Module 6 · Mobile / F2P (2012–2018)

Gacha and the battle pass

Two engines monetize a retained loop in different ways. Gacha sells a random reward — variable-ratio reinforcement, the psychology of a slot machine. The battle pass sells seasonal progress — sunk cost plus FOMO, which pumps revenue and retention at the same time.

~17 min🏠 an economy lab💰 monetization + 🎲 psychology

The gist in 30 seconds

A retained loop is monetized by two machines. Gacha: pay currency → get a random reward from rarity tiers. It hooks through variable-ratio reinforcement (an uncertain reward is more compulsive than a fixed one — Skinner); the money comes from whales (1 whale = $1000+, a 90/9/1 power-law tail); the expected number of pulls before the prize is

1 / p

, and pity (a guarantee at N) cuts off the tail of bad luck. The battle pass: a ~$10 seasonal track of cosmetics unlocked by playing; sunk cost plus FOMO plus progress deliver +$5–10/month per player while also holding retention (challenges = an appointment mechanic). Cosmetics-only (Riot/Epic, "look cool, don't be cool") versus pay-to-win (×3–5 revenue per player, but it kills competitive fairness). Regulation: loot boxes ≈ gambling (Belgium 2018 — they were pulled, China 2017 — odds disclosure, ESRB — a label), and the Battlefront II failure in 2017 produced the most downvoted comment in Reddit history (~668K).

The mechanism: randomness and seasonality

Gacha — selling uncertainty

Gacha (from Japanese capsule machines): spend currency → get a random item from tiers (Common → Rare → Epic → Legendary). What hooks you is not the item but variable-ratio reinforcement: not knowing the outcome activates the same dopamine mechanism as a slot machine and holds harder than a fixed reward. The probability of getting what you want at least once in $n$ pulls:

P (\geq 1 in n) = 1 - {(1 - p)}^{n}

With a tiny $p$ the tail of bad luck is long — and that is where pity comes in: a guaranteed rare after N pulls. Genshin is the canonical example: a base 5★ chance of 0.6%, soft pity around 74 (the chance rises sharply), hard pity at 90 (guaranteed). Without pity, $P (\geq 1 in 90) = 1 - {0.994}^{90} \approx 42 %$ — nearly half of players would walk away with nothing; pity makes the 90th pull 100%, capping the variance: "by 90 you WILL get it". Pity is both a softening of predation and a retention device (the ceiling on bad luck is visible).

The battle pass — selling seasonal progress

The battle pass (popularized by Fortnite, season 2, Dec 2017; the format itself comes from the Dota 2 Compendium, 2013): you buy a pass (~$10) → play challenges → unlock 100 "tiers" of cosmetics → the season ends (≈10 weeks) and whatever you did not claim is gone forever. Four hooks at once:

Sunk cost: you paid $10 → you feel obliged to "get your money's worth" by finishing the pass.
FOMO: the cosmetics are time-limited — miss the season and you miss them forever.
Progress: daily challenges give direction instead of aimless grinding.
Social pressure: your friends are leveling their pass, and falling behind means falling out of the shared thing.

The brilliance is that a battle pass is monetization and retention at once: the challenges work as an appointment mechanic (log in every day, don't lose the season), and the pass itself brings in +$5–10/month per player. Which is why everyone adopted it after Fortnite (Overwatch 2, Valorant, Apex, Halo Infinite).

Cosmetics versus pay-to-win

The fork: what exactly you sell.

Cosmetics-only (Riot, Epic): skins and emotes that do not affect power. "If you pay, look cool, don't be cool." The upsides: a social signal (a skin is visible), no pay-to-win resentment, no rebalancing needed, streamer-friendly. LoL is cosmetics-only, with ~180M quarterly players and $2B+/year.
Pay-to-win (many mobile games): you sell power (equipment, stats, exclusive units). ×3–5 revenue per player, but the downside is real: a newcomer cannot compete, endless rebalancing is required, and it is toxic to the competitive core.

The choice = the audience: a competitive game has to be cosmetics-only (otherwise it loses its core and its streamers), while a casual gacha RPG lives on "power for money". The middle ground (Clash Royale) is a hybrid: mostly cosmetics plus a slight advantage from levels.

🕹 What to play — and what to notice

You can see the mechanics in the shop and on the pull screen within a minute: look for disclosed odds, a pity counter, a season timer and the line between cosmetics and power.

Genshin Impact gacha + pity 0.6% / 74 / 90

The reference implementation of modern gacha: "wishes" with odds disclosed at China's insistence, soft/hard pity and a 50/50 "guarantee" system. The theatrics of a pull (the animation, the colored beam of rarity) are variable reward in action.

🎮 Play: open the wish screen and find the odds disclosure (0.6% for 5★) and the counter toward pity. Notice the pull animation — the delay and the color of the beam exist to maximize the dopamine peak of uncertainty. That is design, not decoration.

Fortnite 2018 · the original battle pass

The very pass that redefined the industry: ~$10, 100 tiers, a season timer, cosmetics only. The challenges pull you in daily, and the season timer applies FOMO pressure.

🎮 Play: look at the battle pass screen — notice the countdown to the end of the season (the FOMO clock) and the daily challenges (appointments). Ask: what here pulls you to come back tomorrow, and what pulls you to pay? That is retention and monetization on one screen.

Star Wars: Battlefront II 2017 · a pay-to-win disaster

A textbook case of how NOT to do it: power progression through loot boxes in a full-price $60 game. EA's response on Reddit became the most downvoted comment in history (~668K), forced paid loot boxes to be cut at launch and accelerated the wave of regulation.

🎮 Watch: read a breakdown of the 2017 Battlefront II scandal. Notice the trigger for the rage — not "loot boxes" in general but power for money in a premium game: a violation of competitive fairness the core will not forgive. The contrast with cosmetics-only is night and day in terms of reaction.

Deep end · economics: the whale curve, gacha expectation and the arithmetic of pityskippable

F2P money is extremely uneven — a power-law tail (see F2P economics): roughly 90/9/1 (90% never pay, 9% are "dolphins", 1% are "whales" and supply almost all the revenue). Gacha is a machine for extracting from that tail: one whale pulls for $1000+.

Expectation and variance

The expected number of pulls before the first prize is $1 / p$ for a geometric distribution (at $p = 0.006$ that is ~167 pulls). But the variance is enormous: one person gets it on the 5th, another still has nothing at the 150th. Pity cuts off the right tail: a hard guarantee at 90 turns "maybe $300, maybe $1000" into "no more than N" — which is both about ethics (a bounded maximum of pain) and about conversion (a visible ceiling → pulling stops feeling scary).

Battle pass arithmetic

The pass is cheap ($10) and works through volume plus top-ups: if a share $q$ of the retained base buys it, revenue per player = $10 \cdot q$ per season, plus "pay to unlock the remaining tiers before the season ends" collects the FOMO tail. Unlike gacha it does not depend on whales and does not provoke pay-to-win resentment (it is cosmetics), which is why it became the standard: predictable mass revenue plus retention in one mechanism.

Deep end · psychology and regulation: Skinner, FOMO and "a loot box is gambling"skippable

Why a variable reward is stronger

Variable-ratio reinforcement (a reward after a random number of attempts) produces the most persistent and compulsive response in Skinner's experiments — stronger than a fixed schedule. The brain cannot "count down" to the reward, so it keeps going. Gacha commercializes exactly that; the theatrics of the pull amplify the peak of uncertainty.

FOMO and sunk cost

The battle pass stacks two biases: sunk cost (I paid, so I have to get my money's worth) and FOMO (it will be gone forever). Both push toward "finish it / top it up", even when the enjoyment is no longer worth the time.

The regulatory patchwork

Paid loot boxes are gambling in form (you pay for a random outcome of varying value): Belgium 2018 — the gaming commission ruled against them and publishers voluntarily pulled paid loot boxes; China 2017 — mandatory drop-rate disclosure; ESRB — a "Loot Boxes" label. There is no outright global ban — there is patchwork regulation and self-regulation (pity, odds disclosure) in response to pressure. The ethical core: targeting minors, hidden odds, exploiting the vulnerable.

Analogy

Gacha is a slot machine with a guaranteed ceiling on the jackpot: you pull the lever for the uncertainty (dopamine on every "almost"), but pity promises "by the 90th it will definitely drop", turning a casino into a casino-with-a-guarantee. The battle pass is a prepaid gym membership: you paid up front and now feel obliged to "walk off the money" by showing up every day (sunk cost), until the membership expires at the end of the season (FOMO). Both machines sell not an item but a psychological state — uncertainty and obligation.

Why it matters

Gacha and the battle pass are the two dominant ways to turn retention into money, and both run on psychology rather than the value of the item: variable rewards and FOMO/sunk cost. Understanding the mechanics (probabilities, pity, the whale tail, pass conversion) is mandatory if you are going to build or analyze live games. But it matters even more to see the ethical line: the same levers that raise engagement slide easily into predation. This is direct practice for "optimizing a metric against the user's wellbeing" — the argument waiting for you in any engagement system.

🔁 Beyond games — where this transfers

The lesson is monetization through designed psychology: variable rewards, uncertainty, FOMO — and the ethical line around them.

ML / AI (your domain): gacha's variable-ratio reinforcement is literally a schedule of reinforcement from RL, and it is the same dynamic that engagement-optimizing recommenders exploit (an infinite feed = variable-ratio dopamine). The math of gacha ( $1 - (1 - p)$ raised to a power, an expectation of $1 / p$ , pity trimming the tail) is the same expectation/variance you deal with in sampling and exploration. The key part is Goodhart and dark patterns: A/B-optimizing a revenue or engagement metric drifts on its own toward predation (the model finds what works predictively without asking whether it is ethical) — precisely what explicit reward design and human constraints exist to prevent. Knowing where NOT to maximize variable reward is part of professionalism, not an optional extra.

Product / growth: sunk cost and FOMO are standard growth levers (limited-time offers, streaks, progress bars); cosmetics-only vs pay-to-win = monetization that does not break the core experience versus monetization that does.

Finance / behavioral economics: a loot box ≈ a lottery; variable rewards, loss aversion and sunk cost are the same biases that operate in gambling and trading.

The principle: the most powerful engagement mechanics run on cognitive biases. Learn to build them and do their math — and learn to draw the line past which "engagement" becomes exploitation.

🏠 Lab — do the math on gacha and whales

The interactive F2P economy lab already contains a gacha calculator (a slider for the chance

p

→

1 / p

and

P (\geq 1)

over 10/50/100 pulls) and the 90/9/1 whale curve, plus LTV/ARPDAU/retention. Sweep the chance and watch pity "trim the right tail". Open the lab →

Best moment: set $p$ = 0.006 and compare $P (\geq 1)$ over 50 versus 100 pulls — you will see why, without pity, the unlucky tail is so long that a guarantee is required.

🔧 Run it and poke at it — on your home machine

What to play is above (🕹). This part is about taking the monetization apart numerically.

🔧 Poke at it (deconstruction) ~40 min

Take a gacha game and write down: the top-rarity chance, the pity rules, the currency packs in dollars. Compute the real expected cost of the "guarantee" (pulls to pity × price per pull). Then take somebody's battle pass apart: price, season length, cosmetics vs power, where the "pay to unlock early" button sits. Classify the game: cosmetics-only / hybrid / pay-to-win.

🧪 Test it (with ethical-designer eyes) ~15 min

For each of the game's mechanics answer: does it "help the player get what they value" or "create pain in order to sell the relief"? Where are the odds disclosed and where are they hidden? Are there protections for minors or spending limits? Sketch how you would make the same monetization less predatory without losing revenue.

Checklist: computed the expected cost of the pity guarantee; took a battle pass apart along price/duration/power; classified the monetization; ran an ethical audit of the mechanics.

Connections

foundation

The core loop and retention — gacha and the pass monetize a retained loop; with no retention there is nobody to monetize.

foundation

F2P economics — the 90/9/1 whale curve, LTV/ARPDAU and the gacha calculator itself (the lab).

contrast

Virtual economies — there the economy is player↔player (emergent); here it is designer↔player (extractive monetization). Two sides of "money in games".

Analytics — how prices, odds and offers get A/B-tuned (and where that turns into dark patterns).

Questions worth asking

Why is a variable reward more addictive than a fixed one if the expectation is the same?

Because what hooks you is not the mean but the uncertainty. Variable-ratio reinforcement (a reward after a random number of attempts) produces the most persistent response in Skinner's experiments: the brain cannot predict when the reward is coming, so it never "settles" and keeps going — every "almost" delivers a dopamine spike of anticipation. A fixed reward is predictable and the anticipation fades. Gacha commercializes exactly that unpredictability; an equal expectation does not mean equal compulsiveness.

Does pity make gacha ethical — or is it just a shop window?

Honestly, both. Pity really does bound the maximum pain (there is a ceiling on spending before the guarantee) and lowers the variance — a genuine benefit to the player and protection against a bottomless pit. But it also raises conversion: a visible ceiling removes the fear of "what if I spend $1000 and get nothing", so people pull more readily. So pity simultaneously softens predation and optimizes revenue — it does not make the mechanic "not gambling", it just puts boundaries on it. Whether it is ethical also depends on odds disclosure, targeting of minors and spending limits, not only on the existence of a guarantee.

Why is a battle pass both retention and monetization rather than one of the two?

Because one mechanism covers both goals. The pass's challenges are an appointment mechanic: they give a reason to log in daily (retention), a direction instead of grinding, and a fear of losing the season. And the pass itself ($10 plus top-ups for early unlocks) is revenue — mass and predictable revenue at that, not whale-dependent the way gacha is. Sunk cost and FOMO make you both play (retention) and pay (monetization). That is why it displaced loot boxes in many games: it brings money without pay-to-win resentment and holds players at the same time.

Cosmetics-only leaves money on the table (P2W gives ×3–5). Why doesn't Riot sell power?

Because for a competitive game, power-for-money destroys the product itself. Pay-to-win kills competitive fairness → the core leaves (there is no point competing if a wallet decides) → the streamers leave (P2W is toxic for content) → the ecosystem that LTV rests on collapses. Cosmetics-only gives up ×3–5 short-term ARPU for a sustainable mass base (LoL: $2B+/year without selling power). It is a choice between "extract more per player, but fewer players and not for long" and "less from each, but 180M for a long time". For a genre that lives on competition and spectatorship, the second strictly dominates.

Is a loot box gambling? If so, why isn't it banned everywhere?

In form, yes: you pay for a random outcome of varying value, as in a lottery. But legally, "gambling" usually requires a payout of real money (a cash-out), and game items are not officially cashable — the industry's defense rests on that technicality. So regulation is patchwork: Belgium (2018) ruled against them and forced paid loot boxes out, China requires odds disclosure, the ESRB applies a label, but there is no global ban. The pressure produced self-regulation (pity, drop-rate disclosure). The argument continues, especially around targeting children; "no cash-out, therefore not gambling" is an increasingly shaky position.