Meta & Children's Online SafetyHave we made people addicted to Facebook?
Raw text
Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 1 of 29 AMENDED Exhibit 140 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 cebook/1766984813316375/?he location=ufi Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 2 of 29 Seriously, I need to work! LOOK AT ME! LOOK AT ME! f Have we made people addicted to Facebook? THURSDAY, AUGUST 3, 2017 READING TIME: 9 MINUTES manager This summer, I've been interning on the News Feed Ecosystems project as a Data Science Intern working to answer the following: "Are there people addicted to Facebook and if so, how can we identify and size this population so we can help them?" When my intern first called me to tell me this, the timing seemed comical. Just an hour before, I had deleted the Facebook app from my phone because I thought I was obsessively using Facebook. By deleting the app, I was hoping to reclaim my time and hopefully decrease my internalized instinct to keep opening Facebook. Now, I had lez mv 12 weeks here would be around figuring out how we can improve the Facebook Chats MacBook Pro Confidential EXHIBIT 9 HAUGEN-9/25/25 Haugen_00016893 Case 4:22-md-03047-YGR W bk/1766984813316275/?hc_location=ufi W Document 288557 File 0326/26 Page 3 of 29 W This summer, I've been interning on the News Feed Ecosystems project as a Data Science Intern working to answer the following: "Are there people addicted to Facebook and if so, how can we identify and size this population so we can help them?" When my intern first called me to tell me this, the timing seemed comical . Just an manager, hour before, I had deleted the Facebook app from my phone because I thought I was obsessively using Facebook. By deleting the app, I was hoping to reclaim my time and hopefully decrease my internalized instinct to keep opening Facebook. Now, I had learned, my 12 weeks here would be around figuring out how we can improve the Facebook experience to make sure people like me don't feel the need to just leave Facebook. As I wrap up my internship, here's what I learned this summer in terms of addiction to Facebook. What does it even mean to be addicted to Facebook? How can we understand what being addicted to Facebook means? At the start of this project, Ime Archibong, and myself talked with Nir Eyal, author of the book Hooked. In this book, Nir discusses his extensive research into creating habit-forming products and the Hook model, "a four-step process embedded into the products of many successful companies to subtly encourage customer behavior" to ultimately get users to voluntarily choose to open your app without the need for external advertising. Companies like ours have implemented the hook model well. However, Nir notes that using apps like ours in some cases leads a user from having a habit of using the app to being addicted. The difference between habit and addiction from Nir's perspective is harm: an addicted user is one who cannot stop using a product to the point where it can cause them harm. Chats Haugen_00016894 Confidential W W W Case 4:22 -md-03047-YGR bok/1766984813316275/?hc_location=ufi W W Document 2885-57 Filed 03/26/26 Page 4 of 29 Companies like ours have implemented the hook model well. However, Nir notes that using apps like ours in some cases leads a user from having a habit of using the app to being addicted. The difference between habit and addiction from Nir's perspective is harm: an addicted user is one who cannot stop using a product to the point where it can cause them harm. "When we feel overly stressed, we seek serenity, perhaps finding relief in sites like Pinterest. When we feel lonely, destinations like Facebook and Twitter provide instant social connections" While there are many factors that could lead to someone becoming addicted to Facebook, Nir noted that he expected the number of addicted people to be low. I hoped so too based on his definition. Now let's get into the data behind this project. Deactivations While working on studying how to find addicted users, we realized that deactivations data could be quite helpful in this process. To explain how, I'll show the current deactivati flow Chats MacBook Pro Confidential Haugen_00016895 W Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 5 of 29 k/1766984813316275/?hc_location=ufi Deactivations While working on studying how to find addicted users, we realized that deactivations data could be quite helpful in this process. To explain how, I'll show the current deactivations flow. f Search Facebook Are you sure you want to deactivate your account? Q REDACTED-SCA Home Deactivating your account will disable your profile and remove your name and photo from most things you've shared on Facebook. Some information may still be visible to others, such as your name in their friends list and messages you sent. Your 2,100 friends will no longer be able to keep in touch with you. i miss you will miss you Redacted SCA miss you will miss you will miss you Redacted - SCA Rodacted-SCA Redacted-SCA BLA Send Message Send a Message Send KAJ Message Send Message Send Message Reason for leaving (required) This is temporary. I'll be back. I have another Facebook account. I spend too much time using Facebook. I don't find Facebook useful. I don't feel safe on Facebook. I have a privacy concern. Please explain further My account was hacked. I get too many emails, invitations, and requests from Facebook. I don't understand how to use Facebook. Other, please explain further: Emall opt out Opt out of receiving future emails from Facebook Even after you deactivate, your friends can still invite you to events, tag you in phetes, or ass you to join groups. If you opt out, you will NOT receive these email invitations and notifications from your friends. Delete applications Permanently delete all applications for which I'm the only developer Note You are listed as the only developer for the fellowing Platform plications. You must either delete or transfer ownership of th Chats belfore deactivating. MacBook Pro Confidential Haugen_00016896 2 Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 6 of 29 f Search Facebook Are you sure you want to deactivate your account? Q REDACTED-SCA Home Deactivating your account will disable your profile and remove your name and photo from most things you've shared on Facebook, Some information may still be visible to others, such as your name in their friends list and messages you sent. Your 2.100 friends will no longer be able to keep in touch with you. miss you will miss you miss you Redacted-Sciss you miss you Redacted-SCA SCA Redacted-SCA RIA 2 Send Message Send Message Send Message SendMessage Message 3... b... 3 16 15 10 7 Reason for leaving This is temporary. I'll be back. (required) 1 have another Facebook account. I spend too much time using Facebook. Please explain further I don't find Facebook useful. I don't feel safe on Facebook I have a privacy concern. My account was hacked. I get too many emails, Invitations, and requests from Facebook. I don't understand how to use Facebook. Other, please explain further A Emall opt out Opt out of receiving future emails from Facebook Even after you deactivate, you there can still invite you to events, lag you in photos, or ask you to join graves γου αρχους, γ NOT receive these em tions and notifications from your friends Delete applications Permanently delete all applications for which I'm the only developer Note: You are fated as the only devemper for the following Platform applicatione. You must either deste sender Messenger Redacted-SCA Redacted-SCA dit Opt out of receiving messages on Messenger Even if you comctivate your Facebook account, you can still chat on Messenger. Your profile picture and people wil stl be able to search for you by rame to said you a message. Vault where they can ressenyου η γου ορίους, γου ΜεNOT be sois to chat on Messenger or appler to your handtagg Deactivate Cancel Haugen_00016897 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 7 of 29 #C0000円/1/00984813316275/?hc_location=ufi The flow is fairly aggressive as we try to stop people from leaving Facebook. We list your some of your friends to remind you that they will no longer be able to contact you through the site and more importantly, we require the user to choose a reason for why they're leaving. For my analysis this summer, we focused on two specific categories: people who said they spent too much time on Facebook and those who selected that their leave was temporary and that they planned to return (since many of these people listed similar reasons for deactivating their accounts). This subset provided a good signal for people who could be addicted, who ultimately leave Facebook as a solution. In a given week, approximately 5.9 million people leave Facebook for just these two reasons alone so clearly we have work to do here if we want to improve the Facebook experience so they do not feel compelled to deactivate their accounts. After settling on analyzing these two categories, I dug into these users' metrics and found a few interesting things. For the sake of keeping this note succinct, I'll only include a few of the findings and link to my full deck at the end of this for those interested in more information about what I found. Haugen_00016898 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 8 of 29 few interesting things. For the sake of keeping this note succinct, I'll only include a few of the findings and link to my full deck at the end of this for those interested in more information about what I found. People who cite too much time/temporary have a higher number of sessions Comparing Sessions of All Users and TMT/T Users (FB Blue) PDF (bucketed by a person's average sessions - blocks of 5) 35% 30% All Users: 16.19 sessions TMT/T Users: 24.41 sessions 25% 20% 15% 10% 5% 0% 5 30 55 80 105 -% of All Users 130 155 % of TMT/T Users 180 205 230 PDF distribution of number of sessions across TMT/T and all users. Haugen_00016899 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 ted-to-facebook/1766984813316275/?hc_location=ufi S M W W Page 9 of 29 Q W W X 0% 5 30 55 80 105 -% of All Users 130 155 % of TMT/T Users 180 205 230 PDF distribution of number of sessions across TMT/T and all users. One interesting insight discovered was that while people deactivating their accounts citing too much time/temporary had a similar time spent to the average user, they had a higher number of sessions in a day (we define a session as an instance where the user opens the FB app on their phone). This user subset also spent less time per session. From this, it appeared that an addicted user is not necessarily just someone who spends several hours a day on Facebook, but rather someone who had a high number of sessions compared to their time spent. Analyzing notifications data, people in the TMT/T subset both received more notifications than the average user and also responded to them faster. TMT/T users get 60% more push notifications Push Notifications Delivered (7d Average) ☆ + W Haugen_00016900 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 10 of 29 Analyzing notifications data, people in the TMT/T subset both received more notifications than the average user and also responded to them faster. TMT/T users get 60% more push notifications Push Notifications Delivered (7d Average) 16 15 14 12 10 8 6 4 14 9 2 0 TMT/T users deactivated, non TMT/T all users Visualization of average push notifications a user gets in a day. Haugen_00016901 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 11 of 29 TMT/T users also click them sooner How long it takes people to click a notification (minutes) 35 32.6 30 25 20 15 10 5 0 all users 15.9 14.6 deactivated, not tmt/t tmt/t Average time it takes a user to click on a delivered push notification. nilding Model Haugen_00016902 Confidential W nd-03047-YGR W book/1766984813316275/?hc_location=ufi Document 2885-57 M W W W Filed 03/26/26 Page 12 of 29 Q Average time it takes a user to click on a delivered push notification. Surveying Users/Building a Model We wanted to learn about whether people would admit they were addicted to Facebook. To do this, I worked with and the Communications team to make a branded survey that asked users the following questions: How often in the last month have you: •Spent a lot of time thinking about Facebook or planned use of Facebook? (0.61) •Thought about how you could free more time to spend on Facebook? (0.42) •Thought a lot about what has happened on Facebook recently? (0.55) •Spent more time on Facebook than initially intended? (0.68) •Felt bad if you, for different reasons, could not log on to Facebook for some time? (0.58) We left the survey running for a week and focused on Facebook users living in the United States. In that week, we received around 1300 responses. The main insight we found was that for most questions posed to these survey takers, people who marked often/very often had an average higher time spent and number of sessions than someone who said rarely/never (I've excluded notifications data here because while we log them well, in their current form it's hard for us to analyze the data over long periods for all users). Chats Haugen_00016903 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 13 of 29 excluded notifications data here because while we log them well, in their current form it's hard for us to analyze the data over long periods for all users). There are noticeable trends in people's answers During the past month, how often During the past month, how often have you spent more time on Facebook than initially intended? 120 have you thought about how you could free more time to spend on Facebook? 103 120 100 89 104 103 98 78 78 80 100 80 83 86 80 60 60 38 40 31 34 25 29 40 31 32 36 38 38 20 20 0 0 Never Rarely avg_sessions -avg_ts Sometimes Often Very Often Never Rarely Sample of results from user survey. Sometimes Very Often avg_sessions -avg_ts Haugen_00016904 Confidential Often Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 14 of 29 Sample of results from user survey. Next, we built out a model to try and size the number of addicted users. To build this, we used both sessions data and average time spent and looked at the deactivation rate in each bucket. Preliminary Model Results TMT/T Deactivation Rate ...... 0.9% A model of mapping addicted users. The darker the area, the more deactivated TMT/T users that were in that bucket. Note: This is only U.S. users. Haugen_00016905 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 15 of 29 liminary Model Results Tx Bucket O TMT/T Deactivation Rate Sessions Bucket 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 5 10 15 20 95 100 105 110 115 120 125 130 135 140 145 150 155 160 165 170 175 180 185 190 195 200 0% 0.9% Haugen_00016906 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 16 of 29 cebook/1766984813316275/?hc_location=ufi A model of mapping addicted users. The darker the area, the more deactivated TMT/T users that were in that bucket. Note: This is only U.S. users. From here, we plotted out the number of people that could be considered addicted based on the threshold we set. This threshold is based on the TMT/T deactivated users as a percentage of total DAP. At a minimum threshold of 1%, that would be ~15k people we could call addicted, but if we loosen the threshold to 0.2%, that's nearly 200k people. 100,000,000 Haugen_00016907 Confidential of Potentially Addicted DAP 50.500.000 1.000.000 10.000 10.000 1,000 100 Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 17 of 29 10.00% 100% 0.20% Pct Threshold 0.01% A plot with approximate line of best fit of how many users could be considered addicted based on threshold set. Note: the trend line does not fit well after 0.02 %. Haugen_00016908 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 18 of 29 How do we help these users? New tools to help people not obsessively check their account We're here to help if you need a break We want you to feel comfortable with your relationships on Facebook. As a way to support you, we created these options so you can change when you see someone on Facebook, limit what you share with that person or edit past posts with them. Select someone a X Beach Mode Beach mode hides some FB-only content from your Facebook feed and prevents FB-only groups from loading. Please note that this also disables notifications for such content (including email notifications for all FB groups activity). Enable Beach Mode + Night Mode: No Notifications between 9:00 PM To: 6:00 AM Enable Night Mode Weekend Mode: No Notifications between Friday 9:00 PM and Monday 9:00 AM Enable Weekand Mode Messenger Beach mode mutes notifications from all Messenger threads that are 90% employees. Feedback? Post in Messenger Feedback Enable Messenger Deech Mode Screenshot of Take a Break Tool and Beach Mode Haugen_00016909 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 19 of 29 Messenger Beach mode mutes notifications from at Messenger threads that we 30% employees. Feedback? Post in Messenger Feedback Enable Mesaenger Beach Mode Screenshot of Take a Break Tool and Beach Mode In order to help these users, we can create new tools that don't stop them from using Facebook, but help them feel more in control of when they choose to visit the app. To this end, we could leverage the easy to navigate structure of the Take a Break tool by the Protect and Care team. The tool, designed for breakups, makes it easy for someone to adjust their settings to see less of a person in their feed, photos, or messages. Rather than digging through the settings tab, this tool makes it easier to find the relevant settings a person may want to change. Haugen_00016910 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 We're here to help if you need a break Here are some changes we think might be helpful so you can take a break and stay friends on Facebook. Don't worry, we won't notify about changes you make. Take a break from someone else. See Less of See on Facebook only if you visit his profile. See Options Limit What Will See Update your privacy settings so he won't see your posts. See Options Edit Who Can See Past Posts Change who can see past posts of you and See Options Want to make more changes? Visit the Help Center to see more options. DONE Take a Break tool by the Protect and Care team. Confidential X Page 20 of 29 Chats Haugen_00016911 Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 21 of 29 3316275/?hc_location=uf We're here to help if you need a break Here are some changes we think might be helpful so you can take a break and stay friends on Facebook. Don't worry, we won't notify about changes you make. Take a break from someone else. See Less of See on Facebook only if you visit his profile. X See Options Limit What Will See Update your privacy settings so he won't see your posts. See Options Edit Who Can See Past Posts Change who can see past posts of you and See Options Want to make more changes? Visit the Help Center to see more options. DONE Confidential Haugen_00016912 Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 W W 1766984813316275/?hc_location=ufi W ③ Page 22 of 29 B W W Q Something like this tool could be made in conjunction with the Beach Mode option we use internally. This could range from allowing users to choose periods where they receive no notifications to choosing what types of notifications they receive (ex. choosing not to be notified when someone likes their post). This could help “twitchy" users who may keep entering the app because of the push notifications they receive. In addition, we could also change some of our strategies for increasing app usage. One main example of this is badging. In the screenshot below, I captured just some of the badging used in the current Facebook family of apps as well as some intended badging strategies. ... and probably change some of our growth Haugen_00016913 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 23 of 29 -people-addicted-to-facebook/1766984813316275/?hc_location=ufi ... QQ X and probably change some of our growth "hacks" "What shared responsibilities do we so-called growth hackers ... have to our users, to our future generations, and to ourselves?" Bonfire Redacted-SCA Redacted SCA Redacted - SCA Redacted SCA Instagram Redacted SCA TECH TALK: Learning to Write Code, A Most Recent T-A A Calls we can be very aggressive with our notifications to create a habit Chat 2 and Good morning, I hope you Haugen_00016914 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 24 of 29 Most Recent Screenshots of badging added to different parts of the Facebook family of apps. Now this is not to say that we should never notify users. But, if people are addicted to our products, we do have an obligation to help them. Rather than inundating users with excessive notifications, we should focus on creating experiences that users to willingly want to come back. As we work towards creating meaningful connections on Facebook, we should focus more on quality time spent than just time spent. Where do we go from here? Which quadrant are we targeting? ☆ Haugen_00016915 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 25 of 29 Where do we go from here? Which quadrant are we targeting? Manipulation Matrix Where we want to be Materially improves the user's life Peddler Does not improve the user's life Dealer Facilitator Entertainer The maker does not use it The maker uses it NirAndFar.com The manipulation matrix (from Nir's blog). Haugen_00016916 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 26 of 29 Which quadrant are we targeting? Manipulation Matrix Where we want to be Materially improves the user's life Peddler Does not improve the user's life Dealer Facilitator Entertainer The maker does not use it The maker uses it NirAndFar.com The manipulation matrix (from Nir's blog). Haugen_00016917 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 27 of 29 The maker does not use it The maker uses it NirAndFar.com The manipulation matrix (from Nir's blog). I'll end this note with the Manipulation Matrix that Nir discusses in his book. The matrix has four sections: each defining the type of product you can create based on whether we as Facebook employees would actually use it and whether it helps people's lives. As Mark said when he announced the company's new mission statement, "we have a unique opportunity and platform to bring people closer together, and now more than ever, we each have a responsibility to make the most of our opportunity to do this.” As we work to build community on Facebook, it is important that as we create new products and tools we continue to build things that we both use and know help people's lives. While boosting metrics with products that are not helping people's lives can work in the short term, in the long term we need to focus on making products that align with our company's mission and make sure people feel in control of their time spent on Facebook. 2 Shares Haugen_00016918 Confidential Case 4:22-md-03047-YGR Document 2885-57 Filed 03/26/26 Page 28 of 29 52 Share Link to Powerpoint deck: https://www.facebook.com/media/set/... 1 3y This is awesome! 1 3y Save 9 Comments 2 Shares great job i'm really glad this was your project this summer and it was great working with you! for posterity, i'll add here that we had some good conversations as to whether we could actually label these people "addicted" which is a pretty specific term. for instance, the survey items that you used were the most tame ones from an existing scale so may not really be picking up on "addicting" but likely picking up on patterns of use that we wouldn't want to be encouraging. I think all of your research and analyses suggest we should continue to dig deeper here, so thanks for all of your hard work! cc: Moira, 3y Moira Burke Agree that this is really interesting and something that a lot of people here care about! Great work, point is good, as well, that "addicted" has clinical implications (including that the addiction interferes with other aspects of your life), but that in general, we care about providing people with meaningful, high-quality social connections and not just a junk food time-sink. 6 3y 3y It was great working with you as well We should definitely look into other ways to define these users. While I think addicted is the easiest word, I think we could have a clearer word for these users because our data confines to how they act on Facebook and we don't know if they're suffering from other non-online addictions. One good adjective I've been thinking about is "twitchy" since it seems these users just feel jumpy and keep coming into the app. 3y This is awesome, nice job To your point around allowing users to choose periods where they receive no notifications/choosing what types or notifications they may want to receive, I saw that Google announced "Inbox Pause" today which allows people to adjust their settings so emails come in at a designated schedule and cadence. http://www.boomeranggmail.com/inboxpause/?ref=producthunt. Is this similar to what you had in mind for notifications? pit back against email overto MacBook Pro Chats 5 Haugen_00016919 Confidential W Case 4:22-md-03047-YGR bok/1766984813316275/?hc_location=ufi Document 2885-50 ed Q3/26/26 Page 29 of 29 ☆ to be encouraging. I think all of your research and analyses suggest we should continue to dig deeper here, so thanks for all of your hard work! cc: Moira, 3y Moira Burke Agree that this is really interesting and something that a lot of people here care about! Great work, s point is good, as well, that "addicted" has clinical implications (including that the addiction interferes with other aspects of your life), but that in general, we care about providing people with meaningful , high-quality social connections and not just a junk food time-sink. 3y 3y It was great working with you as well We should definitely look into other ways to define these users. While I think addicted is the easiest word, I think we could have a clearer word for these users because our data confines to how they act on Facebook and we don't know if they're suffering from other non-online addictions. One good adjective I've been thinking about is "twitchy" since it seems these users just feel jumpy and keep coming into the app. 3y This is awesome, nice job To your point around allowing users to choose periods where they Teceive no nomcations/choosing what types of notifications they may want to receive, I saw that Google announced "Inbox Pause" today which allows people to adjust their settings so emails come in at a designated schedule and cadence. http://www.boomeranggmail.com/inboxpause/?ref=producthunt. Is this similar to what you had in mind for notifications? ght back against email overlo BOOMERANGGMAIL.COM INBOX PAUSE Inbox Pause: Eliminate Interruptions, Stop Stress, and Feel Free Boomeran 3y yeah, this is a cool idea! This is by Baydin (but their main products run as gmail extensions). I think This is a good start followed by general ways to filter out specific notification types. For example, choosing to not get notified if someone liked your status (since you can't really do anything because of it). 3y Edited 3y Makes sense, that's an interesting idea! MacBook Pro Chats 6 5 1 1 Haugen_00016920 Confidential