Chapter 6 Exercises: Who Builds These Systems?

These exercises are organized into five tiers of increasing cognitive complexity, following Bloom's Taxonomy. Complete lower tiers before attempting higher ones.

Tier 1: Recall

These questions test your memory of key facts and concepts from the chapter.

Exercise 1.1 What was Frances Haugen's role at Facebook before she became a whistleblower, and approximately when did she provide documents to Congress and the press?

Exercise 1.2 What does the acronym "OKR" stand for, and which company is credited with originating the framework?

Exercise 1.3 Name three specific metrics that were tracked in Velocity Media's product review meetings as described in the chapter's Dr. Aisha Johnson narrative.

Exercise 1.4 What was the title of Sophie Zhang's internal Facebook memo that was published by BuzzFeed News, and what general category of platform problem did it address?

Exercise 1.5 What was Timnit Gebru's position at Google, and what was the name of the paper that precipitated her dismissal in December 2020?

Tier 2: Comprehension

These questions test your understanding of how the mechanisms and concepts described in the chapter work.

Exercise 2.1 Explain in your own words why the demographic homogeneity of the tech workforce is described as a "design concern" rather than merely a fairness concern. What specific causal mechanism connects workforce demographics to product design outcomes?

Exercise 2.2 The chapter describes A/B testing as a "moral distance-creator." What does this mean? Walk through the specific steps by which A/B testing shifts the appearance of moral agency from human beings to data.

Exercise 2.3 Distinguish between "ethics theater" and genuine organizational ethics commitment. What specific organizational features — reporting structure, authority, budget, independence — determine which category a given ethics hire falls into?

Exercise 2.4 What is "diffusion of responsibility" as described by Albert Bandura, and why does it operate with particular force in large platform companies? Give a specific example from the chapter.

Exercise 2.5 Explain the temporal mismatch problem with OKR culture and social media harm. Why does a quarterly measurement framework systematically underweight long-term harms even without anyone consciously deciding to do so?

Exercise 2.6 What did Sophie Zhang mean when she wrote "I have blood on my hands" in her internal memo? What made her account different from a straightforward external critique of Facebook, and why did that difference matter for its credibility?

Exercise 2.7 Describe the "advantageous comparison" mechanism of moral disengagement. Why does the chapter argue this form of reasoning, while "not entirely illegitimate," is problematic as a justification for continued participation in harmful systems?

Exercise 2.8 The chapter describes the Google Project Maven case as an example of internal advocacy producing policy change. What specific conditions enabled it to succeed, and why does the chapter suggest these conditions are not typical?

Tier 3: Application

These questions ask you to apply chapter concepts to new scenarios.

Exercise 3.1 A product manager at a social media company is designing a new feature that allows users to see how their content is performing (views, likes, shares) in real time, updated every 30 seconds. Apply the chapter's analysis of A/B testing as a moral distance-creator to this scenario: what specific choices are embedded in the experimental design of an A/B test for this feature before any "data-driven" decision is made?

Exercise 3.2 A tech company announces it is hiring a "Chief Ethics Officer" who will report directly to the CEO and have a seat on the executive team, but the role has no veto authority over product launches and is not included in the product OKR review process. Using the chapter's framework, evaluate whether this constitutes genuine organizational ethics commitment or ethics theater. What additional information would you want to know?

Exercise 3.3 An engineer at a social media company discovers in a code review that a new notification algorithm sends 40% more notifications than the current system. The A/B test shows a positive impact on 7-day return visits. The engineer is concerned about notification overload but has no data on its effects. What specific actions are available to this engineer, and what are the realistic limitations of each action, based on the chapter's analysis?

Exercise 3.4 Using the concept of "euphemistic labeling" from the chapter, rewrite the following platform-culture descriptions in language that more accurately reflects their behavioral intent, and then analyze how the vocabulary change affects what questions an engineer would naturally ask: - "Engagement optimization" - "Re-engagement campaign" - "Content quality ranking" - "Personalized experience"

Exercise 3.5 The chapter describes Velocity Media's product review process as tracking six specific metrics and no wellbeing metrics. Design an alternative product review process that incorporates user wellbeing while maintaining the empirical rigor valued in platform engineering culture. What metrics would you add? What review steps? What authority mechanisms?

Exercise 3.6 Apply the chapter's analysis of the "talent pipeline" and its effects to a hypothetical social media company whose engineering team was recruited 60% from liberal arts universities, has a median age of 38, and is 52% women and 34% non-white. Predict, based on the chapter's framework, how the design of this company's platform might systematically differ from those built by a typical platform engineering team.

Exercise 3.7 Facebook's internal research showing Instagram's negative effects on teenage girls' body image was not published and did not result in design changes for several years. Using the chapter's moral disengagement mechanisms (diffusion of responsibility, euphemistic labeling, advantageous comparison, metric normalization), explain how researchers, product managers, and executives at Facebook might have each experienced this situation without feeling morally responsible for the outcome.

Exercise 3.8 Dr. Aisha Johnson's question about the animated notification's effect during sleep hours was acknowledged but did not stop the rollout before the product review meeting ended. Based on what the chapter describes about her organizational position, write a brief strategic memo she might send to her VP making the case for pausing the rollout pending time-of-day data. What framing would be most likely to succeed given the organizational culture described?

Tier 4: Analysis

These questions ask you to break down complex situations, evaluate arguments, and identify assumptions.

Exercise 4.1 The chapter argues that both individual responsibility and structural change are necessary — a "both/and" position — and criticizes the "heroic engineer" narrative as "emotionally satisfying but empirically weak." Analyze this argument. What evidence does the chapter provide for the weakness of the heroic engineer narrative? What evidence does it provide for the effectiveness of individual action? Are these claims in tension, and if so, how does the chapter resolve or fail to resolve the tension?

Exercise 4.2 The Facebook Papers, revealed by Frances Haugen in 2021, showed that Facebook had internal research documenting specific harms and had not acted on it. Haugen's disclosure produced significant public attention and some regulatory movement but did not change the fundamental business model. Analyze why the structural impact of the disclosure was limited, drawing on at least three specific mechanisms described in the chapter.

Exercise 4.3 The Timnit Gebru case at Google involved a procedural dispute (the paper review process) as the proximate cause of termination but is widely interpreted as retaliation for research challenging commercial interests. Analyze this interpretation: what evidence supports it? What would need to be true for the Google management account (that it was purely a procedural matter) to be accurate? What does the subsequent behavior of Google (Mitchell's firing, team restructuring) contribute to the analysis?

Exercise 4.4 The chapter describes the hiring filter at tech companies as "systematically underweighting precisely the skills that would be most useful for anticipating harm." Analyze this claim critically. Is it possible to design a hiring process that selects for both technical rigor and ethical reasoning? What tradeoffs would such a process involve, and what would a company need to believe — about what its employees are for — to implement it?

Exercise 4.5 Compare the chapter's account of moral disengagement in platform culture with Philip Zimbardo's account of ordinary people participating in harmful institutional systems. What are the structural similarities? What are the important differences? Does the platform engineering context make moral disengagement more or less likely than in the institutional settings Zimbardo studied?

Exercise 4.6 The chapter notes that the Google Project Maven case succeeded (internal advocacy changed policy) while most cases of internal dissent fail to change policy. Identify and analyze the factors that made Maven unusual. Then evaluate whether the conditions for Maven-style success are systematically more or less likely to be present in the specific domain of social media design decisions, compared to military AI contracting.

Exercise 4.7 The chapter's description of Sophie Zhang's "I Have Blood on My Hands" memo presents her as simultaneously a critic of Facebook's practices and an honest acknowledger of her own complicity in those practices. Analyze the organizational and psychological factors that likely made it difficult for Zhang — and people in similar positions — to act more decisively on her concerns while still employed at Facebook. What does this analysis imply about how we should evaluate individuals who participate in harmful organizational systems?

Tier 5: Synthesis and Evaluation

These questions require you to form original arguments, synthesize across sources and perspectives, and evaluate competing claims.

Exercise 5.1 The chapter ends by saying that closing the gap between intent and effect requires "unglamorous, structural, patient work to change the processes, metrics, and accountability mechanisms" combined with "external pressure." Design a specific, realistic reform program for a social media company like Velocity Media that would constitute this structural work. Your program should address: (a) the OKR system and what gets measured, (b) the product review process and what has blocking authority, (c) the hiring and training pipeline for ethical reasoning, and (d) the reporting structure for ethics professionals. Evaluate the political feasibility of each element.

Exercise 5.2 Evaluate the following argument: "Hiring ethicists at tech companies is inherently counterproductive because it creates the appearance of ethical commitment without the substance, thereby reducing external pressure for regulatory change. It would be better to have no internal ethics function and let external regulation do the work." Use the chapter's evidence, including the Gebru case and the Velocity Media narrative, to construct the strongest case for and against this position. Where do you land, and why?

Exercise 5.3 The chapter's description of Frances Haugen's whistleblowing is largely sympathetic. But whistleblowing involves costs: to the whistleblower (professional and personal risks), to the organization (reputational and operational disruption), and potentially to other employees. Write a nuanced evaluation of Haugen's decision to leak documents rather than pursuing internal channels, taking seriously both the case for her decision and the legitimate concerns about whistleblowing as a form of institutional action.

Exercise 5.4 The chapter argues that demographic homogeneity in the tech workforce contributes to design blind spots. A counterargument holds that good engineers can design empathetically for populations unlike themselves through user research, testing, and systematic feedback mechanisms. Evaluate this counterargument in light of the chapter's evidence. Under what conditions is it plausible? Under what conditions does it fail? What does your analysis imply about whether diversity hiring is a necessary condition, a sufficient condition, or neither for reducing design harms?

Exercise 5.5 The chapter describes the "move fast and break things" ethos as adaptive in Facebook's early phase and harmful at scale. Consider three other industries where a "move fast" ethos has been dominant — biotech, financial services, aviation — and analyze how each managed (or failed to manage) the transition to appropriate caution as scale and stakes increased. What lessons, if any, do these cases offer for platform technology? What are the disanalogies that complicate the comparison?

Exercise 5.6 Dr. Aisha Johnson is drafting a proposal for a formal wellbeing review process at Velocity Media. Write the core section of that proposal, addressed to the CEO, making the business case for giving wellbeing review the same procedural weight as legal and security review. Your proposal must engage directly with the objection that such a review process would slow down product development, increase costs, and reduce Velocity's competitive position relative to platforms that do not have such a process.

Exercise 5.7 The chapter focuses on the gap between individual intent and structural effect within platform companies. But there is another dimension of the problem: the users of these platforms are also not simply victims of structural forces — they make choices, have preferences, and in some cases actively want the engagement experiences that platforms provide. Write an essay (750-1,000 words) that engages seriously with this complexity: How should we think about user agency and structural manipulation simultaneously? What does taking user agency seriously imply about the limits of structural reform? What does taking structural manipulation seriously imply about the limits of individual responsibility?