Chapter 33: Game Design Ethics — Engagement vs. Exploitation, Dark Patterns, and Responsibility

There is a designer somewhere right now, in a meeting room with frosted glass walls, looking at a chart that shows player retention as a function of how aggressively the game pushes a limited-time offer. The chart is going up. The product manager is happy. The designer is being asked to push it harder — to add a third pop-up, to shorten the timer, to make the discount badge red instead of orange. The designer does the math in their head: this will move retention by another two points. It will also make the game worse for every player who is not already addicted to it. They open their mouth to push back. They close it. The meeting ends. They go home and do not sleep well.

This chapter exists for that designer. It also exists for the designer one cubicle over, who pushed back successfully last quarter and is wondering what it cost them. It exists for the student who is about to take their first job at a free-to-play studio and has no idea what they are about to be asked to do. And it exists for the player, downstream of all of these decisions, who will pick up the game and either be delighted or be drained.

Game design is one of the most powerful behavior-modification technologies humans have ever built. We know — because we have measured it — exactly which patterns of feedback, reinforcement, scarcity, and social pressure cause people to spend more time and money than they intend. We can A/B test our way to any retention number we want, given enough players and enough patience. The question that has not been settled, and that the industry has spent twenty years failing to settle, is whether we should.

This is not a chapter about ethics in the abstract. It is a chapter about specific games, specific patterns, specific harms, and specific choices designers make every day. We will name games. We will name patterns. We will draw a line — imperfect, contested, but real — between engagement (the player wants to come back because the game is good) and exploitation (the player comes back because the game has hooked into something deeper than wanting). And we will end with a practical framework: not a checklist, but a set of questions you can ask in the meeting room when the chart is going up.

Why This Chapter Exists

For most of game design's history, ethics was a topic for the academics. Working designers had craft problems — pacing, feedback, balance — and the moral weight of what they were doing was limited to "is the violence too gratuitous" or "are the female characters dressed sensibly." These are real questions, but they are surface questions. They are questions about content. The deep ethical questions in games are about systems — about what the game is doing to the player's brain when they play it for an hour, six hours, six hundred hours.

Three things happened that forced the conversation.

The first was the rise of free-to-play, somewhere between Zynga in 2009 and Candy Crush in 2012. F2P broke the old assumption that you sell a game once and the player owns it. Instead, the game became a service that monetized continuously, which meant the design goal shifted from "give the player a satisfying experience and let them go" to "give the player a satisfying-enough experience that they keep paying." This is a profound shift in the designer's incentive structure, and it took the industry a decade to grapple with what it had become.

The second was loot boxes — specifically, the Star Wars Battlefront II launch in November 2017, which we will examine in Case Study 1. Battlefront II made loot boxes legible to the mainstream. Suddenly U.S. senators were asking about them. Belgium and the Netherlands were investigating them. Gamers were posting screenshots of their kids' surprise credit-card statements. The industry was forced to admit, in public, that some of what it called "engagement" was indistinguishable from gambling, and that the distinction mattered.

The third was the slow rotting of the workplace itself — the Red Dead Redemption 2 crunch reporting in 2018, the Activision Blizzard harassment lawsuits in 2021, the Riot Games lawsuits, the Ubisoft "boys' club" reporting. The industry that was selling people exploitative products was also exploiting the people who built them, and these turned out not to be unrelated facts.

Put those three things together — extractive monetization, gambling-adjacent mechanics, exploitative labor — and you arrive at the present moment, in which "game design ethics" is no longer a niche academic concern. It is a working condition. You will be asked, sooner than you think, to design something whose ethics you are not sure about. This chapter is meant to give you the vocabulary and the categories to think about that moment before it arrives.

The good news is that the industry is also producing some of the most ethically-thought-through work in its history — Celeste's assist mode (Ch 11), Hades's representation, Tunic's lack of microtransactions, the wave of indie titles that simply refuse to participate in the F2P playbook. Ethical design is not a fantasy. People are doing it. It is harder, and it pays less in the short run, and it requires more spine. But it is not theoretical.

💡 Intuition: The ethics of game design is not a separate skill from the craft of game design. It is the craft of game design, taken seriously. Every design choice — every pacing decision, every reward schedule, every UI button placement — is also an ethical choice, because it is a choice about what to do to a person. The pretense that you can separate "designing a fun system" from "designing a system that affects a human being" is the central evasion this chapter is trying to break.

Engagement vs. Exploitation — The Defining Tension

Begin with the words. Engagement and exploitation sound like opposites, but in actual design practice they often look identical from the outside. A player logs in every day. A player spends three hours in the game tonight. A player buys the cosmetic. From the analytics dashboard, you cannot tell whether this player loves your game or whether your game is consuming their life against their better judgment.

Here is the working distinction. Engagement is the player wanting to come back because the game offers them something they value — challenge, expression, community, story, the joy of mastery. The player's wanting and the player's playing are aligned. If you took the game away tomorrow, they would miss it the way you miss a good book or a good restaurant. The game adds to their life.

Exploitation is the player coming back because the game has installed a compulsion in them that operates beneath the level of wanting. The player plays, but they are not sure they want to. They feel guilty about the time they spent. They check the game while the partner is talking. They spend money they did not intend to spend and feel bad about it after. If you took the game away tomorrow, they might feel relief. The game subtracts from their life, but they cannot stop playing it.

Notice that the same game can be engaging for one player and exploitative for another. World of Warcraft is a beloved game for some people and a quiet ruin for others. Candy Crush is a pleasant five-minute distraction for some and a $200/month addiction for others. The mechanics are the same. The difference is in the player's vulnerability and the way the game's systems intersect with their psychology.

This is why ethical game design cannot be reduced to "if a game does X, it is unethical." A given mechanic can be fine for the median player and devastating for the vulnerable player at the tail of the distribution. The ethical question is not just "what does this do to the average player" but "what does this do to the player who is most susceptible to it, and is the harm to them justified by the benefit to everyone else?"

This framing — average vs. tail — is uncomfortable, because it demands that you take seriously the existence of players you will never meet who will be hurt by your choices. The industry's standard evasion is to pretend the tail does not exist. "Most players don't spend that much." "Most players don't play that long." Statistically true, ethically irrelevant. The industry's revenue is not coming from the median player. The industry's revenue is coming from the tail.

Here is a number to remember. In free-to-play mobile games, somewhere between 0.15% and 2% of players generate 50% to 80% of revenue. These are the players the industry calls "whales." Whale revenue is not whale users — it is a small group of users spending huge amounts. When you optimize a game for revenue, you are not optimizing it for the median player. You are optimizing it for the whale. And the design choices that make a game most extractive for the whale are usually the choices that make it least pleasant for everyone else.

The defining tension of contemporary game design is this: the techniques that maximize revenue and the techniques that maximize player wellbeing are not the same techniques, and a designer working in a commercial studio is constantly being asked to choose. Pretending the choice does not exist is an ethical position. It is the wrong one.

Operant Conditioning Revisited

We covered motivation and reward in Chapter 12. We are going to cover it again, because the same psychology that makes a game fun is the psychology that makes it addictive, and the difference between the two is mostly a matter of how cynically the designer deploys it.

B.F. Skinner taught pigeons to peck levers in the 1930s and 1940s. He discovered that the pigeons pecked harder, longer, and more compulsively when the lever delivered food on a variable ratio schedule — that is, a random average number of pecks per pellet — than under any other reinforcement pattern. Continuous reward (one peck, one pellet) made pigeons that pecked when hungry. Variable ratio reward (one pellet every ten pecks on average, but with the actual number unpredictable) made pigeons that pecked compulsively, that pecked past hunger, that pecked until the experimenters intervened.

The same pattern is the operating principle of every slot machine ever built. It is also the operating principle of loot boxes, gacha systems, and most of the random-reward systems in modern games. This is not a metaphor. It is the same psychological mechanism, applied for the same reasons, and verified to produce the same compulsive behavior.

A slot machine and a loot box differ in a few legal details and almost nothing else. Pull lever, watch wheels spin, get reward of variable size including occasionally a big one. Repeat. The dopaminergic system in the brain — the wanting system, distinct from the liking system — fires hardest in the seconds before the reward arrives, when the outcome is uncertain. This is what makes variable-ratio reinforcement so powerful. The wanting fires every time. The liking fires only sometimes. The brain, given enough repetitions, learns to want the activity even when it does not particularly like it.

Now — and this is important — variable ratio reinforcement is not inherently unethical. It is the basic structure of how humans learn most skills. Practicing scales on a piano is variable ratio (sometimes a phrase comes out beautifully, sometimes not). Trying to land a hard combo in a fighting game is variable ratio. The randomness of life is variable ratio. The brain is built for it.

The ethical line is not "no variable ratio." The ethical line is about what is being conditioned, and at whose expense, and with what costs to the player. A player practicing a hard mechanic in Hollow Knight is being variable-ratio conditioned to pursue mastery — but the reward is genuine skill, the cost is time freely chosen, and the conditioning makes the player better at something they wanted to be good at. A player opening loot boxes for a chance at a cosmetic is being variable-ratio conditioned to spend money — and the reward is a digital trinket, the cost is real money sometimes from real savings, and the conditioning makes the player into a slightly worse version of themselves.

The same neural mechanism. Different ethical valences. The valence depends on what the conditioning is for.

⚖️ Ethical Framework: When you build a system around variable-ratio reinforcement — and you will, because almost every game does — ask three questions. What skill or outcome is the player being conditioned toward? (Mastery and play are usually fine; spending and time-sink are usually not.) What is the cost to the player of the conditioning? (Time on its own is okay; money and self-worth are not.) Could the same fun be produced with a less compulsive schedule? (If yes, you owe it to the player to use the less compulsive schedule and accept the engagement loss.)

Loot Boxes and Gambling Mechanics

Loot boxes are the case study that broke the conversation open. They deserve their own treatment.

A loot box is a virtual container, purchasable with real money or earnable through play, whose contents are randomized. The player pays, opens, and receives one or more items from a defined pool, each item with a defined (and usually undisclosed) probability. The most desirable items have the lowest probabilities. The player can buy as many boxes as they want.

This is gambling. It is gambling with three legal fig leaves: (a) the prize is virtual rather than redeemable for cash, (b) every pull is "guaranteed" to return something, and (c) the platform owners have lobbied successfully in most jurisdictions to keep gambling regulators away from games. None of these fig leaves alters the psychological mechanism. The brain processes a loot box opening using the same dopaminergic pathways it uses for a slot machine. The compulsion is the same. The harm to the vulnerable player is the same.

The pattern entered the mainstream in the early 2010s. Team Fortress 2 introduced the "Mann Co. Supply Crate" in 2010, requiring a separately-purchased key to open. FIFA Ultimate Team shipped in 2009 and quickly became EA's most profitable franchise — by 2021 it generated roughly $1.6 billion per year, almost all of it from "FUT" packs, which are loot boxes. Overwatch (2016) launched with cosmetic-only loot boxes, drawing initial praise for keeping random rewards out of competitive gameplay. Star Wars Battlefront II (2017) crossed the line into pay-to-win loot boxes that affected combat performance, and the world noticed. (See Case Study 1.)

The regulatory response has been uneven and country-specific. Belgium, in 2018, declared paid loot boxes a form of illegal gambling, forcing publishers to remove them from Belgian players' versions of FIFA, Overwatch, and several other games. The Netherlands issued a similar ruling, fought it in court, and eventually settled into an enforcement pattern that pressured EA to remove FUT packs from the Dutch market in 2022. The UK, after a parliamentary inquiry in 2019, opted for industry self-regulation rather than legal classification — a decision that has not aged well. The U.S. has not regulated loot boxes federally; the FTC has held workshops, the ESRB added the "In-Game Purchases (Includes Random Items)" label in 2020, but federal action has not happened. China has required disclosure of probability rates since 2017, which is why almost every major game now publishes its loot-box odds in some form.

The ESRB label deserves comment. "In-Game Purchases (Includes Random Items)" is a label, not a regulation. It tells parents what the game contains. It does not restrict the sale of loot-box games to children. A 13-year-old can still buy FIFA and spend a parent's credit card on FUT packs. The label is a fig leaf the size of a label, designed to forestall actual regulation by demonstrating the industry is "doing something."

The defense the industry offers — and you will hear this in meetings — runs roughly: players choose to buy loot boxes; we offer them as one option among many; the player gets value (the items, the dopamine, the surprise) for their money; calling this gambling is paternalistic. Every clause of this defense is technically true. None of it engages the actual issue, which is that variable-ratio monetization, deployed on millions of players, will reliably produce a tail of players whose participation is compulsive and whose spending is harmful to themselves and their families. The industry knows this. The industry has the data. The defense is a deflection.

🎮 Case Study Note: The single most-downvoted comment in Reddit's history (as of this writing) is an EA community manager's response to a Battlefront II complaint, posted November 2017, defending the loot-box pricing with the line "the intent is to provide players with a sense of pride and accomplishment for unlocking different heroes." This single post, more than any regulatory ruling, marks the moment the gaming public stopped accepting the loot-box framing.

Gacha and Pity Systems

If loot boxes are the Western face of variable-ratio monetization, gacha is the Asian face — older, more refined, more openly-acknowledged, and arguably more sophisticated. The word comes from gachapon, the Japanese coin-operated capsule-toy machines you have probably seen at a mall: drop in a coin, twist a knob, get a random toy in a plastic capsule. Mobile gacha games translate the mechanic into virtual currency, virtual capsules, and virtual rewards.

The major gacha games, as of this writing, include Genshin Impact, Honkai: Star Rail, Fate/Grand Order, Honkai Impact 3rd, Arknights, Blue Archive, Uma Musume, Granblue Fantasy, and a long tail of Korean and Chinese titles. Together they generate tens of billions of dollars annually, with Genshin Impact alone reportedly grossing over $5 billion in its first three years.

The mechanics are honest about what they are. You spend a "pull" — purchased with a premium currency, which is purchased with real money — for a chance at a featured character or weapon. The "banner" advertises a 5-star (highest-rarity) reward. The probability of pulling that reward on any single pull is usually around 0.6%. The probability of pulling any 5-star is usually around 1.6%, of which only half might be the featured one (the other half is a random non-featured 5-star — the "50/50").

To prevent the mathematically-inevitable run of bad luck from producing total despair, gacha systems include a pity system. Pity is a counter that tracks how many pulls you have done without a 5-star. At a defined threshold — usually 90 pulls — a 5-star is guaranteed. There is also typically a "soft pity" that boosts the rate from around pull 75 onward. If you "lost" the 50/50 (got a non-featured 5-star), the next 5-star you pull is guaranteed to be the featured one.

This is mathematically more humane than the pure-RNG loot box of Battlefront II's era. It guarantees an upper bound on disappointment. It makes the variable-ratio reinforcement somewhat more predictable, which dampens the worst compulsive behavior. The industry uses pity as a defense: we are not gambling; we have guarantees; the player will always eventually get the prize.

But pity systems also do something else. They convert a gambling game into a grinding game with a known target price. Players quickly learn to calculate the worst-case cost to guarantee a featured character — typically ~180 pulls if you lose every 50/50, which at retail prices is around $300. Players treat this number as the real price tag. They save toward it. They budget. They spend. The pity system makes the spending feel rational and budgetable, which makes it easier to spend more.

And the pity system masks the underlying economic structure, which is that gacha games are explicitly designed around a small number of whales funding the experience for a large number of free or low-spending players. Industry ARPPU (average revenue per paying user) figures suggest the top 10% of payers contribute 70-85% of revenue, and within that top 10%, the top 1% contribute the bulk. Whales in gacha are not casual customers — they are users who routinely spend thousands of dollars per character to "max out" their builds.

The ethics here are complicated. Gacha games are often genuinely good games — Genshin Impact is gorgeous and richly designed; Arknights has tactical depth comparable to any premium strategy game; Honkai: Star Rail tells a real story. The pity systems do reduce the worst gambling outcomes. The economic model — many free riders subsidized by a few whales — is not necessarily exploitative if the whales are spending money they can afford and choose to spend.

But the economic model is also the design model. The game's combat balance, character power curves, content release schedule, and event timing are all tuned to push the whales to spend. Featured banners appear on a rotation; if you do not pull this character now, you wait six months for the rerun. Powerful character combinations require multiple banner units. Power creep ensures that yesterday's whale-purchased character is today's mid-tier and tomorrow's bench-warmer. The whale is always one purchase away from being competitive again.

A designer working on a gacha game is not designing a game in the traditional sense. They are designing a treadmill calibrated to extract money from a small group of users at the upper bound of what those users can be conditioned to spend. The work is interesting; the visuals are beautiful; the game can be genuinely fun to play; and the underlying purpose is what it is.

🛠️ Practitioner Tip: If you take a gacha-game job — and many of you will, because that is where much of the industry's hiring is — keep your eyes open about what you are designing. Internal monetization meetings will use language like "engagement runway," "spender progression," "soft-cap economy." Translate these terms back to plain English when you go home. "How much can we extract from this person before they leave or run out of money" is the question being asked, even when it is not the question being said.

Dark Patterns in Games

The term dark pattern was coined by UX designer Harry Brignull in 2010 to describe deceptive interface designs in e-commerce — the pre-checked subscription box, the obscured unsubscribe link, the fake countdown timer. Games adopted these patterns and invented new ones. We covered the UI manifestations in Chapter 29. Here we take them up as ethical objects.

A dark pattern is a design choice that intentionally exploits a known cognitive bias to produce a behavior the player would not freely choose. It is distinguished from "good design" by two features: first, it works against the player's interests; second, it relies on the player not noticing it works that way.

The major dark patterns in games:

FOMO (Fear of Missing Out) via limited-time content. The event runs for 14 days. The character is featured for one banner cycle. The skin is on sale until midnight. The mechanic exploits loss aversion: humans feel the prospect of losing something more strongly than the prospect of gaining the equivalent. Limited-time events make every player who is not currently logged in losing something, even if they have no particular need for the item. Fortnite's rotating cosmetic shop, Genshin Impact's banner schedule, Destiny 2's seasonal exotics — all built on FOMO. The pattern works. It is also corrosive to the player's relationship with the game, because it converts play from a thing you do when you want to into an obligation you have to keep up with or fall behind.

Anchoring via fake original prices. The skin's "original price" is shown crossed out at $20, with the "sale price" of $8 highlighted in green. The original price is fictional — the item has never been sold at $20. The crossed-out number exists only to make the $8 seem like a deal. This is a textbook anchoring exploit, and it appears in nearly every cosmetic shop in nearly every F2P game.

Sunk-cost via daily login streaks. The 30-day login streak rewards a special item on day 30. Miss day 17 and you lose your streak — the counter resets to zero. The 16 days you already logged in were "wasted." This converts logging in from a choice into an obligation, weaponizing the sunk-cost fallacy against the player. Genshin Impact, Destiny 2, Hearthstone, and most mobile games use some version. The most cynical version requires daily paid activity (a daily quest tied to a battle pass) to maintain the streak.

Artificial friend pressure (life-gifting). Candy Crush invented the pattern in 2012. You are out of lives; you can wait an hour, pay 99 cents, or ask a friend on Facebook for a free life. The friend gets a notification: "Sarah wants to keep playing and needs a life from you!" Half the friends do not play the game; the notification is just spam designed to bring them into the game. The other half play the game and now feel obligated to send a life back, lest they be the friend who didn't help. The pattern weaponizes social ties for retention.

Bait-and-switch monetization. The game is free. The first three chapters are charming. At chapter 4, a paywall: pay $9.99 to continue, or grind for 200 hours. The player has invested four hours and developed an emotional connection to the characters; the sunk cost makes the $9.99 feel like a small price. The pattern has been present since Dungeon Keeper Mobile in 2014, which EA managed to make so egregious that even the 2014-era press noticed.

Pay-to-skip-the-pain. Build a tedious progression system. Then sell a "speed-up" or "auto-clear" item that bypasses it. The progression system was designed to be unpleasant in order to make the speed-up feel valuable. Diablo Immortal pioneered the modern version: legendary gem grinding so slow that even committed free players experience it as torture, with the pay option always available.

Currency obfuscation. The game has 4 currencies (gems, coins, crystals, dust). Some are earned, some are purchased, conversions are not 1:1 and are not the same in both directions. The player cannot easily translate "I want this $5 item" into "I am about to spend $5." This exploits the well-documented finding that humans spend more freely with abstract currency than with cash. Clash of Clans, Hay Day, and most mobile RPGs do this.

The infinite-scroll lobby. The game's main menu is not a menu — it is a feed of events, offers, news items, login bonuses, and friend activity, designed to keep you in the lobby (and therefore in the data-collection environment) even when you are not playing. Mobile gacha games are the worst offenders. The lobby is the casino floor.

⚠️ Pitfall: Dark patterns sneak into well-meaning teams because each individual pattern can be defended on its own merits. The countdown timer "creates excitement." The login streak "rewards loyalty." The crossed-out price "shows value." Defended one at a time, each pattern sounds reasonable. Stacked together — and they always stack — the patterns convert a game into a behavior-modification machine. The audit you do for the progressive project this chapter is meant to make the stacking visible.

Ethics of Difficulty and Accessibility

The "Dark Souls should have an easy mode" debate has been litigated to death online. Let us treat it honestly here, because the way we treat it reveals our ethics.

The case against an easy mode, made by FromSoftware fans and sometimes by FromSoftware itself, runs: the difficulty is the game. The triumph after twenty deaths against Ornstein and Smough is not separable from the twenty deaths. To make the game easier is to give the player a worse experience disguised as a more-accessible experience. The artist has a right to their vision; the player has the right to choose another game.

The case for an easy mode runs: difficulty is a mechanic, not a sacred object. The player who is disabled, the player whose reflexes have aged out of high-difficulty play, the player who has limited time and wants to see the story — all of these players are excluded by a design choice that presents itself as artistic necessity but that is, in fact, a choice. Most games offer difficulty options. Why not this one?

The honest version of the debate is this. Dark Souls without its difficulty would be a different game; it might be a worse game. The artist's vision is real. AND — both/and — the design choice excludes a real population of real players, and the artist owes that population at least an honest reckoning rather than the dismissive "git gud" that the community offers. Both things are true. Designers who pretend the first thing is the only true thing are dodging the second; designers who pretend the second is the only true thing are dodging the first.

Celeste is the gold-standard counter-example. It is a punishingly difficult precision platformer. It also includes an "Assist Mode" that lets the player slow the game down, give themselves invincibility, give themselves infinite stamina, or skip chapters entirely. The design choice was deliberate. Maddy Thorson, the designer, wrote: "Celeste is a game about overcoming personal challenges, including but not limited to twitch-platforming challenges. If the platforming gets in the way, that's a barrier to engaging with the larger themes of the game, and the game should adapt." Assist Mode does not break the game's vision; it widens its audience without compromising its core. It is a design choice made by an artist who took accessibility seriously.

Compare Hades, which offers "God Mode" for players who want a less punishing roguelite experience. Compare the wave of "Story Mode" difficulties in narrative-heavy games (Spider-Man, God of War, The Last of Us). Compare the deep accessibility work in The Last of Us Part II, which won awards for the breadth of its options — high-contrast modes, audio cues, motion-control alternatives, navigation assists, adjustable difficulty per-system.

The false opposition — "artistic vision vs. accessibility" — has been refuted in practice many times over. The accessibility options are not a watering-down of the vision; they are a recognition that the vision can be experienced by more people if the surface controls allow it. A designer who refuses to offer accessibility options is making a choice to exclude. They have the right to make that choice. They do not have the right to pretend they did not make it.

🛠️ Practitioner Tip: The accessibility appendix of this textbook (Appendix I) gives you a checklist. Run your game through it. If a category does not apply (your game has no audio, your game has no precise inputs, etc.), note that. If a category applies and you have not addressed it, that is a known gap, not a successful design — and gap is okay if you acknowledge it. The thing to avoid is the unexamined exclusion that you simply did not think about.

Representation and Inclusion

Who is in the game. Who is not. Who is the protagonist; who is the NPC; who is the boss; who is the joke.

These are design choices. They are also ethical choices, because they participate in the construction of who the medium thinks is normal, who is heroic, who is dangerous, and who is invisible.

For most of game history, the protagonist was a white man with brown hair and stubble, the female characters wore impractical armor and existed to be rescued or seduced, the villains had non-Western accents, and queer characters did not exist. This was not a conspiracy. It was a default. The default was produced by a workforce that was overwhelmingly white and male, designing for an audience they assumed was overwhelmingly white and male, in a genre vocabulary inherited from action movies and pulp novels with the same defaults.

The default has shifted, slowly. The Last of Us (2013) gave us Ellie. Horizon Zero Dawn (2017) gave us Aloy. Celeste (2018) gave us Madeline, who is canonically trans (as the designer revealed in 2020 — Maddy Thorson came out at the same time). Hades (2020) made the protagonist canonically bisexual, in a casual matter-of-fact way that did not turn his sexuality into the plot. Spider-Man: Miles Morales (2020) put a Black-Latino teenager in the lead. The Forgotten City (2021) and Pentiment (2022) did serious work with non-default protagonists. The work continues.

The work is also fragile. For every step forward, there is a backlash. The harassment campaigns against women and BIPOC developers (covered in the next section) are partly about the demographic fact that the medium is changing and a portion of the audience is angry about it. Designers who push representation know they are taking on a personal cost — sometimes the cost is online harassment, sometimes it is internal pushback, sometimes it is "marketing" telling them the protagonist needs to be changed for sales.

A few things to remember about representation as design ethics:

Representation is not "adding diverse NPCs." A game with a white-male protagonist and a token Black sidekick and a token female love-interest and a token gay best friend has not done the work. Representation is who the protagonist is, who has narrative agency, whose perspective the player is asked to inhabit. The peripheral NPCs are easy. The protagonist is the question.

Representation should be material, not decorative. A character whose race or gender or sexuality is incidental to the story does not necessarily have to make it the plot, but the character should also not feel like a re-skin of the default protagonist. Hades's Zagreus is bisexual not because the plot requires it but because the writers wrote him as a real character whose sexuality is part of who he is. Tell Me Why (2019) made trans identity central to the narrative without reducing the protagonist to his transness.

The genre conventions are themselves the problem. A first-person military shooter is a genre that historically excluded most representation; the genre's conventions assume a white-male soldier protagonist. To make a genuinely representative shooter, you do not just change the protagonist's skin — you have to interrogate the genre. Spec Ops: The Line (2012) is the canonical example of a shooter that turns its genre against itself. The deeper representational work is genre work, not character work.

Representation in the credits matters too. Who is on the team. The hiring pipeline. Who gets to speak in design meetings. Who gets credited; who gets paid; who gets promoted. A studio with a homogeneous team will produce homogeneous work, no matter how many representation guidelines they write down. The work of inclusion in the game and the work of inclusion in the studio are the same work.

🎮 Case Study: The reception of The Last of Us Part II (2020) is instructive. The game put a queer woman as protagonist for a substantial portion of its runtime; it took the player through a story about cycles of violence and the cost of revenge; it made design choices that some players hated. The harassment of cast and crew was extensive and well-documented. The game also won Game of the Year at multiple ceremonies and is now widely considered a defining work of the medium. The lesson is not "everyone will love the brave choice" — they will not. The lesson is that the brave choice is the right choice, and the cost of making it is part of the work.

Gamer Culture, Harassment, and Designer Complicity

Gamergate happened in August 2014. The pretext was an indie developer's relationship history; the actual content was a coordinated harassment campaign against women in games that produced doxxing, death threats, and forced relocations of multiple targets. The campaign reshaped gaming culture in ways that are still being felt. It also exposed something the industry had been avoiding: a substantial fraction of the audience treats games as their identity, treats criticism of games as personal attack, and treats the people making games as legitimate targets.

This is a culture. It is not a small culture. The same dynamics produced the harassment of The Last of Us Part II's cast, the review-bombing of any game perceived as "woke," the harassment of voice actors over performance choices they had no creative control over, and the daily dogpiling of women and BIPOC developers on social media. None of this is incidental to game design. It is part of the cultural ecosystem your game ships into.

What can a designer control? Not the culture. Not the harassment campaigns. Not the trolls in your Twitter mentions. You did not cause it; you cannot end it. But you can make choices about what your game does within it.

You can build anti-harassment features into your multiplayer (we covered this in Chapter 28). You can build robust moderation systems. You can not promote toxic content creators. You can not make community features that allow voice chat with strangers without an opt-in. You can give your customer service team the resources to actually respond to abuse reports. You can design report flows that treat the reporter as a customer rather than as a nuisance.

You can also not pretend your game is neutral. A multiplayer shooter that has no anti-toxicity infrastructure is making a choice. A community tool that lets players send rich-media messages to strangers is making a choice. A game that incentivizes one player to steal from another, with no recourse, is making a choice. The choice has consequences. Designers who say "we just made the systems; what people do with them is on them" are dodging — they made the systems, and the systems shape the behavior.

The other thing you can do is not contribute to the culture from inside the building. Studios that tolerate harassment internally — and many of them do — produce designers who carry those norms into their work. The Activision Blizzard lawsuits in 2021 documented a "frat-boy culture" that included harassment, suicide of a female employee, and a leadership response that prioritized PR over change. The Riot Games lawsuits documented similar patterns. The Ubisoft reporting in 2020 produced a wave of senior departures. None of this is unrelated to the games these studios produced. A culture that tolerates the harassment of its own people will not produce games that take harassment of players seriously.

The work of changing this is largely not the individual designer's. It is structural. But the individual designer can do three things. Listen to the colleagues you have who are being treated worse than you. Believe them when they describe what is happening to them. Push back internally when leadership defends bad behavior, even at cost to yourself — there is no such thing as a private complicity. Support unionization efforts (next section), because the only durable counterweight to power abuse is collective power.

Data, Telemetry, and Privacy

Chapter 31 covered telemetry as a design tool. It is also an ethical surface. Every event your game logs is a piece of data about a real human being. What you do with that data is an ethical choice.

Most games today log far more than the design team needs. The default mobile-game SDK ships with hooks into ad networks (AdMob, Unity Ads, Facebook Audience Network) and analytics platforms (Firebase, AppsFlyer, Adjust) that collect device IDs, location, app usage, and a lot of behavioral data. The data is sold or used to target ads. The player typically does not know.

The free-to-play mobile industry's dirty secret is that the game is sometimes not the product. The player is the product, and the game is the harvesting tool. A "hyper-casual" game that you play for two minutes and then forget exists primarily to install an SDK on your phone that will track you across other apps and feed an advertising graph. The game is bait.

This is legal in most jurisdictions, in the sense that the user accepted a 47-page Terms of Service when they installed the game. It is not legal everywhere — GDPR in Europe requires meaningful consent and the right to data deletion; COPPA in the U.S. restricts data collection from children under 13; California's CCPA requires disclosure and a right to opt out. Compliance with these regimes is patchy, mostly because the cost of non-compliance is small relative to the revenue.

A designer working on a game that does this should know it is happening. The data team is not lying when they say "we use telemetry to improve the game" — they may use some of it for that. But the data is also feeding the ads. The data is also being aggregated and sold. The data is also building a profile of your players that will follow them across apps for years.

The minimum ethical bar for a designer dealing with telemetry:

• Know what data your game collects. Ask. Get a real answer.
• Know what the data is used for. "Game improvement" is a partial answer; the rest is ad-tech.
• Know which third parties have access. Look at the SDK list.
• Push back on collection that is not necessary for the game. Most of it isn't.
• Make your privacy policy actually readable. Most of them are written to be unreadable.
• If your game targets minors, double everything.

The kids' games question deserves its own section.

Kids' Games and a Higher Bar

A kid is not a small adult. A kid's brain — the prefrontal-cortex-mediated impulse-control circuitry, especially — is not fully developed until the mid-twenties. A 9-year-old presented with a loot box, a battle pass, an FOMO timer, and an in-app purchase prompt is not making the kind of choice we mean when we use the word "choice" with adults. They are being conditioned, by an industry of professional conditioners, with developmental machinery that is not equipped to fight back.

The ethical bar for games marketed to children is therefore higher. It should be higher in your design. It should be higher in regulation. It is higher in some places — COPPA (the Children's Online Privacy Protection Act) in the U.S. restricts data collection from under-13s, PEGI 3 in Europe requires content suitable for very young children, the ESRB rates games and labels in-game purchases. None of these regimes do what they should do, but they are not zero.

Two cases.

Roblox is, by player count, the largest game in the world for children — over 70 million daily active users, the majority under 16. Roblox is a platform, not a single game; users create games, other users play them, the platform takes a cut. The moderation problem is structural: there are too many user-created games for human review, and the automated moderation has reliably failed to catch grooming, sexual content, gambling-style mini-games, and IP theft. Roblox has been the subject of multiple investigations into both child safety and labor practices (children developing games for trivial or no compensation while Roblox takes 70% of revenue). The company's response has been incremental and reactive. The structural problem persists because it is structural.

Fortnite settled with the FTC in December 2022 for $520 million — the largest gaming-industry settlement to that point — over two charges. First, COPPA violations: collecting personal data from under-13 players without parental consent. Second, "dark patterns" that made unintended purchases easy and refunds hard, including a "quick-purchase" UI that let kids spend V-Bucks without meaningful confirmation. Epic agreed to refund affected players, change its UI, and submit to FTC monitoring. The settlement is significant because it explicitly named dark patterns as the violation — the FTC was treating the design choices, not just the data collection, as the harm.

The lesson for designers working on anything that might be played by children: the bar is higher, and it should be. No loot boxes in kids' games. No purchase flows that a 9-year-old can complete without an adult. No endless feeds. No engagement-maximizing design optimized against developmental brains. This is one of the few places where the ethical answer is unambiguous: do not deploy psychological-manipulation techniques against children.

The pushback you will hear: "but if we don't, our competitors will, and we'll lose the market." This is true, and it is also the standard logic of every race-to-the-bottom in human history. It is not a defense. It is an explanation of why regulation, not industry self-restraint, is the only durable solution. As a designer you can refuse the work; as a citizen you can support the regulation; as a member of the industry you can vote with your career.

Labor Ethics — Crunch, Contracts, and Credits

Game design ethics that does not include the ethics of how games are made is incomplete. The conditions of production shape the product. A studio that exploits its workers will produce games that exploit their players, more often than not, because the same management mindset is responsible for both.

Crunch. "Crunch" is the industry term for the period before launch when teams work 60-80-100 hour weeks for months at a time, often unpaid (because most game-dev employees are salaried), to ship the game. Crunch is sometimes presented as a one-time emergency; in practice it is structural. Almost every major game ships through crunch. Red Dead Redemption 2 was reported in 2018 to have involved 100-hour weeks for substantial portions of its development. Cyberpunk 2077's 2020 launch debacle was preceded by a year of crunch. The Witcher 3's acclaimed 2015 release was built on crunch. God of War Ragnarök's 2022 launch, despite Sony's stated anti-crunch policy, included crunch.

Crunch destroys people. It produces burnout, depression, divorce, miscarriages, and well-documented physical health problems. It also produces worse games — fatigued people make bad decisions, ship bugs they would have caught when rested, and design systems with a brittle desperation rather than a confident craft. Crunch is presented as the cost of ambition; in reality it is mostly the cost of bad project management and a refusal to cut scope.

Harassment. The Activision Blizzard lawsuits filed by California's Department of Fair Employment and Housing in July 2021 documented a "frat boy culture" of harassment, including a female employee's suicide, retaliatory firings of women who complained, and a leadership cover-up. The CEO, Bobby Kotick, was reported to have been informed of allegations and to have intervened on behalf of accused executives rather than victims. Microsoft acquired the company in 2023; many of the implicated executives remained. Riot Games settled a class action in 2021 for $100 million over similar patterns. Ubisoft lost multiple senior executives in 2020 after a wave of harassment reporting.

These are not isolated incidents. They are the visible tip of an industry-wide pattern in which the people who get hurt the most — women, BIPOC employees, junior staff, contractors — are also the people with the least power to push back.

Credits. A professional standard exists in film: every person who worked on the movie, including the contractors, is in the credits. Game industry standards are looser. Contractors are routinely omitted. People who left the studio before launch are routinely omitted. People who built the game's foundational systems but were not on the team at ship are omitted. Several years' worth of designers' careers can be invisible because they shipped under contract or left during development. The omission has career consequences — your portfolio is what other people can verify you worked on.

Unionization. The countervailing force is collective. Game workers have been organizing. The CWA's CODE-CWA initiative has supported unionization efforts at multiple studios. The first U.S. game studio union of QA testers was certified at ZeniMax (Microsoft) in early 2023. Workers at Activision's Blizzard Albany studio organized. Multiple studios have followed. Unionization is slow, contested, and management-resisted, but it is the durable path to changing the labor conditions that produce the industry's worst behaviors.

This is a chapter on game-design ethics. Why labor? Because the same studio that grinds its developers will grind its players. The cultural defaults are the same. The management mindset is the same. A designer who is exhausted, demoralized, and worried about being fired will not push back on the dark pattern in the next sprint. The conditions of design ethics start in the meeting room where the decision gets made; they are downstream of the conditions of work in the building where the meeting happens.

🛠️ Practitioner Tip: Pay attention, when interviewing, to questions about crunch, harassment policies, contractor practices, and credits. The studio's answers will tell you a lot about whether you will be able to do ethical design work there. Studios that say "we never crunch" are sometimes lying; studios that say "we sometimes crunch but here is what we do to avoid it" are usually being honest. The honest answer is the one to look for.

Environmental Ethics

Games have a carbon footprint. It is not the largest carbon footprint of any technology sector; it is also not zero, and it is growing.

Three sources matter.

Streaming and downloads. A 100GB game download, multiplied by 10 million players, is 1 exabyte of data over the network. Cloud gaming services (Stadia, GeForce Now, Xbox Cloud Gaming) shift the rendering load from the player's hardware to data centers, which require constant power and cooling. The carbon footprint of a streamed game is meaningfully larger than the same game played locally on the same hardware once downloaded.

Always-online services. A game with always-on servers (every MMO, every live-service shooter, every gacha game) requires datacenter capacity that runs 24/7 for the game's lifetime. The carbon cost is small per player but multiplies across years and player counts. When Battlefield 2042 keeps servers running for 8 years for a steadily-shrinking player base, the per-player carbon cost rises.

Crypto and NFT games. Ubisoft Quartz, launched in December 2021, was Ubisoft's experiment in NFT-based cosmetic items in Ghost Recon Breakpoint. The reaction was a near-universal industry rejection. The Quartz items required substantial blockchain transactions, the energy footprint of which (especially on the proof-of-work chains that dominated at the time) was orders of magnitude higher than equivalent in-game items. Ubisoft quietly shut down Quartz in 2023. Most major publishers have backed away from NFT integration. A subset of the industry has not — "play-to-earn" and "Web3" games persist, mostly in markets and demographics where the regulatory and cultural pushback is weaker.

The environmental ethics here are not the largest issue this chapter covers, but they are not nothing. Designers can:

• Optimize game install size; resist the trend toward 200GB AAA installs that are mostly uncompressed assets.
• Build offline-playable games where the design admits it.
• Sunset always-online services responsibly when they are no longer viable, rather than running them at low utilization for years.
• Refuse blockchain integration where it is being added for hype rather than function.

These are small, but the small things add up.

Building an Ethical Design Practice

This chapter has named a lot of problems. The temptation, at this point, is either to shrug ("the industry is what it is") or to despair ("I cannot fix this from where I sit"). Both responses are wrong. Most ethical design happens in small, daily decisions made by designers who are paying attention.

A practice — a set of habits — is what produces ethical design over time. Here is a practice you can adopt, tomorrow, in your next design meeting.

Ask: who am I designing for? Not "who is the target demographic." Who is the actual person — what are they doing in their day, what are they hoping for from this experience, what would a good outcome look like for them. Hold that person in mind. Their wellbeing is the test.

Ask: who am I extracting from? Where is the revenue coming from? What is the median spend, the mean spend, the top-1% spend? If the top 1% is funding the game, what is happening to those people? Are they spending money they can afford? Or money they cannot?

Ask: what would this look like on the front page of a newspaper? This test has limitations — many ethical violations would not be newsworthy because they are too common — but it surfaces the cases where you would feel embarrassed by the public version of your decision. If you would not want to read the headline, do not make the decision.

Ask: what is the worst-case outcome for a vulnerable player? Not the average player. The 14-year-old who has access to a parent's credit card. The adult with a gambling history. The lonely player using your game to fill the absence of human contact. What does your design do to that person? Is it acceptable?

Ask: would I let my child play this? A useful test if you have children, or know children. The answer is sometimes yes; the answer is sometimes no; the question forces specificity.

Ask: am I being asked to design something I would not freely choose to design? The honest version of this question separates the cases where you have a real ethical objection from the cases where you are uncomfortable for other reasons. If the answer is yes, the next question is what you do about it.

These questions are not a checklist; they are a practice. Run them every week, on the work you are doing. Most weeks, the answers will be fine. Some weeks, the answers will be uncomfortable. The uncomfortable weeks are the ones where the practice matters.

⚖️ Ethical Framework: A useful frame is consent. Ethical engagement systems work because the player has been given enough information to consent to participating in them, and the design respects the consent the player has given. Exploitative systems work because they bypass the player's capacity to consent — they hide the cost, exploit a moment of weakness, or condition the player to behave against their own interest. "Could the player have meaningfully consented to this if they understood what it was doing?" is a strong test.

Progressive Project Update

For the action-adventure game you have been building since Chapter 3, do an ethics audit and write an ethics statement.

Step 1: Audit. Run your current build through this checklist. For each item, score yourself: green (clearly ethical), yellow (defensible but worth examining), red (problematic).

• Monetization. Are there in-game purchases? If yes, what kind (cosmetic, convenience, power)? Are odds disclosed if random? Is it clear what currency translates to what real money? Are children able to spend without an adult?
• Time pressure. Are there limited-time events? Daily login streaks? FOMO mechanics? Can a player who plays once a week still meaningfully progress?
• Difficulty and accessibility. What accessibility options does the game offer? Is there a difficulty option, an assist mode, or a way for players who cannot complete twitch challenges to experience the story?
• Representation. Who is the protagonist? Who is the antagonist? Who are the NPCs? What demographic patterns appear in your character roster? Are the non-default characters written as full characters or as tokens?
• Multiplayer (if applicable). What anti-harassment infrastructure exists? Can players block, mute, report? What happens to a report?
• Data. What does the game collect? Who has access to it? Is the privacy policy readable?
• Labor. This is about you and your team — including the friends you might pull in for art, music, or playtesting. Are people being credited? Compensated? Treated well?

Step 2: Ethics statement. Write a one-page document — single side, around 400-600 words — that covers three things.

What you did. The ethical design choices you made: assist mode, no random monetization, accessibility options, etc.

What you won't do. The lines you have decided not to cross: no loot boxes, no FOMO mechanics, no targeted ads, etc.

What you're uncertain about. The honest version of "I don't know yet." Maybe you are unsure how to handle multiplayer toxicity. Maybe you are not sure whether to include a difficulty option. The honesty is part of the ethics.

This document should ship with the game, on its itch.io page or website, as a public commitment. The act of writing it down — and showing it to other people — is what makes it a constraint. Designers who do not document their ethics tend not to keep them, because the pressure of release windows erodes everything that is not written down.

Save the file as ethics-statement.md in your project folder. We will reference it in Chapter 38 when you build your itch.io page and again in the Ch 40 capstone.

When You Disagree with Your Employer

You will be asked, at some point, to design something you do not want to design. This section is about that moment.

The realities first. Most game-industry jobs are precarious. Many of you reading this are on visas tied to your employment, are paying student loans, are supporting families, are working through a probationary period, or are otherwise not in a position to walk out of a meeting and never come back. The advice "just refuse" is cheap if you have not had to give it from inside the actual situation. The honest advice is more complicated.

A staged response, in roughly increasing order of cost:

Push back internally. State the concern. Be specific about the harm you see. Offer an alternative. Most ethical violations in design are not made by villains; they are made by managers who have not thought through the consequences and who will adjust if a respected colleague raises a substantive objection. The first move is always the internal push-back.

Document. Write the email. Send it to your manager. Keep a copy. The documentation matters for two reasons: it forces specificity (you have to articulate the concern in writing, which clarifies it), and it creates a record (if the project goes badly later, the documentation establishes that someone raised the issue). The documentation is for the present and for the historical record both.

Escalate. If your manager is the problem, escalate to skip-level or to ethics infrastructure if your studio has any. (Many do not; some do.) Escalation has career cost; it also has career cost not to do it, in cases where the issue is serious.

Refuse the specific work. "I cannot work on this feature" is a stronger move than "I disagree with this feature." It puts the burden on the studio to find someone else, or to drop the feature. It also signals seriousness. In many cases the studio will reassign you and proceed with the feature anyway, but the signal matters.

Leave. The last resort. Sometimes the right one. A studio whose direction you cannot work with is a studio that will use up your career on work you do not believe in. Leaving is sometimes the only ethical option remaining.

Speak publicly. Sometimes — rarely — the situation warrants going public after leaving. The cost is large and the protection minimal. NDAs constrain what you can say. The industry retaliates against public whistleblowers. People who have done it have paid for it. People have also done it anyway and changed the conversation. The decision is yours.

What I want you to take from this section is permission. You are allowed to push back. You are allowed to refuse work. You are allowed to leave. The industry will not tell you you have these options; you have them anyway. The career cost of using them is real, and the career cost of not using them — of becoming someone who designs things you are ashamed of — is also real, and is ultimately larger.

Common Pitfalls

Treating ethics as marketing. The studio publishes a "values statement" full of words like "respect" and "inclusivity" and uses it as a recruiting tool while shipping products that do the opposite. This is not ethics; it is branding. Real ethics costs money. If your studio's values statement has not constrained any product decision, it is not a values statement.

Checklist ethics without culture. A studio adopts a "responsible design" checklist that all features must pass. The checklist is filled out, signed, filed. The product still ships predatory mechanics, because the checklist is a paperwork ritual rather than a design constraint. Checklists are useful only when they have teeth — when items can be flagged red and the feature is actually killed.

Assuming your game is different. "Our loot boxes are fine because we make sure the items are achievable through play." "Our gacha is fine because we have generous pity." "Our daily login is fine because the rewards are small." Every studio in the industry has some version of this sentence. It is almost always self-serving. The patterns are the patterns; the local justifications do not change what the patterns do to vulnerable players.

Ignoring kid players in adult games. The game is rated M; therefore the design does not need to consider minors. In practice, kids play M-rated games — GTA, Call of Duty, Fortnite (which is rated T but functionally a kids' game). If your game can be acquired by a minor, your design has obligations to that minor. The rating absolves you legally, not ethically.

"It's the market's fault." "We have to do this because our competitors are doing it." "Our players want this." "If we don't make money, we can't make games." These are the standard moves. Each one is true in some narrow sense; none of them is exculpatory. The market is not a thing that happens to you; you are the market. The competitors are other designers making other choices. The players want what they want partly because the industry has trained them to want it. The financial pressure is real; it is also the same pressure every industry uses to justify its worst behavior.

Conflating "the player kept playing" with "the player was happy." Engagement metrics measure presence. They do not measure satisfaction. They do not measure wellbeing. They do not measure regret. The most engaged players are sometimes the most miserable players — players who cannot stop playing a game they no longer enjoy. Telemetry will not surface this; only player conversation will. Talk to your players, including the ones who quit.

Summary

The ethics of game design is the craft of game design with the consequences taken seriously. Every design choice is a choice about what to do to a person, and you are responsible for the choice.

Engagement and exploitation look identical from the analytics dashboard but are different in kind. Engagement is the player wanting to come back because the game adds to their life. Exploitation is the player coming back because the game has installed a compulsion that operates beneath their wanting. The ethical designer's work is to build for engagement and to refuse the patterns that produce exploitation, even when those patterns also produce revenue.

The major patterns to know: variable-ratio reinforcement (the operating principle of every loot box and gacha system), FOMO (limited-time content that exploits loss aversion), sunk-cost (login streaks that punish departure), bait-and-switch (free entry, paywall later), currency obfuscation (multiple currencies that hide real-money cost), pay-to-skip-the-pain (intentionally tedious systems with paid bypass). These patterns are not accidents. They are the result of decades of A/B testing optimized against the vulnerable player at the tail of the spending distribution.

Accessibility and representation are ethical surfaces, not optional features. The artist's vision and the player's right to engage with it are not in opposition — Celeste's assist mode demonstrates that they coexist. Representation is a design choice in protagonist, not a checkbox in NPC count. Both are made hard by industry defaults inherited from a prior demographic moment; both are getting better, slowly, against pushback.

Labor ethics, environmental ethics, data ethics, and harassment ethics are all part of the design ethics package because the conditions of production shape the product. A studio that exploits its workers will exploit its players. A studio that ignores harassment of its developers will ignore harassment of its players. The ethics of the industry is one ethics, not several.

Finally: you will be asked, sooner than you think, to do something you do not want to do. Have your responses ready. Push back, document, escalate, refuse, leave — in roughly that order, depending on the severity. The career cost of resisting is real. The career cost of becoming a designer who ships things they are ashamed of is larger. Build the practice now, before the meeting.

Next chapter (Ch 34) we look at the business models that drive these design choices, and at how an industry built around different models would produce different games. Chapter 38 covers marketing ethics specifically, because how a game is sold is part of the same problem as how it is built. The ethics conversation does not end with this chapter; it threads through everything that comes after.

What I want you to leave this chapter with is a sentence you can say in a meeting: I don't think we should ship that, and here's why. The work of game-design ethics is largely the work of being someone who can say that sentence, when it needs to be said, in the room where the decision is being made. Build the muscle. The medium needs more designers who have it.