Chapter 4 Quiz: The Business Model of Engagement
Instructions: Complete all questions. For multiple choice, circle the best answer. For true/false, write T or F and briefly explain your reasoning. For short answer, write 2–4 sentences.
Part A: Multiple Choice (10 questions)
Question 1. What does CPM stand for in the context of digital advertising?
A) Cost per minute B) Cost per mille (thousand impressions) C) Content performance metric D) Consumer perception measurement
Answer
**B) Cost per mille (thousand impressions)** CPM is the price an advertiser pays for one thousand ad impressions. It is the fundamental pricing unit of the attention economy — the cost of reaching one thousand pairs of eyes with a single advertisement.Question 2. In a Real-Time Bidding auction, what typically determines the final price the winning advertiser pays?
A) The highest bid submitted B) A fixed rate negotiated in advance with the platform C) The second-highest bid submitted (Vickrey auction format) D) The average of all bids submitted
Answer
**C) The second-highest bid submitted (Vickrey auction format)** RTB auctions typically use a Vickrey (second-price) auction format, where the winner pays the second-highest bid. This format incentivizes honest bidding — advertisers can bid their true maximum value for an impression without worrying that winning will cost them more than that impression is worth to them.Question 3. Which of the following user signals produces the highest CPM rates in programmatic advertising?
A) A user browsing entertainment news B) A user watching a viral video C) A user actively searching for a specific product or service D) A user scrolling a social media feed
Answer
**C) A user actively searching for a specific product or service** High-intent signals — explicit search queries indicating purchase interest — command the highest CPMs because the probability of conversion is highest. A search for "best personal injury lawyer in Chicago" can command CPMs over $400, while passive content consumption carries CPMs of $3–15.Question 4. Facebook's EdgeRank algorithm, introduced between 2009 and 2011, ranked News Feed content based on which combination of factors?
A) Follower count, post length, and political alignment B) Affinity, weight, and recency C) Advertiser relevance, user demographics, and post timing D) Comment count, share count, and profile verification
Answer
**B) Affinity, weight, and recency** EdgeRank used three variables: Affinity (how closely connected you are to the poster, measured by your past interactions with their content), Weight (the content type, with photos and videos ranked higher than text links), and Recency (how recently the post was made, with older posts decaying in rank).Question 5. According to the chapter, how many users had their News Feeds manipulated in Facebook's emotional contagion study?
A) Approximately 10,000 B) Approximately 68,000 C) Approximately 689,000 D) Approximately 6.89 million
Answer
**C) Approximately 689,000** The emotional contagion study, conducted in January 2012 and published in June 2014, manipulated the News Feeds of approximately 689,003 Facebook users. Half received feeds skewed toward positive emotional content; half received feeds skewed toward negative emotional content.Question 6. Velocity Media's Series A VC term sheet required the company to report monthly on which set of metrics?
A) Monthly recurring revenue, customer acquisition cost, and churn rate B) Content quality score, creator satisfaction, and learning outcomes C) DAU/MAU ratio, average session duration, and Day-7 user retention D) Total video uploads, average watch time per video, and comment rate
Answer
**C) DAU/MAU ratio, average session duration, and Day-7 user retention** These three engagement metrics were specified in the term sheet from David Park's VC firm. Sarah Chen's concern — that none of these metrics measured whether users actually learned something useful — illustrates the incentive trap: the metrics that matter financially are not the metrics that reflect the platform's stated purpose.Question 7. Which of the following best describes the "data flywheel" that creates competitive advantages for established advertising platforms?
A) More data leads to better targeting, which leads to higher CPMs, which funds more data infrastructure, which generates more data B) More users share more data voluntarily, which enables better product features, which attracts more users C) Advertising revenue funds engineering talent, which builds better products, which attracts users who generate revenue D) Network effects create switching costs, which lock in users, who generate advertising revenue
Answer
**A) More data leads to better targeting, which leads to higher CPMs, which funds more data infrastructure, which generates more data** The data flywheel is specifically about the self-reinforcing relationship between behavioral data and advertising precision. More interactions generate richer behavioral profiles; richer profiles enable more precise targeting; more precise targeting commands higher CPMs; higher CPMs fund infrastructure and algorithmic investment that extracts more behavioral signal from each interaction.Question 8. According to the chapter, what is the primary limitation of Substack as an alternative to advertising-supported social media platforms?
A) Its subscription prices are too high for most users B) It lacks the technical infrastructure to serve large audiences C) It works for individual writers but doesn't obviously extend to general social media use cases D) It still relies on advertising for a significant portion of its revenue
Answer
**C) It works for individual writers but doesn't obviously extend to general social media use cases** Substack solves the journalism and newsletter problem effectively, but the use cases that define dominant social media platforms — sharing with friends, discovering new connections, building community around shared experiences — are not obviously served by the Substack model. The chapter notes that "Substack solves the journalism problem; it does not solve the Facebook problem."Question 9. What does the chapter identify as the key difference between what Facebook's News Feed algorithms were optimizing for and what users believed they were receiving?
A) Users believed they saw all posts; the algorithm actually showed a curated subset based on advertiser preferences B) Users believed the feed was relevance-ordered by their stated preferences; the algorithm was actually constructing an emotional environment to maximize engagement C) Users believed the feed was chronological; the algorithm was actually showing them politically targeted content D) Users believed they controlled their experience; the algorithm was actually controlled by advertisers
Answer
**B) Users believed the feed was relevance-ordered by their stated preferences; the algorithm was actually constructing an emotional environment to maximize engagement** The chapter frames this as the central gap revealed by the emotional contagion study. Users consented to relevance-based curation — having their feeds prioritize content their behavior suggested they cared about. What they received was something different: a deliberately constructed emotional environment whose composition was optimized for an engagement metric that served the advertising model, not user preferences.Question 10. DuckDuckGo's business model is best described as which of the following?
A) Subscription-based search with a free tier B) Contextual advertising matched to search queries without behavioral profiling C) Selling anonymized user data to third-party advertisers D) A nonprofit funded by user donations
Answer
**B) Contextual advertising matched to search queries without behavioral profiling** DuckDuckGo shows ads based on what you search for, not who you are. If you search for hiking boots, you see ads for hiking boots — but DuckDuckGo does not track you across the web, does not build a behavioral profile, and does not sell your data. This is the contextual advertising model, which has lower CPMs than behavioral targeting but avoids the surveillance infrastructure.Part B: True/False with Explanation (7 questions)
For each statement, indicate True or False and provide a 2–3 sentence explanation.
Question 11. The Real-Time Bidding auction typically completes before the user is consciously aware that the page has loaded.
Answer
**True.** The RTB auction completes in under 100 milliseconds — the entire sequence of bid request, data enrichment, competitive bidding, winner selection, and ad creative retrieval happens in less time than it takes for the human visual system to consciously register that a page has begun loading. This invisibility is not incidental; it is structural to the system's design and essential to why users cannot meaningfully consent to or observe the auction in which their attention is being sold.Question 12. According to the chapter, platforms with higher engagement metrics can charge higher CPMs, meaning that engagement increases not just ad volume but also ad price.
Answer
**True.** Advertisers pay a premium to reach users who are deeply engaged with a platform, because engaged attention is more valuable — it is more likely to be noticed, processed, and acted upon. This "engagement multiplier" means that increasing engagement has a compound financial effect: more time on platform means more ad impressions, and a more engaged audience means each impression can be sold at a higher price. The incentive to maximize engagement is therefore doubly reinforced.Question 13. The chapter argues that the advertising business model is the sole cause of harmful platform behaviors, and that subscription-based platforms have no problematic incentive structures.
Answer
**False.** The chapter explicitly acknowledges that subscription models have their own incentive tensions. Netflix still optimizes for engagement (though the alignment is tighter), and Spotify has an incentive to make its free tier unpleasant enough to drive upgrades — what the chapter calls "friction manipulation." The chapter's argument is that the advertising model creates particularly severe misalignment between platform incentives and user wellbeing, not that subscription models are problem-free alternatives.Question 14. High-income users command higher CPMs than lower-income users, meaning platforms have stronger financial incentives to maximize engagement among wealthy users.
Answer
**True.** Advertisers pay more to reach high-income demographics because their purchasing power makes conversions more valuable. This creates a disturbing distributional dynamic: the advertising model gives platforms stronger financial incentives to design engaging experiences for wealthy users than for poor ones. The implications extend to everything from content recommendation (what gets surfaced to whom) to product investment (which user needs get resources) to content moderation (whose concerns get prioritized).Question 15. The Like button was primarily a social feature designed to replace longer comment responses, and its data collection significance was a secondary and unanticipated consequence.
Answer
**False (or substantially misleading).** While the Like button had obvious social utility, the chapter characterizes it as "the most important data collection instrument Facebook ever built." Whether its data significance was fully anticipated is uncertain, but the effect was transformative: it gave Facebook explicit emotional signal at scale — billions of data points per day indicating which content produced positive affective responses in which users. This data transformed the algorithmic capabilities of EdgeRank's successors in ways that shaped the platform's entire subsequent trajectory.Question 16. Wikipedia demonstrates that a platform serving hundreds of millions of users can be financially sustainable without advertising or subscription fees.
Answer
**True.** Wikipedia is funded by user donations of approximately $150–160 million annually and operates as a nonprofit with no advertising revenue. It is among the most-visited websites in the world. However, the chapter also notes that Wikipedia's cooperative model has significant structural problems — governance that skews toward certain demographics, slow consensus processes, hostility to new contributors — and that its use case (reference encyclopedia) is sufficiently different from social media that the model's transferability is limited.Question 17. The chapter argues that individual engineers and product managers bear primary moral responsibility for the attention economy's harms because they build the systems that exploit users.
Answer
**False.** The chapter explicitly argues the opposite: that the attention economy's harms are primarily structural, not personal. Individual actors — Sarah Chen, Marcus Webb, the engineers at Facebook — are making locally rational decisions within a system whose incentive structure produces harmful aggregate outcomes. The chapter does not excuse executives who knowingly suppress research documenting harm, but its central analytical move is to locate the primary cause of the problem in the business model and incentive structure, not in individual moral failings. Structural problems require structural solutions.Part C: Short Answer (5 questions)
Answer each question in 3–5 sentences.
Question 18. What did the emotional contagion study reveal about the relationship between Facebook's stated purpose (connecting people) and the actual operation of its News Feed algorithm?
Answer
The emotional contagion study revealed a profound gap between Facebook's stated purpose and the actual operation of its systems. The study showed that Facebook could deliberately alter users' emotional states — making them happier or sadder — through systematic manipulation of News Feed content, without users' knowledge or consent. More importantly, it revealed that Facebook had this capability but had not been using it to make users happier, implying that user emotional wellbeing was not the optimization target. The advertising model's demand for engagement — not emotional health — was driving the algorithm's behavior. The study made visible a capability that had presumably been operating, in less deliberate form, for years.Question 19. What is the "incentive trap," and why does it make it difficult for even well-intentioned platform companies to prioritize user wellbeing over engagement maximization?
Answer
The incentive trap is the structural dynamic whereby every layer of a platform's environment — measurement systems, competitive markets, capital markets, talent markets, and regulatory frameworks — rewards engagement maximization and punishes deviation from it. A platform that sincerely wanted to prioritize user wellbeing would face: no validated way to measure wellbeing at scale in real time; competitors who would not reciprocate and would gain users; investors who would cut valuations when engagement metrics declined; and a talent market that rewards engagement optimization expertise more than ethical design expertise. Each layer creates a separate obstacle, and the layers compound each other, making unilateral change by any single platform commercially perilous even when executives genuinely want to make it.Question 20. Using the Velocity Media example, explain the difference between what a product feature's designer intends it to do and what it actually does within the advertising business model's incentive structure.
Answer
Marcus Webb's product features — autoplay, personalized feed, push notifications, social sharing — each had genuine user-benefit rationales: helping users discover content, not miss uploads from favorite creators, and build community around shared learning. But each feature also, mechanically, increased engagement metrics. Within the advertising model, this dual function means the feature's commercial justification is engagement, not user benefit — and when the two diverge (as when Dr. Aisha Johnson showed that the algorithm was surfacing dramatic content over educational content), the engagement rationale is financially protected while the user-benefit rationale is merely aspirational. The designer's intention is real, but the incentive structure selects for the engagement effect regardless of intent.Question 21. Describe what "brand safety" means in the programmatic advertising context, and explain the unintended consequences it has for content ecosystems.
Answer
Brand safety refers to advertisers' concern that their ads not appear alongside content that could embarrass their brand — violence, hate speech, graphic imagery, or political controversy. In programmatic advertising, brand-safety enforcement is largely automated: algorithms scan content and apply advertiser-specified exclusion lists. The unintended consequence is a system of financial penalties on certain types of content: controversial or politically sensitive content receives deeply discounted CPMs or no advertising revenue at all, even when that content is legitimate journalism or important public discourse. This creates a perverse incentive structure where engagement-generating sensationalism may be financially rewarded while careful analysis of difficult topics is financially punished — not by deliberate editorial policy, but by the automated aggregation of advertiser risk-aversion.Question 22. The chapter ends with the claim: "The system worked as designed. The design was the problem." What does this mean, and why does the distinction matter for thinking about solutions?