So 16 17 18 19 20 21 22 23 24 25 a even limiting that is hundred percent in power of Meta. You could I think we even talked about, like, making it black and white at nighttime. It'd be, like, easier on the eyes in the dark, but also just less stimulating so you're more likely to, you know, shut down the app and go to sleep. That's one example. All of these, like, you know, message notifications and these badges 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 36 of 43 Page 136 1 and these engagement triggers, they 2 3 can be suppressed. Those are some examples, but, you know, you can 4 imagine many solutions here. 5 QUESTIONS BY MR. WARREN: 6 Ω. Okay. George, were you 7 responsible in whole or in part for 8 determining the metrics for success for the 9 Quiet Mode feature? 10 A. Definitely in part. I yes. 11 I don't recall who was all involved. It was 12 probably some combination of me and people 13 from the product product growth analytics 14 team. Yeah. 15 Ω. Okay. And can you turn to 16 slide 7, please? 17 A. 7. 18 Ω. I'm actually going to focus you 19 on the comments which are on the next page, 20 but why don't you take a second to review 21 this slide first. 22 A. 23 Success Definition. Okay. Okay. 24 Ω. And the slide is entitled 25 "Success Definition." Golkow Technologies, 877-370-3377 A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 37 of 43 Page 140 1 2. Okay. So Ms. indicates 2 here that "counter-metrics will be 3 4 IG-critical metrics and surface-specific guardrails for notifs, IGD, active presence, 5 et cetera." 6 I think you've talked about 7 some of these things, but can you just decode 8 that for the jury, please? 9 A. Uh-huh, yeah. So 10 MR. CHAPUT: Object to form. 11 Foundation. 12 THE WITNESS: Counter-metrics, 13 14 15 16 essentially, yes, as you said, don't guardrails. So we want to drive some outcome, but at the same time want to harm the ecosystem sort of we 17 thing, right? 18 19 20 21 22 23 24 25 And the teams that are involved specifically with Quiet Mode are and the notification teams, IGD is IG Direct, which is the messaging within Instagram. Active presence, don't recall, but it's probably some team that it would be involved somehow, right? other 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 38 of 43 Page 141 1 2 3 4 5 a 6 7 we need to 8 we see that, we 9 Because when we're suppressing notifications, we're suppressing badges and other stuff. You'll imagine that there will be see kind of a hit on those metrics. So that's why we would, like when we if and when need to go to those teams and be, like, hey, at least we we'll 10 are improving the situation with late 11 night time spent. So maybe you can 12 work with us and give us an approval. 13 QUESTIONS BY MR. WARREN: 14 Ω. 15 16 Okay. How important did those counter-metrics wind up being in shaping the version of Quiet Mode that Meta ultimately 17 released on Instagram? 18 MR. CHAPUT: Object to form. 19 Foundation. 20 THE WITNESS: From what I 21 22 we kind of saw remember, not very important, because a similar thing that we 23 saw previously with Take a Break, is 24 25 that there's just not enough impact across -- there's not the board, you 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 39 of 43 Page 142 1 know. 2 3 4 a 5 6 7 8 9 Like, the late night time spent, I don't remember if like, statistically significant impact there or not in our testing. And same thing for these critical metrics and other teams' guardrails. So it's just, like, kind of doing nothing. we saw, 10 And as I 11 was saying, the optimal scenario is that it is 12 doing you know, it is reducing our 13 main problematic use metric, in this 14 15 case late night time spent, and also doing some damage on the those 16 guardrails. That's how we know it's, 17 like, moving the needle. 18 QUESTIONS BY MR. WARREN: 19 Ω. So if counter-metrics related 20 to growth and usage had not been considered, 21 22 do you think Instagram could have released version of Quiet Mode that was more a 23 protective of teen health? 24 MR. CHAPUT: Object to form. 25 Foundation. Golkow Technologies, 877-370-3377 A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 40 of 43 Page 143 1 THE WITNESS: I believe So, 2 yes. 3 (Volichenko Exhibit 7 marked 4 for identification.) 5 QUESTIONS BY MR. WARREN: 6 Ω. 7 Okay. Let me direct your attention to another document. This ends in 8 the Bates stamp 504271, and it will be 9 Volichenko 7. 10 So as a very preliminary 11 matter, is this an e-mail from ? 12 A. Yes. 13 Ω. And are you on the "to" line of 14 this e-mail? 15 A. I am, though I 16 working there at the time? I guess was I even I 17 guess I was. 18 Ω. Okay. 19 A. It was January. Yes. Yes. 20 Yes. Okay. Sorry. Yes. 21 Ω. I mean, did you have accesS to 22 your meta.com e-mail after you left the 23 company? 24 A. No. 25 Ω. 877-370-3377 Okay. So you would have been Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 41 of 43 Page 164 1 financially. My wife is mad at me to this 2 day. But that's her problem, right? 3 Ω. All right. Thank you, George. 4 I just have a few more 5 6 7 8 questions about Quiet Mode, and I think we'll be able to wrap at least my portion of the examination relatively soon. So you've spent a lot of time 9 talking about opt-in versus opt-out. 10 Was Quiet Mode launched as an 11 opt-in feature or an opt-out feature? 12 A. I'm pretty sure, if my memory 13 is accurate, it was launched as an opt-in 14 feature. 15 Ω. Okay. Would it have been 16 better for teen safety, in your experience, 17 if it had been launched as an opt-out default 18 for teenagers? 19 MR. CHAPUT: Object to form. 20 Foundation. 21 THE WITNESS: Yes, I think much 22 like Take a Break, it would have 23 been it would have been a good 24 move. 25 877-370-3377 There are some differences, Golkow Technologies, A Veritext Division www.veritext.com Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER Page 42 of 43 Page 165 1 however, because we 2 would need to set, like, in that case we a default 3 time window of what constitutes 4 5 nighttime and, you know, that might be challenging just from a technical 6 perspective. 7 8 9 10 11 shift at 12 13 14 But also, like, what if you have a teenager that, like, works you know, doesn't go to school but, a night like, works at a a -- works - toll booth, right? So they're not, like, sleeping anyway. Maybe that's how they kind of pass time while there's no cars at night. 15 So there's there were, like, 16 17 18 19 20 circumstantial and kind of, like, tactical things that can make it challenging. But overall, I think that would have been more effective for sure. 21 QUESTIONS BY MR. WARREN: 22 Ω. 23 24 25 Okay. And speaking of default hours, even for the opt-in version of Quiet Mode, did Instagram have to set what the default nighttime hours were going to be? 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com 1 2 3 4 5 6 7 8 9 Case 4:22-md-03047-YGR Document 2885-4 Filed 03/26/26 PURSUANT TO PROTECTIVE ORDER CERTIFICATE Page 43 of 43 I, CARRIE A. CAMPBELL, Registered Diplomate Reporter, Certified Realtime Reporter and Certified Shorthand Reporter, do hereby certify that prior to the commencement of the examination, George Volichenko, was duly sworn by me to testify to the truth, the whole truth and nothing but the truth. I DO FURTHER CERTIFY that the foregoing is a verbatim transcript of the testimony as taken stenographically by and before me at the time, place and on the date hereinbefore set forth, to the best of my ability. Page 263 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 neither a I DO FURTHER CERTIFY that I am relative nor employee nor attorney a relative nor nor counsel of any of the parties to this action, and that I am neither employee of such attorney or counsel, and that I am not financially interested in the action. Cume A. Campbell CARRIE A. CAMPBELL, NCRA Registered Diplomate Reporter Certified Realtime Reporter California Certified Shorthand Reporter #13921 Missouri Certified Court Reporter #859 Illinois Certified Shorthand Reporter #084-004229 Texas Certified Shorthand Reporter #9328 Kansas Certified Court Reporter #1715 New Jersey Certified Court Reporter #30XI00242600 Louisiana Certified Court Reporter #2021012 Notary Public Dated: January 3, 2025 25 877-370-3377 Golkow Technologies, A Veritext Division www.veritext.com