The Economics of Gambling

When someone loses money, the explanation is always roughly the same: they lacked willpower. They got hooked, carried away, couldn't walk away in time. Aristotle called this state akrasia—acting against your own understanding of what's best.

It's an ancient explanation, and honest in its own way. It's also extremely convenient for the industry: the entire problem is placed inside the individual. Weak willpower—so it's the willpower that needs fixing.

Yet behind the slot machine, the betting app interface, and the casino floor layout aren't abstract temptations at all. There are several decades of applied science: reinforcement theory, neurobiology of reward, behavioral economics, user experience design. The industry isn't selling luck—it's selling the time a person spends playing. And it's learned how to stretch that time.

The scale has long ceased to be niche. The Lancet Public Health Commission on gambling estimates net player losses—the difference between what's wagered and what's paid out—at nearly $700 billion annually by 2028. Gambling in one form or another is legal in more than 80% of countries. Over the past year, 46.2% of adults and 17.9% of adolescents worldwide have gambled.

A seventy-year-old invention

The industry's main tool was invented in a laboratory.

In the 1950s, B.F. Skinner described reinforcement schedules—the rules by which a reward follows an action. It can come after a fixed number of presses, after a fixed interval, after a random interval, or after a random number of presses. The last schedule, with a variable ratio, turned out to be abnormally powerful. Behavior reinforced by it almost never extinguishes: the animal keeps pressing the lever long after the reward has stopped coming.

The reason is that an empty attempt communicates nothing. It doesn't mean "they're not giving anymore"—it means "not yet." The next one could still hit.

Skinner himself cited gambling as a textbook example of this schedule. He's also credited with claiming he could turn a pigeon into a pathological gambler.

The reinforcement schedule is a characteristic of the environment, not the player. It's set from the outside by the machine's designer, and it works roughly the same on a pigeon, a rat, and a person with two graduate degrees. A slot machine is literally a device for implementing this scheme: a win is possible on any spin, but you don't know which one. That's why a losing streak sustains the behavior.

Everything else is superstructure built on this foundation.

"Near miss"

The most discussed technique is the "near-miss" effect. An outcome that looks like a win but isn't: two matching symbols out of three, a reel frozen one notch away from the jackpot.

In 2009, a team led by Luke Clark at Cambridge placed test subjects with a slot machine simulator inside an fMRI scanner. It turned out that "near-misses" activate the same reward-related brain structures as actual wins. Yet the participants themselves rated such outcomes as more unpleasant than ordinary losses—and felt a stronger urge to keep playing.

This discrepancy is worth dwelling on. It undermines the standard model of humans as creatures that seek pleasure.

Back in the 1990s, Kent Berridge and Terry Robinson proposed separating two processes that everyday language conflates: "liking"—the hedonic response—and "wanting"—the motivational salience of a stimulus. They're distinct not just conceptually but neurochemically. The "near-miss" is a textbook case: you don't like it, but you want more. The industry needs the latter.

Pharmacology confirmed the hunch. In a 2016 study, 22 people with gambling addiction and 22 controls performed the same task twice—once under placebo and once under a dopamine D2-receptor blocker. In those with addiction, the response to "near-misses" was amplified, and intervention in the dopamine system altered it.

A caveat that popular accounts usually omit: the effect doesn't always replicate. In a 2020 review with three original experiments—on pigeons and humans—the authors found no confirmation. The mechanism is more likely real than not, but the debate over its strength remains open. This is common in gambling literature: a clear signal at the brain level and a murky picture at the behavioral level.

There's far less debate about another technique—"losses disguised as wins." On some machines, a combination registers as a win, triggering sound and animation, but the payout is less than the bet placed. Formally, the player won. In reality, they lost money. The brain responds to the celebration, not the arithmetic.

The metric everything is built around

Industry engineers have a metric from which nearly all design decisions flow: time on device.

Anthropologist Natasha Dow Schüll spent fifteen years doing fieldwork in Las Vegas—from slot machine manufacturers' conventions to Gamblers Anonymous meetings. Her main finding is counterintuitive: slot machine regulars don't play to win. They play to enter a state they call "the zone." A trance where time, money, and the sense of one's own body disappear. Winning is more of an interruption here; it breaks the rhythm. The point is to continue.

If this is true, the familiar schema of "people risk for the sake of winning" is wrong at its very foundation. They play for the process, in which the outcome is an annoying pause. The industry's economics are built on the customer's desire to disappear.

The architecture of the gaming floor serves the same metric. Bill Friedman, former manager of two casinos on the Las Vegas Strip, analyzed the layout of more than eighty Nevada establishments and derived the principles of "gaming design": machines right at the entrance, labyrinthine floor plans, low ceilings, no clocks or windows. That very set everyone's heard about.

Then things get interesting. Over the past thirty years, the industry has largely moved away from this canon. Designer Roger Thomas did the opposite: high ceilings, natural light, clear navigation—and antique clocks right in the gaming hall. The model became known as "playground design," and it matches the classic approach in visitor retention. In a comfortable space, people simply tire less and stay longer voluntarily.

Two opposing architectural philosophies yield the same result. Manipulation, it turns out, doesn't have to be repressive.

The Illusion of Control and the Cost of What's Already Lost

Next come cognitive biases that psychology studies independently of casinos.

The "illusion of control"—the tendency to overestimate one's influence on random outcomes—was described by Ellen Langer back in 1975. The "gambler's fallacy"—the belief that after a series of losses a win must come, even though the probability of each subsequent event has no memory of previous ones.

Both biases aren't merely tolerated in casinos—they're actively fed. Roulette wheels invariably display boards showing the history of numbers that have come up: mathematically useless, psychologically—fuel. Slot machines let you choose lines and "stop" the reels. It has zero effect on the outcome, but it sustains the sense of authorship.

The third mechanism is the sunk cost trap. The more money, time, and effort invested, the harder it is to walk away: leaving means admitting what was invested was wasted. Japanese researchers compared 32 men with gambling disorder and 37 people in a control group using fMRI. The former showed altered sensitivity to sunk costs—from there it's a straight path to "chasing losses," attempting to win it back by increasing bets.

There's a nuance here that's often overlooked: the trap hits hardest those who consider themselves experts. A poker player or bettor who's spent years studying statistics perceives leaving the game as admitting incompetence. The stake is no longer about money—it's about self-definition.

The Casino That Lives in Your Pocket

Everything described was designed for physical venues. The smartphone hasn't changed the essence, but the density of impact.

In a report by the Lancet commission, certain features of betting apps are explicitly called "dark patterns"—interface techniques that exploit cognitive biases to the user's detriment. An audit of ten popular British apps conducted by the Behavioural Insights Team documented, among other things:

• registration without functioning age verification;
• default deposit amounts and odds are noticeably higher than the minimums—a classic anchoring effect;
• one-click betting and odds selection;
• self-limitation tools buried deep in the menu;
• minimum account balance requirements for withdrawals;
• a new bet offer popping up immediately after the previous one;
• no on-screen display of the amount lost during a session;
• notifications creating urgency around "expiring" bonuses;
• accounts that are difficult or impossible to close permanently.

Meanwhile, the mechanics have accelerated. Sports betting was historically a slow-burn form of gambling: place your bet on the outcome—then wait ninety minutes. Today there are real-time bets on micro-events (who will take the next corner kick, will the player make the free throw), parlays combining a dozen such micro-events, and "cash out"—early closure of a bet, a decision made in seconds and driven by emotion. Psychologist Jamie Torrance from Swansea University puts it this way: sports betting isn't yet as harmful as a slot machine, but it's moving in that direction.

There's also something that offline casinos never had at all—data. The app sees when a person bets, on what, how much, after which results, at what time of day. This forms the basis for deciding which offer to send them. Heather Wardle, who chaired the commission, suggests imagining if tobacco companies knew about every cigarette smoked: when, how many, under what circumstances.

The practical result of such personalization looks mundane. Someone who decided to quit two weeks ago receives a push notification with "ten free bets, today only."

In a 1971 work, Harry Frankfurt described the figure of the unwilling addict. He has a first-order desire—to take the substance—and a second-order desire not to want to want it. Will, according to Frankfurt, is determined precisely by the second level; addiction is a situation where the first-order desire wins out despite who the person wants to be. The notification is addressed precisely to the first level and arrives at a moment calculated from data about when resistance is weakest. Formally, freedom of choice remains intact: no one is forcing you to click. In reality, the offer is constructed to bypass the very faculty that makes a person the author of their own decisions.

A separate story involves loot boxes in video games—paid bundles with random contents. A survey of 7,422 gamers published in PLOS ONE found a link between spending on loot boxes and symptoms of problem gambling, one notably stronger than with regular in-game purchases. The key factor is randomness. Belgium and the Netherlands already regulate loot boxes as a form of gambling. The direction of causality hasn't been established, but from an age-restriction standpoint the difference is minimal: the "pay-and-try-your-luck" mechanic is learned long before adulthood.

Who's Really Paying for the Party

This brings us to the main point—how revenue is actually structured.

The industry's public rhetoric is built around the image of a mass casual customer: millions betting small amounts, someone wins, everyone's happy. The data tell a different story.

Heather Wardle and colleagues analyzed the spending of participants in a British longitudinal study of regular bettors. The distribution turned out to be extremely uneven: a Gini coefficient above 0.70 for most types of games. The 14.1% of participants showing signs of problem gambling accounted for 43.5% of all gross spending, and for online casinos—over 80% of revenue. For lotteries, dependence on this group is noticeably lower.

The picture is even starker in research by Gemini Research, commissioned by Connecticut's Department of Mental Health and Addiction Services for $1.2 million. Gaming problems affect 1.8% of state residents, around 50,000 people. They account for 12.4% of lottery revenue—and 51% of sports betting revenue. Across all forms of legal gambling—21.5%. And if you add the "at-risk" category (another 4.9% of the population), then fewer than 7% of state residents generate about 71% of all legal gaming revenue.

The Lancet commission looks at it from the product side and arrives at a consistent result. Gaming disorder may affect 15.8% of adults and 26.4% of adolescents who play online casinos and slots, and 8.9% of adults and 16.3% of adolescents among those betting on sports. In absolute numbers—around 80 million adults with gaming disorder and approximately 448.7 million people experiencing gambling-related harm.

Three independent datasets converge on one point: the fastest-growing products are those most dependent on players with disorders. This is the load-bearing structure of the business model.

The Russian Picture

In Russia the same processes are unfolding, adjusted for the pace of regulation.

The annual study by Reyting Bukmekerov, released in spring 2026, estimates gross gaming revenue for the entire industry at roughly 1 trillion rubles—up 12.6% from the year before. Of that sum, 598.3 billion comes from the illegal segment, meaning nearly 60% of the market sits outside the regulated zone. And within the shadow portion, 418.8 billion comes from online casinos and only 179.5 billion from bookmakers.

Then things get really interesting. At least one bet in 2025 was placed by 13 million Russians—virtually the same as the year before, growth of just 0.3%. The number of new customers, meanwhile, plunged 34.6% to 1.7 million. But active players, those placing bets at least once a week, rose 9.6% to 6.9 million people. The average bet increased from 1,750 to 1,890 rubles.

In other words, the market has stopped expanding outward and started growing deeper.

The players themselves confirm this shift. The share of those who consider betting a regular source of additional income rose over the year from 15.3% to 19.1%, while the share of those who see it as entertainment fell from 53.1% to 47.9%.

The state responded on two fronts simultaneously—fiscal and medical.

As of January 1, 2026, a law signed in late November 2025 took effect: 7% on the difference between bets and winnings plus 25% corporate income tax. Earmarked contributions to sports rose from 2% to 2.25%, and will reach 2.5% in 2028. Deputy Finance Minister Alexey Sazanov illustrated the change vividly: whereas previously the budget received around 1 billion rubles annually from bookmakers and totalizators, now it expects just over 60 billion. The burden has increased sixtyfold.

The industry has cut back on advertising. Before the reform, bookmakers spent 100–120 billion rubles annually on promotion and sponsorship contracts; by mid-summer 2026, according to estimates from the Unified Gambling Regulator, that figure could drop by 70–75%. Fonbet has reduced marketing by 40–50% and forecasts profits will fall five to six times. Turnover itself, however, according to ERAI projections, will decline by only 4–5%.

The difference between these figures is the answer to where the industry's money is. And OLIMPBET's Development Director Konstantin Gusev spells out the strategy explicitly: companies will focus on retention marketing—maintaining and developing their current client base rather than aggressively acquiring new players.

Two more measures take effect on September 1, 2026. The first is voluntary self-exclusion: an adult citizen submits an application through Gosuslugi or an MFC and enters the ERAI registry, after which bookmakers, totalizators, casinos, and slot machine halls can neither accept bets from them nor send them advertising. The minimum period is twelve months, and early withdrawal is not permitted.

The mechanism is older than any regulator. Ulysses, ordering himself tied to the mast and his oarsmen's ears sealed with wax, does exactly the same thing: he doesn't count on future fortitude but preemptively deprives his future self of the ability to yield. Jon Elster analyzed such "Ulysses contracts" as a paradoxical case of rationality that operates through self-limitation. The impossibility of revoking the ban before a year is up isn't bureaucratic rigidity but a condition of effectiveness. A rope that can be untied from within offers no protection from the sirens.

The second is Ministry of Health Order No. 666n, which includes pathological gambling disorder (ICD-10 code F63.0) in the care protocol for "psychiatry-narcology." Problem gamblers will be treated by addiction psychiatrists under the compulsory health insurance system, free of charge, on par with patients with alcohol and drug dependencies.

And a caveat about numbers that Russian publications almost never make. The press widely circulated an estimate by Kaplan Panesh, deputy chairman of the State Duma Committee on Budget and Taxes: up to 12% of the adult population, around 13 million people, may suffer from severe forms of gambling addiction, and only 5–10% of them seek help.

The first half of this estimate diverges from international data by an order of magnitude—the Lancet commission estimates the prevalence of problem gambling among adults at approximately 1.4%. And the coincidence of 13 million with the number who placed at least one bet suggests where the figure may have come from. The second half, unfortunately, sounds plausible: very few people seek help for behavioral addictions—stigma is high, and specialized care formally didn't exist until September.

What Actually Works

The list of measures researchers are discussing is fairly mundane. That's its virtue: no bans for the sake of banning, only targeted intervention in specific mechanisms.

Protection by default. Lia Nower of Rutgers University, whose team analyzes all betting activity in New Jersey annually, found that only 1–4% of bettors under 25 use deposit limits and self-exclusion. The conclusion is obvious: limits should be set by default, and users should have to disable them deliberately and consciously.

The objection is equally obvious—isn't this paternalism? Richard Thaler and Cass Sunstein answered it this way: there's no such thing as a neutral default setting. Some value must appear in the deposit field, and someone chose it. Today, it's chosen by those whose revenue grows along with that number. The question, then, isn't whether to intervene in the architecture of choice, but whose interests it serves by default.

Friction instead of seamlessness. A delay between decision and bet placement, a delay in deposit processing, restrictions on live betting. In Australia, once a match begins, you can't place a bet through the app—only by phone. The approach seems archaic right up until you remember why cash-out features and micro-bets exist. Friction restores the gap in which reflection becomes possible.

Honest interface. Show not just the current balance, but the amount lost during the session. This isn't even a ban—simply a refusal to hide that information.

Advertising. Ohio banned the terms "free" and "risk-free," New York introduced mandatory harm warnings, the UK in 2022 banned celebrities from appearing in bookmaker ads, and Italy banned such advertising entirely back in 2018.

The industry has a counterargument. American Gaming Association representative Joe Maloney explains aggressive engagement tactics as a product of the competitive environment: legal operators compete not only with each other but with illegal platforms, where the interface is just as good but there are no restrictions whatsoever. Add friction to the legal product—and customers will go where no one protects them. Russia's 60% shadow market shows this argument isn't empty.

But it also exposes the heart of the problem. As long as profit grows from depth of engagement rather than number of customers, the industry has no internal incentive to reduce the intensity of its impact. Self-regulation here isn't even hypocrisy—it's a structural impossibility. Demanding that business voluntarily cut what constitutes its revenue means demanding the impossible. The task falls to regulators, to medicine, and only lastly to the individual. Not in reverse order, as is commonly assumed.

Conclusion

The gaming industry doesn't exploit human greed—greed would actually be harder to work with. Instead, it works with mechanisms that everyone possesses and that activate automatically: variable reinforcement, the gap between "liking" and "wanting," the illusion of control, and the reluctance to accept sunk costs. None of these are character flaws. All are documented in scientific literature, reproducible in laboratory settings, and carefully engineered into the product—from the placement of slot machines near entrances to notifications that arrive precisely when someone has stopped opening the app.

This doesn't mean players bear no responsibility. Explanation isn't justification, and understanding the mechanism doesn't turn anyone into an automaton. But responsibility is distributed, and the data suggests how to distribute it more fairly.

The British figures: 14.1% of players generate 43.5% of revenue. Connecticut: 1.8% of the population accounts for half of sports betting income. The Russian market: customer numbers aren't growing, but revenue is. This means the conversation about "responsible gaming" is missing the point. A responsibly playing customer brings this business almost nothing; economically, the industry depends on those who have precisely lost the ability to play responsibly. Kant would have called this treating a person purely as a means, and it's hard to find a more accurate formulation for a model where revenue structurally depends on those most harmed.

That said, placing a bet on a football match doesn't become evil, nor does visiting a casino on vacation become an offense. Only one thing follows: the construct in question is not a natural phenomenon but a sum of design decisions. Each was made by a person. Which means each can be reconsidered.