Tetris and Tic Tac Toe have economies. I’d argue any game does at the fundamental level because there is constantly a decision to be made about tradeoffs, cost, and benefits of one move over the other.—Catalin Alexandru

Catalin Alexandru is the founding economist of C/A, a video game industry-focused consultancy based in Bucharest, Romania. Educated at the University of Bucharest, Alexandru earned a degree in political science in 2006, but it wasn’t long before he was working in games.

“I was working as a generalist game and level designer at EA” in the late 2000s, he told me in an email exchange.

One of the first projects he worked on was an early iOS game, Command & Conquer: Red Alert. The 2.5D real-time strategy title debuted in 2009 and cost $4-7.

Alexandru saw the free-to-play (F2P) monetization wave roll in, and caught it. He pivoted out of “necessity and curiosity into free-to-play game design when the industry started shifting to free-to-play.”

I’d initially reached out to Alexandru in February after a video game designer I respect reposted a slide of Alexandru’s that caught my eye. He agreed to be interviewed for Player Driven, but given GDC and our other commitments, it took us until now to complete two rounds of interviews and for me to put his comments into context. If you’re as fascinated in behavioral economics in a video game design/production context as I am, I think you’ll agree the wait was worth it!

Since his EA days, Alexandru has worked on a wide range of projects, including some that involved iconic IPs like Star Trek, Medal of Honor, Bioshock, Mafia, Thomas & Friends, Octonauts and Peanuts. He founded C/A in 2011, and now bills himself as a behavioral game economist, consultant, and designer who covers “all aspects of virtual economies – content, macro and behavioral.”

He’s also tinkering with something mysterious called Intersheets, which he describes as “Figma for game design.” We’ll come back to that.

An early question I put to Alexandru involved faucets, sinks, and drains. If you’ve ever spoken with anyone toiling at the intersection of game design and math, chances are they’ve said something about this classic analogy that involves inputs, storage, and outputs.

He waved my question off like an annoying fly.

“The faucets/drains analogy comes from system dynamics,” Alexandru said, “It's now second nature to economy designers. The drawback is it narrows the overall view a lot, and it makes it easy to lose perspective and think you can describe all of the important parts of a game just in those terms.”

“It very easily becomes a crutch to justify any decision,” he continued, “especially ones with a financial stake because it can be graphed out and explained very neatly, even if it goes against the grain/shape of the system otherwise.”

Translation: The faucet/drain analogy isn’t the kiddie pool of game economics, it’s more like a deceptive, dangerous tar pit.

Okay, so what’s his better mousetrap? I asked him to define game economy.

“It’s hard to beat Brain Giaime’s definition for straightforwardness: The dynamic exchange of resources, time, and power between multiple systems and entities,” was Alexandru’s response. “These days, with the rise of UGC platforms and secondary economies, I would specify real world and virtual resources.”

Brian Giaime is the lead systems designer for Unknown Worlds Ent’s Subnautica 2. This underwater survival-adventure game is set on an alien world, just launched, and has sold >2 million copies within 12 hours. Maybe Giaime is on to something?

He has his own site if you’d like additional color; he and Alexandru are indeed two peas in a pod since Giaime wrote on his site that in “an alternate timeline I'm probably a behavioral economist. This timeline is more fun, though!”

I bet it is atm!

Alexandru isn’t complaining.

As we walked through his working definition of a game economy, it became clear that the math of these dynamic “systems and entities” is hopelessly enmeshed—or it should be, at least—with the rest of the math that’s baked into a game. And there’s generally a lot of it. Identifying where the “economy” of a video game ends and the rest of a game’s math starts can be a fool’s errand. All this math needs to hang together for it to work well, Alexandru implied, including the math relating to player behavior, irl money, and “fake” in-game currencies.

Even some game economists underestimate the importance of these connections, according to Alexandru.

“What most people with ‘game economy’ in the job title don’t realize,” he said, “is that this also describes the systems powering the core gameplay loop. There is almost no level at which a game can be played where a player is not making a decision about the cost of performing an action over another to achieve a goal.”

That’s a working definition of behavioral economics.

It’s often “harder to quantify and express that part numerically,” he went on, so “the move is, usually, to clone the core [gameplay loop’s math/mechanics], then ‘innovate’ the theme or small Live Ops parts of the game, like events, meta-progression features or minigames.” (Note: If you’d like some context on Live Ops, my February discussion with ArenaNet’s Crystin Cox is a good primer.)

Alexandru maintained that the “resource exchange portion” of a game economy is optional for a large swath of traditional games, including Tetris and Tic Tac Toe. Many of today’s most popular video games, though, include a “crafting or deep loot system, in twitch shooters or action games, but it doesn’t always gel, and it’s easy enough for players to engage with only minimally, or largely sidestep” these resource exchanges, which have explicit links to long-standing economic models and theories.

There is, however, always a core gameplay loop in a video game. This loop forces players to make choices between scarce resources (i.e., time, irl money, virtual items/resources). That loop also has a mathematical foundation, and these choices put core game loops in the domain of behavioral economics.

“For mobile games,” added Alexandru, “the resource management portion, or the metagame, is the majority of design work and the main focus, while the power [core loop] side of it is mostly perfunctory or cloned wholesale.”

“It’s come to dominate and almost define the profession. For better (a lot of game developers have made a living from that for a long time) and for worse (the huge financial success that can be achieved through superficial customizations disincentivizes structural innovation).”

Alexandru believes it’s “impossible to be a fully rounded out practitioner without understanding all of these sides. Without understanding how mechanics generate demand in your game, you can only work on the supply side, which leads to this cycle of increased UA [user acquisition] spending and complicated, but not complex, superficial additions that are only meant to extend time a player spends in a game, even as the experience has long since ceased to be meaningful or beneficial.”

That sounds less than ideal.

“You can have wonderful, playful experiences like Proteus or Panoramical where there is very little goal orientation and the point is sensory experience more than structure or progression. You could make a good argument that those are the exceptions, but they also have their own logic and tuning curves, just not exposed to the player.”

Translation: Even “goalless” games have an economy under the hood. In 2013’s Proteus, players navigate a procedurally generated environment and it’s mostly an auditory experience as the flora/fauna emit music. Panoramical debuted in 2015 and this single-player adventure game also features music composed by David Kanaga.

A follow-up question involved Alexandru’s view of how a game’s economy related to other design goals, like players having fun, or finding a large audience and turning a profit.

This question was swatted like another annoying fly.

“The crucial difference is not between the game and the economy but between the economy and the monetization, which often tend to be conflated with one another,” he said.

“Most fun in mass market games,” he continued, “comes from an economy of intrinsic intentions that drive players and setting up the game to balance fulfilling one over the other. Whatever goal a studio has for their game is dependent not on making a ‘fun’ game or one that ‘sells,’ but on a game’s structural factors working together.”

There needs to be a certain combination of core, meta-progression, and business model if you want to reach a large audience, and a different one if you want to reach a smaller one while staying profitable. A less fun game with barely any player skill involved can make huge amounts of money, while a super fun game with the wrong theme, platform or business model can fail in a big way.

“A game where you can fulfil every desire instantly isn’t fun,” he added. “Fun is often defined by the friction between all the things you want to do versus what the game lets you do. Take [Konami’s] Metal Gear Solid, for instance. The player fantasy is being a hypercompetent super-soldier, and that may mean, for a lot of players, having an unlimited arsenal of rockets and bombs to level everything.”

Metal Gear is an iconic series of action-adventure-stealth games created by the legendary Japanese video game designer, writer, and director Hideo Kojima (who, from what I hear, is a real diva).

“But that would be fun for about 10 minutes,” Alexandru continued. In Metal Gear, the “item and ammo economy mediates that to constrain the player from wrecking their own fun and the game’s identity, while also occasionally letting the players let loose into action.”

So, it appears, “player fun” depends as much the constraints that designers and programmers bake into their environments as anything else. Those limits generate tension. Those tensions force tough player choices. Sometimes they make the wrong decision and their characters are baked to death in a microwave hallway.

Failure is the best teacher…and a vital source of video game fun.

Could he name a game that had changed its business model, held all else equal, and damaged the player experience?

He was able to fight off my curveball.

“There are instructive examples like the F2P games released without monetization on Apple Arcade which, as it turns out, made them less fun, and made a powerful contrast to the success of other very similar mobile games.”

Some studios, Alexandru was saying, took their F2P games, ripped out their in-game stores (and often their ads), put them in Apple Arcade’s subscription service, and many players liked them less precisely because that change mitigated the tension the original games had between players having to grind to advance and giving players the choice to pay irl money to advance.

“A good example of the reverse dynamic (game designed for premium but with the wrong monetization bolted on) is Rocksteady’s Suicide Squad,” he told me. “All of the complaints about the game were centered on the monetization, while the gameplay, characters and story were all considered on par with previous Rocksteady efforts and would likely have been as successful as their previous work.”

Rocksteady’s 2024 game, Suicide Squad: Kill the Justice League, indeed landed with a thud. The $70 title launched with a Live Ops pipeline: Post-launch spending on battle passes, cosmetic items, multiplayer perks, and related content was optional. It was a break from Rocksteady’s well-known, venerated Batman: Arkham approach. According to a June 2024 Game Informer article, Suicide Squad wound up losing $200 million.

Ouch.

Alexandru thinks that game’s core gameplay loop was just fine. Rocksteady erred when it took its Batman: Arkham template and “bolted on” a Live Ops content and monetization system that turned many players off, and that may have been because this system’s math was poorly balanced against the core gameplay loop’s underlying math.

Read on for three other behavioral game economy insights from C/A’s Catalin Alexandru:

• How Wii Sports Tennis became a poster child for an oft-used mathematical curve

• Why Diablo 3’s auction house melted down and presaged Web3 gaming’s ongoing tailspin

• Why reconciling design theory and the practice of making video games will likely remain impossible.

We’ll cover the first bullet point below, the last two next week.

Subscribe to our newsletter if you haven’t, and be sure to share the goodness with a friend!

Exponential Skill Decay in Nintendo Wii Sports Tennis

Exponential curves are the most satisfying for players because they rapidly increase rewards from player input, which is why they usually need some sort of hard cap or balancing sinks to absorb those huge numbers and keep them from wrecking the economy long term.—Catalin Alexandru

While at EA in the late 2000s, Catalin Alexandru’s interests were already starting to drift from the rising F2P wave to more interesting waves behind it.

F2P, he explained to me over email, went from “a new and promising model at the unexplored frontier to boring standardization around UA, built around blatant and constant cloning of what’s already in the market (which to be fair had always been an industry problem but mobile just takes it to the absolute extreme).”

So, what waves did he ride in the 2010s?

“The shift to my current practice came as I was looking for a space that’s not completely calcified and inimical to innovation,” he said. “This took me on a short detour through Web3, which I quickly realized was a space where the only thing anyone was interested in innovating in was grifting.”

Wipeout! We’ll come back to this topic next week.

“The idea remained of innovating game economies and finding new models that are not driven predominantly by user acquisition spending and cloning other games, so that’s where I’ve planted my flag. As it turns out, it's a very blue ocean.”

Cowabunga!

Another recent project Alexandru was involved in was Studio Ellipsis' forthcoming Nightholme. On LinkedIn around the start this year, he wrote, “Having helped shape the economy and core systems of this project at various points and seen it evolve over the past year or so, I know the team has something special that will bring meaningful evolution to the extraction genre while crucially retaining what makes it tick.”

We shall see on that front. It looks spooky. The horror extraction title combines action, survival, and dark fantasy elements as 3-player squads hunt monsters and escape rival teams.

Beyond consulting on projects like Nightholme from Studio Ellipsis in recent years, the video game behavioral economist Catalin Alexandru took a detour into Web3 gaming (he did a stint at Economics Design, a Singapore-based “tokenomics” strategy firm in 2022).

Since he’d mentioned “tuning curves” in relation to his work, I asked Alexandru how topology—a branch of math focused on the properties of geometric objects, such as terrain mapping—related to his day job.

“In its most complete form,” he responded, “it’s absolutely about finding the shape of the system and defining the limits of where, when and how player activity can flow inside a state space. That’s not really the day-to-day of most economy designers, but it’s certainly the aspiration and where we need to be heading as a profession to survive the coming AI wave.”

Ooh, that may become a tsunami.

I pressed on, and asked if and when he used mathematical curves like exponents and logarithms.

“Curves are sort of like drill bits,” Alexandru said. “You pick the right one for the right job.”

“The classic use case is for experience point [XP] thresholds, because it makes for good pacing and allows for increased enemy budgets (can’t have huge mobs of enemies if even one or two levels you up), but even then, usually with a coefficient to keep that growth under control.”

Logarithmic curves are best if you want the player to face diminishing returns for their input. A great example is, famously, the score in Wii Tennis.

I’m had a Nintendo Wii and I even played Wii Tennis back in the day, but I had no idea what he was talking about.

Double-clicking on this topic turned out to be interesting—and bizarrely funny.

Wii Sports was a 2006 sports sim made and published by Nintendo. It included boxing, bowling, golf, baseball, and tennis. What we’ll call Wii Sports Tennis here on is really the tennis piece of this larger game.

The logarithmic curve that related to players’ “diminishing returns” on their effort, it turned out, was related to skill not scoring.

In technical terms, Wii Sports Tennis used an exponential decay function to calculate each’s player skill rating. A player’s skill score behaved asymptotically: it approached a maximum limit but never crossed it. In this case, that limit was 2,400.

If you’d like to geek out on exponential decay, Wikipedia has you covered.

As I researched this aspect of Wii Sports Tennis, I ran across a 2017 Ars Technica article that had a curious headline: “Was Uber’s CEO really the second-best Wii Sports tennis player?”

Inquiring minds suddenly wanted to know.

Uber’s Ex-CEO Surfed the Asymptote

The short version is that Uber’s then-CEO, Travis Kalanick, may well have tied for second place at one point on Wii Sports Tennis’ global skill ranking system.

The problem is the game has no such ranking system…unless you knew where to look.

Before we get to that, the article also pointed out, Kalanick also claimed to be “number seven in the country in Angry Birds” circa 2014.

What the hell was this guy doing all day?

It doesn’t matter: Kalanick resigned as CEO of Uber within three months of this article coming out (and I’m not implying these two things are connected).

Left: Ars Technica had some fun with Uber CEO Travis Kalanick’s Wii Sports Tennis “career” in early 2017, shortly before he resigned in scandal. Right: The only “proof” that anyone hit a perfect 2,400 skill score in Wii Sports Tennis is this photo that some speculate was doctored. Why this guy is also shirtless and has a half-drunk bottle of hard liquor beside him remains a mystery.

Kalanick was obviously a hypercompetitive guy. He was also a grade A jerk. This is where the story took a dark turn.

Kalanick was forced out of Uber by the board after a series of cultural controversies came to light. It began with a cloud of toxic workplace allegations. He and other execs were then named in several sexual harassment lawsuits. A woman in India who was raped by an Uber driver sued Uber execs, claiming they’d illegally accessed private medical records to discredit her claim. Kalanick was caught on video yelling at an Uber driver. An investor fraud lawsuit followed: Investors sued Kalanick, alleging that he’d known about, fostered, and failed to stop the company’s “pervasive cultural issues.”

After selling all his Uber stock, Kalanick is still on the Forbes 400 list of richest Americans. Go figure.

Why So Many Games Use These Curves

Wii Sports Tennis had no online leaderboard but there was, Ars Technica noted, “one in-game measure of progress through each of its component sports.”

Each player had a Mii, their avatar. Each Mii began the game with zero skill. If a player beat an AI NPC, which the article called the “computer opponent,” their skill level would increase—as would the skill of their next AI NPC. The game’s most-skilled AI NPCs, Elisa and Sarah, had skill levels of 2,000 and 1,900, respectively. There was a way players could check their skill level against other players if they knew where to look, which, apparently, Kalanick did, and, he claims, he climbed all the way to a second-place tie at one point (while insufficiently minding, shall we way, the company store).

“To go from skill level 2,398 to 2,399,” the Ars Technica article continues, “requires 14 perfect games in a row. All told, it takes at least 160 wins against the computer AI to reach a skill rating of 2,399, and that’s only if you can pull off dozens of perfect 40-Love matches against the toughest computer opponents the game can throw at you.”

That’s the asymptotic max. That’s what Alexandru was talking about when he said Wii Sports Tennis contained a logarithmic curve of diminishing skill/score returns.

This same article cited some more number crunching that was done by a blogger named pgimeno. According to this 2013 blog post, which was authored by this self-described “geek,” there’s a “famous asymptote of 2,400 that nobody can reach” in Wii Sports Tennis. Apparently, this person just kept playing and playing and playing…like the hypercompetitive Kalanick…and also noted in a spreadsheet what happened to their skill score each time they racked up another 40-Love match against an AI NPC.

The figure below captures only the results of their first 10 wins and their last 10 wins. If you visit their blog, you’ll find waaayyy more numbers between the left and right screengrabs.

Geeks united: A Wii Sports Tennis fanatic, who apparently also loves math, documented their skill score change as they ascended the game’s skill ladder circa 2013, and blogged about the results. The game’s skill scores indeed mapped to a logarithmic curve (aka an exponential decay function).

If you look at the green columns above, which show the net difference in skill score gained per 40-Love match W, you’ll see they’re high and generally falling on the left and are much lower but still dropping on the right. If you graphed the numbers in the “Skill before” column, you’d wind up with an asymptotic curve that approaches 2,400.

Many, many video games contain such progression curves. Why?

Alexandru already answered this: “…because it makes for good pacing and allows for increased enemy budgets.” A veiled faucet/drain reference there!

What’s the last video game you played with some kind of XP leveling system? I’d wager you leveled up quickly at the start, then the “real grind” kicked in, and leveling slowed dramatically per hour of playtime invested. I’d further wager you were hooked by the core gameplay loop by that point, so the progression slowdown didn’t induce you to churn out of the game.

That’s sneaky math.

The studio behind that game, meanwhile, didn’t need to make as much new content because distance between player “watering holes” kept getting further and further apart. That curve saves them money and time.

A noob player has a much greater psychological need for positive reinforcement. “Am I playing the game right?” Rising numbers and achievements/bonuses after each core loop experience tells them, “Yes.” Most vet players have mastered the core loop and have moved on to (hopefully) greener, more interesting/nuanced pastures.

I asked Alexandru if he had an example of a recent game that had a finely-tuned economy—and I meant both the visible, obvious economy-related mechanics and systems as well as the hidden, subtle mechanics and systems that players “feel” but never see expressed in a visible number.

He painted a bleak picture.

“Truly new or compelling economic models have started coming in much slower due to the industry’s consolidation, VC pullback, and macro uncertainties,” he told me. “Constant layoffs are also a factor since expertise has been systematically driven out of the industry in recent years in favor of optimization and cost cutting.”

Ultimately, he fingered Embark Studios' Arc Raiders “because it beautifully reframes old school MMO-style raiding for a modern audience. [IO Interactive's Hitman:] World of Assassination and [Grinding Gear Games’] Path of Exile 2 are other great recent examples, each with their own unique takes on secondary markets and UGC. Roblox is worth a mention due to the Cambrian explosion of game economy models there, but it’s very much its own thing and [I’m] doubtful success there can or should be replicated outside that ecosystem.”

Next week, we’ll wrap up with more brilliant insights from Catalin Alexandru, so come on back!