Meta & Children's Online Safety"Addiction"/Problematic Use Design Audit Plan
Raw text
Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 1 of 20 AMENDED Exhibit 211 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 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 2 of 20 "Addiction"/Problematic Use Design Audit Plan [WIP] POCS: Elena Davis & Jennifer Guadagno [+ depending on degree to which we want to combine with dark patterns work] Design Partners: TBD (ideally someone from Foundations?) Outstanding questions * ॐ Z1 team, possibly someone from FB Design System on Product Is "addiction"/problematic use audit the same as dark-pattern work? Should they be combined or kept separate, with addiction/problematic use audit informing work happening on dark patterns and vice versa? Elena Jennifer to figure out How closely to work with Margaret's team on this? Where are they at, are goals aligned? Include Instagram as well? Who should we partner with externally? Tristan or others? (see pros/cons below) Context There is growing concem in the public that technology companies are deliberately manipulating users by designing addictive products that can harm well-being and subvert self-control. The prevalence of problematic (general) internet use ranges between 1.5-8.2% in the US and Europe (link), and academics are debating whether 1. FB is truly "addictive", and 2. what standardized definitions we should use to identify FB addiction or problematic FB use. Regardless of the clinical definitions here - given that currently there is NO medically defined FB addiction disorder - what are the real people problems in this space? The well-being team has reframed the addiction narrative to focus on the ways that FB may be contributing to use patterns that people find difficult to control despite negative impacts in their lives, and to identify and correct those contributing factors. Specifically, what are the ways that FB can seem rewarding or reinforcing without delivering real value, and what are the types of habitual FB behaviors that people find unwanted and difficult to change? Background for the audit * * * Research on reward leaming and habit formation processes point to certain behavior patterns that promote repeated, automatic behaviors that may begin to operate outside of an individual's control. We hypothesize that unpredictable, low value rewards and cues that promote frequent, automatic, undesired behaviors are contributing to reward leaming and habit formation in a way that is not aligned with users' intentions or preferences. These may lead to feelings of being manipulated, feeling lack of control related to certain behaviors, and feelings of dependence on checking or going on Facebook that could be related to lower well-being - and could be fueling the subjective, colloquial experiences of feeling "addicted" to Facebook. See more about "addiction" narrative here In keeping with the engagement model of the well-being team, we have defined guidelines for how teams throughout the company can be incorporating well-being when designing and building products. Some of these guidelines are particularly relevant for identifying and reconsidering aspects of the platform that are currently promoting "addictive"-like behaviors. (See guidelines in Related work section below). Research Plan ALIGN WITH INTERNAL TEAMS Work with Margaret Stewart and other design ethics leaders to understand what design changes would be in line with the goals and team to understand how this work fits of reducing low value rewards and improving user control. Work with into and can benefit dark pattern work. See if this should be expanded to Instagram as well. CONFIDENTIAL META3047MDL-044-00108564 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 3 of 20 CREATE CLEAR DEFINITIONS/CRITERIA Partner with design to refine a definition of what aspects of features promote "addictive"-like/problematic behaviors to target in the audit and how they're different from/related to dark patterns. Specifically, features are problematic if they promote: • Unpredictable, low-value rewards (e.g., certain notifications) Low-control or automatic behaviors (e.g., autoplay) • Frequent behaviors that are pressured or unwanted (e.g., read receipts) Define other criteria DESIGN AUDIT Perform a design audit of the platform with key external behavioral experts and with internal designers and researchers, to identify features that may lead to habitual and unwanted/low-value behaviors How we'll work with extemal expert: Share and iterate on our definition of characteristics of features that promote "addictive"-like/problematic behaviors. Establish how that's different from/related to dark patterns and specify whether we're after just former or both (TBD from our internal discussions). • Engage with a third party to conduct an independent audit and/or review of the features we have identified Form an alliance where the third party can vouch for the thoroughness and relevance of our approach for targeting the "addiction" daims Possible external experts to partner with: • Tristan Harris (and others in his camp: eg, Justin Rosenstein) 8 ॐ IDEO • Lots of pros and cons here - all included at very end of doc, but summary is biggest risk is that he points out things that we don't agree with or are unfounded and we don't make changes he's suggesting MIT Media Lab Academics specializing in addiction and technology (eg, UK researchers: Darna Kuss and Mark Griffiths - informational meeting set up in mid-September). Others? • Joe-Edelman value-based design (pert of Tristan's-camp) EXPLORE/TEST ISSUES RAISED IN AUDIT Work with relevant product teams to follow up audit with user research, tests, lit reviews/gathering scientific support on certain issues raised in audit. • Eg., test potential drivers of habitual, repeated behaviors by creating variants on the platform and examining associated behavior changes (if badge color being red is raised as an issue test effects of changing notification badge color) GOALS CONFIDENTIAL * Prioritized, validated list of concerning product experiences that should be fixed or improved • Take prioritized list into a design sprint to mock up fixes or alternative features to address main problems Consider expanding or adding to well-being principles to incorporate new "anti-addiction"/ problematic use principles * • E.g., "Behaviors that are frequent should also be valuable." Inform and help move along work on dark patterns team and Margaret Stewart's team) META3047MDL-044-00108565 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 4 of 20 Related work within the company • Well-being engagement model and design principles (link, selected principles copied below): Jennifer Guadagno • Give people the most value in the time they have & Do: recognize that people have different intentions when coming to Facebook Don't: design to increase growth metrics without also benefiting the individual • Provide the transparency needed for people to make decisions with intentionality and clarity Do: make it clear what will happen as a result of their actions and Do: provide defaults that are good for people's well-being, while empowering them to control their Facebook experience Don't: impose behaviors on people that are against their best interests • Create an environment that has a positive impact on people, and that is easy to participate in Don't: create features designed for good that can come off as self-serving 8 Dark Pattern spotting (link): • Margaret's Ethical Design work Additional Thoughts & Info Q: How is an "addictive"-like design element similar vs. different from a "dark pattern"? A: "Dark pattern" is a UX term referring to interactions that are deceptive, or that trick you into doing something you didn't want to do. These include misleading or manipulative features or settings that are designed to encourage a particular behavior that is not necessarily in the user's best interests. These can include single-instance behaviors. "Addictive"-like design elements are those that promote repeated or poorly regulated behaviors that may have a negative impact on the individual over time as they continue behaviors that aren't in their best interests. A dark pattem deviates from user expectations; an "addictive"-like design results in a behavior pattern that is not wanted or valuable for the user over time (the function of the design is to promote repeated maladaptive or low-control behaviors). Examples of cases that may be both "dark patterns" and "addictive"-like: All Roads Lead to Watch (since it seems like it could promote automatic or unintentional behaviors when coupled with autoplay), and low value or "fake" notifications like alerts that you are connected on Messenger or have a particular setting turned off (promoting frequent checking behaviors and perceptions of reward without delivering gratification). CONFIDENTIAL Dark pattern "Addictive"-like design Autoplay Default settings or design nudges towards liberal privacy settings Endless scroll Low-value notifs e.g., "You are now connected on Messenger Frequent, unpredictable but "wanted" notifs e.g., like count updates META3047MDL-044-00108566 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 5 of 20 Previous Notes: Internal at facebook * Work with Margaret Stewart and other design ethics leaders to understand what design changes would be in line with the goals of reducing low value rewards and improving user control ॐ $ * Refine a definition of what aspects of features promote "addictive"-like behaviors to target in the audit, specifically: • Which features constitute unpredictable, low-value rewards e.g., certain notifications • Which features or settings promote low-control or automatic behaviors * e.g., autoplay • Which features encourage frequent behaviors that are pressured or unwanted e.g., read receipts Perform a design audit of the platform with key external behavioral experts, with internal designers and researchers, to identify features that may lead to habitual and unwanted/low-value behaviors Use our refined definition to identify and understand the function of those features in creating "addictive"-like behaviors (ie., a habit loop that is having negative impacts on the person) Follow up with audit with user research on certain features to verify the hypothesized functions (?) oe.g., test potential drivers of habitual, repeated behaviors by creating variants on the platform and examining associated behavior changes (like changing notification badge color) Iterate on the audit following qualitative user research in September-October and quantitative research in (? October) if new aspects of the platform are identified as problematic Consider expanding or adding to well-being principles to incorporate new "anti-addiction" principles • E.g., "Behaviors that are frequent should also be valuable." with an external third party Share and iterate on our definition of characteristics of features that promote "addictive"-like behaviors • Engage with a third party to conduct an independent audit and/or review of the features we have identified ॐ Form an alliance where the third party can vouch for the thoroughness and relevance of our approach for targeting the "addiction" daims Partnering with Tristan Harris on this Pros & Cons • Pros: CONFIDENTIAL & Hehas lots of thoughts in this space - he said it's the area he's best positioned to inform, it'd be helpful to leam from him. • Gives him a chance to be heard • Helpful to deeply partner with him on area he's most informed about to further our relationship 38 He's likely connected to other experts in this area we could leverage and leam from • Possible concems: NDA? • What if we don't agree on much or his claims aren't substantiated by other further research (eg, color red on badges) • We, from the research side, do agree with his advice but product teams are slow or unwilling to change things 粥 He has even more inside info about what's bad with FB • Ideally working with an external expert could on this could make them see we're taking this seriously and they become vocal advocates of ours we can point people to. Likely won't happen with him META3047MDL-044-00108567 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 • Things we can do to help mitigate concerns: Page 6 of 20 • Setting expectations: Be very explicit about us wanting to carefully vet/further research anything suggested so he knows we won't necessarily take whatever he says at face value and run with it. • Identify at the outset that getting product teams to make big changes may be tricky, use this as a challenge for him - how can he help us convince product teams? • Work with additional design firms and experts in addition to him. Have him co-create an expert panel with us (we can both nominate and veto people). This ensures more balanced thinking, other people that aren't us to possibly disagree with him, and builds relationships with other allies to advocate for us. • Overall this is really risky - high risk but high reward. Are the potential gains to our relationship with him and broader reputation worth the potential harm this can do/are we already in a bad enough spot with him that possible gains outweigh harms? Contact Support Conversation History Quipbot#08 left the thread Oct 12 at 10:25 am Quipbot#08 Oct 12 at 10:20 am =====IMPORTANT===== This Quip file has been copied to Google Workspace (https://docs.google.com/document...) as part of the Quip2Google migration efforts. Quip is moving to Read-Only. Please work in the new Google file moving forward. Learn more https://fb.workplace.com/groups... Questions? Visit the Quip2Google Migration Feedback Group (https://fb.workplace.com/groups...). Note New edits in this Quip file will NOT be automatically synced to Google Workspace. Quipbot#08 added Quipbot#08 to the thread Oct 12 at 10:25 am Quipbot#08 added you to a document Oct 12 at 10:20 am Jonny Thaw opened your conversation with Elena Davis, at 6:47 pm CONFIDENTIAL added Jonny Thaw to the thread Aug 20, 2018 at 2:18 pm added you to a document Aug 20, 2018 at 2:12 pm and Jennifer Guadagno Aug 20, 2018 META3047MDL-044-00108568 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 7 of 20 CONFIDENTIAL META3047MDL-044-00108569 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 8 of 20 CONFIDENTIAL META3047MDL-044-00108570 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 9 of 20 CONFIDENTIAL META3047MDL-044-00108571 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 10 of 20 CONFIDENTIAL META3047MDL-044-00108572 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 11 of 20 CONFIDENTIAL META3047MDL-044-00108573 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 12 of 20 CONFIDENTIAL META3047MDL-044-00108574 Case 4:22-md-03047-YGR Elena Davis Aug 8, 2018 at 10:33 pm Joe Edelman here? Document 2757-16 Filed 02/20/26 Page 13 of 20 Elena Davis edited Aug 8, 2018 at 10:48 pm There is growing concem in the public that technology companies are deliberately manipulating users by designing addictive products that can harm well-being and subvert self-control... seem rewarding or reinforcing without delivering real value, and what are the types of habitual FB behaviors that people find unwanted and difficult to change. ? Perform a design audit of the platform with key external behavioral experts, with internal designers and researchers, to identify features that may lead to habitual and unwanted/low-value behaviors Possible extemal experts to partner with: * • We, from the research side, do agree with his advice but product teams are slow or unwilling to change things • Ideally working with an external expert could on this could make them see we're taking this seriously and they become vocal advocates of ours we can point people to. Likely won't happen with him • Overall this is really risky - high risk but big rewards. Are the potential gains to our relationship with him worth the potential harm this can do/are we already in a bad enough spot with him that possible gains outweigh harms? Prioritized, validated list of concerning product experiences that should be fixed or improved • Well-being engagement model and design principles (link), selected principles copied below): Jennifer Guadagno and Jennifer Guadagno edited Aug 8, 2018 at 10:11 pm "Addictive"-Like "Addiction"/Problematic Use Design Audit Plan [WIP] Leads: Elena Davis, Jennifer Guadagno, Design Partners: TBD (possibly someone from Foundations) [WIP] Outstanding questions ॐ Z1 team, possibly someone from FB Design System on Product Is "addiction"/problematic use audit the same as dark-pattern work? Should they be combined or kept separate, with addiction/problematic use audit informing work happening on dark patterns and vice versa? - Elena Jennifer to figure out How closely to work with Margaret's team on this? Where are they at, are goals aligned? CONFIDENTIAL Research on reward leaming and habit formation processes point to certain behavior patterns that promote repeated, automatic behaviors that may begin to operate outside of an individual's control. We hypothesize that unpredictable, low value rewards and cues that promote frequent, automatic, undesired behaviors are contributing to reward learning and habit formation in a way that is not aligned with users' intentions or preferences. META3047MDL-044-00108575 Case 4:22-md-03047-YGR * Document 2757-16 Filed 02/20/26 Page 14 of 20 These may lead to feelings of being manipulated, feeling lack of control related to certain behaviors, and feelings of dependence on checking or going on Facebook that could be related to lower well-being - and could be fueling the subjective, colloquial experiences of feeling "addicted" to Facebook. In keeping with the engagement model of the well-being team, we have defined guidelines for how teams throughout the company can be incorporating well-being when designing and building products. Some of these guidelines are particularly relevant for identifying and reconsidering aspects of the platform that are currently promoting "addictive"-like behaviors. (See guidelines in Related work section below). relevant Well-being design principles In keeping with the engagement model of the well-being team, we have defined guidelines for how teams throughout the company can be incorporating well-being when designing and building products. Some of these guidelines are particularly relevant for identifying and reconsidering aspects of the platform that are currently promoting "addictive"-like behaviors. * Give people the most value in the time they have • Do: recognize that people have different intentions when coming to Facebook • Don't: design to increase growth metrics without also benefiting the individual Provide the transparency needed for people to make decisions with intentionality and clarity • Do: make it clear what will happen as a result of their actions • Do: provide defaults that are good for people's well-being, while empowering them to control their Facebook experience • Don't: impose behaviors on people that are against their best interests Create an environment that has a positive impact on people, and that is easy to participate in Research Plan and Goals ALIGN WITH INTERNAL TEAMS Work with Margaret Stewart and other design ethics leaders to understand what design changes would be in line with the goals of reducing low value rewards and improving user control. Work with and team to understand how this work fits into and can benefit dark pattern work. CREATE CLEAR DEFINITIONS Refine a definition of what aspects of features promote "addictive"-like/problematic behaviors to target in the audit and how they're different from/related to dark patterns. Specifically: * Which features constitute unpredictable, low-value rewards (e.g., certain notifications) Which features or settings promote low-control or automatic behaviors (e.g., autoplay) Which features encourage frequent behaviors that are pressured or unwanted (e.g., read receipts) DESIGN AUDIT Perform a design audit of the platform with key external behavioral experts, with internal designers and researchers, to identify features that may lead to habitual and unwanted/low-value behaviors How we'll work with external expert: Share and iterate on our definition of characteristics of features that promote "addictive"-like/problematic behaviors. Establish how that's different from/related to dark patterns and specify whether we're after just former or both (tbd on our internal discussions). • Engage with a third party to conduct an independent audit and/or review of the features we have identified • Form an alliance where the third party can vouch for the thoroughness and relevance of our approach for targeting the "addiction" daims Possible external experts to partner with: Tristan Harris • Pros: Hehas lots of thoughts in this space - he said it's the area he's best positioned to inform, it'd be helpful to learn from him. Gives him a chance to be heard CONFIDENTIAL META3047MDL-044-00108576 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 ॐ IDEO Page 15 of 20 • Helpful to deeply partner with him on area he's most informed about to further our relationship He's likely connected to other experts in this area we could leverage and learn from • Possible concems: NDA? • What if we don't agree on much or his claims aren't substantiated by other further research (eg, color red on badges) • We, from the research side, do agree with his advice but product teams are slow or unwilling to change things • He has even further inside info about what's bad with FB • Ideally working with an extemal expert could on this could make them see where taking this seriously and they become vocal advocates of ours we can point people to. Likely won't happen with him • Things we can do to help mitigate concems: • Setting expectations: Be very explicit about us wanting to carefully vet/further research anything suggested so he knows we won't necessarily take whatever he says at face value and run with it. • Identify at the outset that getting product teams to make big changes may be tricky, use this as a challenge for him - how can he help us convince product teams? • Work with additional design firms and experts in addition to him. Have him co-create an expert panel with us (we can both nominate and veto people). This ensures more balanced thinking, other people that aren't us to possibly disagree with him, and builds relationships with other allies to advocate for us. • Overall this is really risky - high risk but big rewards. Are the potential gains to our relationship with him worth the potential harm this can do/are we already in a bad enough spot with him that possible gains outweigh harms? Academics specializing in addiction and technology (eg, UK researchers: Daria Kuss and Mark Griffiths). Others? EXPLORE/TEST ISSUES RAISED IN AUDIT Work with relevant product teams to follow up audit with user research, tests, lit reviews/gathering scientific support on certain issues raised in audit. • Eg., test potential drivers of habitual, repeated behaviors by creating variants on the platform and examining associated behavior changes (if badge color being red is raised as an issue test effects of changing notification badge color) GOALS • Prioritized, validated list of concerning product experiences that should be fixed or improved Consider expanding or adding to well-being principles to incorporate new "anti-addiction"/ problematic use principles • E.g., "Behaviors that are frequent should also be valuable." Inform and help move along work on dark patterns team and Margaret Stewart's team) CONFIDENTIAL • Well-being engagement model and design principles (link): • Give people the most value in the time they have and Jennifer Guadagno Do: recognize that people have different intentions when coming to Facebook Don't: design to increase growth metrics without also benefiting the individual META3047MDL-044-00108577 CONFIDENTIAL 200000000000000 000000000000000 0000000000000000000000 0000000 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 ॐ Page 16 of 20 • Provide the transparency needed for people to make decisions with intentionality and clarity Do: make it clear what will happen as a result of their actions • Do: provide defaults that are good for people's well-being, while empowering them to control their Facebook experience Don't: impose behaviors on people that are against their best interests • Create an environment that has a positive impact on people, and that is easy to participate in Don't create features designed for good that can come off as self-serving Dark Pattern spotting (link): • Margaret's Ethical Design work Additional Thoughts & Info Q: How is an "addictive"-like design element similar vs. different from a "dark pattern"? A: "Dark pattern" is a UX term referring to interactions that are deceptive, or that trick you into doing something you didn't want to do. These include misleading or manipulative features or settings that are designed to encourage a particular behavior that is not necessarily in the user's best interests. These can include single-instance behaviors. "Addictive"-like design elements are those that promote repeated or poorly regulated behaviors that may have a negative impact on the individual over time as they continue behaviors that aren't in their best interests. A dark pattern deviates from user expectations; an "addictive"-like design results in a behavior pattern that is not wanted or valuable for the user over time (the function of the design is to promote repeated maladaptive or low-control behaviors). Q: How is an "addictive"-like design element similar vs. different from a "dark pattern"? A: "Dark pattern" is a UX term referring to interactions that are deceptive, or that trick you into doing something you didn't want to do. These include misleading or manipulative features or settings that are designed to encourage a particular behavior that is not necessarily in the user's best interests. These can include single-instance behaviors. "Addictive"-like design elements are those that promote repeated or poorly regulated behaviors that may have a negative impact on the individual over time as they continue behaviors that aren't in their best interests. A dark pattem deviates from user expectations; an "addictive"-like design results in a behavior pattern that is not wanted or valuable for the user over time (the function of the design is to promote repeated maladaptive or low-control behaviors). Previous Notes: * Work with Margaret Stewart and other design ethics leaders to understand what design changes would be in line with the goals of reducing low value... • Which features encourage frequent behaviors that are pressured or unwanted e.g., read receipts Perform a design audit of the platform with key extemal behavioral experts, with internal designers ⚫ accompanied by researcher input ॐ and researchers, to identify features that may lead to habitual and unwanted/low-value behaviors * Use our refined definition to identify and understand the function... META3047MDL-044-00108578 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 • adding to well-being principles to incorporate new "anti-addiction" principles • E.g., "Behaviors that are frequent should also be valuable." Page 17 of 20 Jennifer Guadagno renamed "Addictive"-Like Design Audit Plan to "Addiction"/Problematic Use Design Audit Plan [WIP] Aug 8, 2018 at 9:01 pm Jennifer Guadagno opened your conversation with Elena Davis, at 7:29 pm Elena Davis enabled link sharing View/Edit Aug 8, 2018 at 6:11 pm Elena Davis added Jennifer Guadagno to the thread Aug 8, 2018 at 6:11 pm Elena Davis added you to a document Aug 8, 2018 at 6:04 pm and Jonny Thaw Aug 8, 2018 Elena Davis edited Aug 8, 2018 at 6:06 pm "Addiction" Feature "Addictive"-Like Design Audit Plan [WIP] Context : There is growing concern in the public that technology companies are deliberately manipulating users by designing addictive products that can harm well-being and subvert self-control. The prevalence of problematic (general) internet use ranges between 1.5-8.2% in the US and Europe (link), and academics are debating whether 1. FB is truly "addictive", and 2. what standardized definitions we should use to identify FB addiction or problematic FB use. Regardless of the clinical definitions here - given that currently there is NO medically defined FB addiction disorder - what are the real people problems in this space? The well-being team has reframed the addiction narrative to focus on the ways that FB may be contributing to use patterns that people find difficult to control despite negative impacts in their lives, and to identify and correct those contributing factors. Specifically, what are the ways that FB can seem rewarding or reinforcing without delivering real value, and what are the types of habitual FB behaviors that people find unwanted and difficult to change. Background for the audit Research on reward leaming and habit formation processes point to certain behavior pattems that promote repeated, automatic behaviors that may begin to operate outside... feeling lack of control related to certain behaviors, and feelings of dependence on checking or going on Facebook that could be related to lower well-being and could be fueling the subjective, colloquial experiences of feeling "addicted" to Facebook. relevant Well-being design principles In keeping with the engagement model of the well-being team, we have defined guidelines for how teams throughout the company can be incorporating well-being when designing and building products. Some of these guidelines are particularly relevant for identifying and reconsidering aspects of the platform that are currently promoting "addictive"-like behaviors. ॐ Give people the most value in the time they have • Do: recognize that people have different intentions when coming to Facebook • Don't: design to increase growth metrics without also benefiting the individual CONFIDENTIAL META3047MDL-044-00108579 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 * * Provide the transparency needed for people to make decisions with intentionality and clarity • Do: make it clear what will happen as a result of their actions Page 18 of 20 • Do: provide defaults that are good for people's well-being, while empowering them to control their Facebook experience • Don't: impose behaviors on people that are against their best interests Create an environment that has a positive impact on people, and that is easy to participate in • Don't create features designed for good that can come off as self-serving Research Plan and Goals Internal at facebook $ * ॐ Work with Margaret Stewart and other design ethics leaders to understand what design changes would be in line with the goals of reducing low value rewards and improving user control Refine a definition of what aspects of features promote "addictive"-like behaviors to target in the audit, specifically: • Which features constitute unpredictable, low-value rewards ⚫e.g., certain notifications • Which features or settings promote low-control or automatic behaviors e.g., autoplay • Which features encourage frequent behaviors that are pressured or unwanted ⚫e.g., read receipts Perform a design audit of the platform, with designers accompanied by researcher input, to identify features that may lead to habitual and unwanted/low-value behaviors • Use our refined definition to identify and understand the function of those features in creating "addictive"-like behaviors (i.e., a habit loop that is having negative impacts on the person) Follow up with audit with user research on certain features to verify the hypothesized functions (?) oe.g., test potential drivers of habitual, repeated behaviors by creating variants on the platform and examining associated behavior changes (like changing notification badge color) Iterate on the audit following qualitative user research in September-October and quantitative research in (? October) if new aspects of the platform are identified as problematic Consider expanding or adding to well-being principles to incorporate new "anti-addiction" principles • E.g., "Behaviors that are frequent should also be valuable." with an external third party Share and iterate on our definition of characteristics of features that promote "addictive"-like behaviors • Engage with a third party to conduct an independent audit and/or review of the features we have identified * Form an alliance where the third party can vouch for the thoroughness and relevance of our approach for targeting the "addiction" daims Related work within the company Well-being engagement model and design principles (link) • (?With a third party?) Review features and default settings on the platform to identify unnecessary examples of frequent low-value reward signals, automatic unwanted behaviors that contribute to low perceptions of control, and problematic domains that come from user interviews and/or the problematic use quantitative analysis What Facebook features are implicated in: Frequent, pressured or unwanted behaviors (e.g., read receipts) Low-control or automatic behaviors (e.g., autoplay) Unpredictable, low-value rewards (e.g., certain notifications) CONFIDENTIAL META3047MDL-044-00108580 Document 2757-16 Filed 02/20/26 Page 19 of 20 Case 4:22-md-03047-YGR Principles: Things that are frequent should be wanted and valuable 0000000 000000000000000 Q: How is an "addictive"-like design element similar vs. different from a "dark pattern"? A: "Dark pattern" is a UX term referring to interactions that are deceptive, or that trick you into doing something you didn't want to do. These include misleading or manipulative features or settings that are designed to encourage a particular behavior that is not necessarily in the user's best interests. These can include single-instance behaviors. "Addictive"-like design elements are those that promote repeated or poorly regulated behaviors that may have a negative impact on the individual over time as they continue behaviors that aren't in their best interests. A dark pattem deviates from user expectations; an "addictive"-like design results in a behavior pattern that is not wanted or valuable for the user over time (the function of the design is to promote repeated maladaptive or low-control behaviors). Examples of cases that may be both "dark patterns" and "addictive"-like: All Roads Lead to Watch (since it seems like it could promote automatic or unintentional behaviors when coupled with autoplay), and low value or "fake" notifications like alerts that you are connected on Messenger or have a particular setting turned off (promoting frequent checking behaviors and perceptions of reward without delivering gratification). Elena Davis edited Aug 7, 2018 at 11:39 pm What Facebook features are implicated in: Frequent, pressured or unwanted behaviors (e.g., read receipts) Low-control or automatic behaviors (e.g., autoplay) Unpredictable, low-value rewards (e.g., certain notifications) Principles: Things that are frequent should be wanted and valuable Elena Davis edited Aug 2, 2018 at 6:31 pm "Addiction" Feature Audit Context: Research on reward learning and habit formation processes point to certain behavior patterns that promote repeated, automatic behaviors that may begin to operate outside of an individual's control. We hypothesize that unpredictable, low value rewards and cues that promote frequent, automatic, undesired behaviors are contributing to reward learning and habit formation in a way that is not aligned with users' intentions or preferences. These may lead to feelings of being manipulated, feeling lack of control related to certain behaviors, and feelings of dependence on checking or going on Facebook that could be related to lower well-being. • (?With a third party?) Review features and default settings on the platform to identify unnecessary examples of frequent low-value reward signals, automatic unwanted behaviors that contribute to low perceptions of control, and problematic domains that come from user interviews and/or the problematic use quantitative analysis Work with Margaret Stewart and other design ethics leaders to understand what design changes would be in line with the goals of reducing low value rewards and improving user control Elena Davis created the document Aug 2, 2018 at 6:32 pm CONFIDENTIAL META3047MDL-044-00108581 Case 4:22-md-03047-YGR Document 2757-16 Filed 02/20/26 Page 20 of 20 CONFIDENTIAL META3047MDL-044-00108582