Paradox Interactive, The Little Big Media Conglomerate

The economics of fandom show the core problem for any media company. All else being equal, it's better to produce well-loved stories than not. And some franchises create an absurd amount of love. Make something popular and people will literally name their kids after it (yes, in the year 2070 "Khaleesi" will be a stereotypically old-fashioned name). People will create elaborate costumes, travel for conventions, and spend hours compiling minutia (the Harry Potter wiki article on mistakes in the books is over 22,000 words long).

But, even if they're truly, deeply, dangerously obsessed, fans won't necessarily spend more money. A one-time movie or book purchase can lead to unlimited repeat reading and viewing, and much of the fan-service material is meta commentary that extends the canonical media without replacing it. It's a valuable marketing asset; rabid fans who constantly reference a work are liable to make it a cultural phenomenon (Hamilton was huge on Tumblr in 2015 before being huge everywhere in 2016). But that's hit-or-miss, and amounts to a consolation prize: sure, the biggest fans spend about 0.1% as much per hour of engagement as everyone else, but at least they tweet sometimes.

Media companies have learned to adapt to this. Disney is, of course, the best in the world at handling this problem: they make movies to promote new intellectual property, then convert it into merchandise, spinoffs, crossovers, theme parks, licensing, and now direct-to-consumer streaming. (See this subscribers-only writeup for more on the Disney+ strategy.) Other media companies have been able to run the same model at a smaller scale. Comcast uses parks, streaming, and licensing to monetize its IP, too. Peanuts is not a big enough media franchise to justify an independent theme park, much less a cruise line like Disney's, but it does license its IP for mini theme parks inside Cedar Fair's locations.

Paradox Interactive, a Swedish video game company, has also managed to build something like what these larger and better-known media enterprises have done: a business that converts the hit-driven nature of new creative work into an increasingly recurring revenue stream that collects money roughly in proportion to how much time people actually spend with its content.

Paradox is best-known for a series of grand strategy games: elaborate simulations that try to recreate real-world or at least rules-bound systems in painstaking detail. These include a series of history games (the Crusader Kings franchise is an over-the-top elaborate simulation of dynastic conflict in the middle ages, complete with a genetics system and a “build-your-own-heresy” feature; Hearts of Iron simulates the Second World War), science fiction (Stellaris is a space colonization/war game), and present-day simulations (Cities Skylines is roughly what the SimCity franchise aimed to be early on, though, as always, with vastly more detail).

Calibrating the difficulty of video games is nontrivial. For multiplayer games, it happens by matching players according to skill. But for single-player games, the game needs to be challenging enough that it's not boring, but not so difficult that players give up in disgust.

The general way to design an open-ended video game that doesn't have a single dominant strategy is to build in a set of constraints that have some superlinear scaling and a decay factor. The goal is to ensure that for whatever strategy a player pursues, there's a point at which diversification pays. So a strategy game might allow players to get benefits from their first few diplomatic relationships, but for additional ones to bear an increasing cost. Or it might be designed so that the cost of dealing with rebellions in conquered areas scales based on the amount of conquest the player has done lately. (So if you conquer 10 million people, and then you add another 10 million, you double the amount of rage each one experiences, and the cost of conquest scales at people-conquered-squared.) This dovetails nicely with realism. In theory, the bigger an empire is, the larger its taxable/conscriptable area is relative to the length of its borders, so expansion should accelerate over time; in practice, the ones that grow too fast will suddenly collapse.

Choosing difficulty levels is also straightforward in historical simulators: at any given point in time, some countries are fairly lucky (the UK has a convenient moat between it and every other powerful country) and others are in a tough position. So players can select whichever country and goal approximates how skilled they've gotten, such as creating an alternate history of the Second World War in which Luxembourg conquers the world. The games have "achievements," basically a list of accomplishments players can have, with stats on the percentage of players that reach them. Some of these achievements happen quite easily, and some take continuous effort; using the same mechanics while aiming for different goals is close to, but not quite, playing another game entirely, and tends to add replay value.

How do you monetize a game when a few players will play for ten or twenty hours and get bored, while others will invest hundreds to thousands of hours? What Paradox does is it offers expansions. For its popular games, every few months the company will:

Release a new $10 or $20 expansion that adds new features, and
Change features of the core game.

So some customers will purchase a $50 base game and play it for a while. Others will develop an intense habit of playing the game, and will be a $40/year or so annuity for as long as they stay engaged. Every expansion adds mechanics, and every mechanic gives players something new to optimize. If the audience is selected for being addicted to hyper-optimizing, or addicted to deep realism, then every expansion gives them a fix.

But it's not trivial to manage this. There are two core difficulties:

Any time a new feature is added to a game, it has the potential to create game-breaking interactions with other features. Go from ten core mechanics to 11 and you've gone from 45 two-feature interactions to 55, and from 120 to 165 interactions between three features. And since players only need to discover one way to conquer the exponential expansion barriers, each expansion raises the odds of ruining the experience by creating a single dominant strategy.
On the other hand, for an expansion to be worth buying, it needs to give players features that help them win. So the usual Paradox approach is that each expansion is coupled with free features that make the game harder without the paid features. This is suspiciously close to a "pay-to-win" model. Fortunately, for anyone who can afford a computer, the dollar cost of a $20 expansion is probably a lot lower than the time cost of playing it for a few hours, much less for a few hundred hours.

The main escape valve for both of these dynamics is to continuously launch expansions until players are getting impatient, or the quantity of game-breaking problems is too high, and then launch a sequel title that refactors underlying mechanics and simplifies the game. Paradox released its first Crusader Kings II expansion in mid-2012, and its last in late 2018, and then launched Crusader Kings III in 2020.

It's an interesting way to iterate. With each launch, they get feedback on features that players care about. Since games have to be played online to collect achievements, Paradox knows which games people are playing, which expansions they're buying and then deactivating, and how long a change in the game drives continued interest. And by tracking player behavior at a more granular level, they can also see which parts of the game are unbalanced, and which strategies are dominant. The ideal for a strategy game is that there are high-level choices that have variable payoffs from one playthrough to another, as well as aspects of play that require more technical skill.

Over time, though, dominant strategies tend to converge, because it's too hard to change them without altering the fundamentals of the game in a deep way that might upset players and introduce bugs. So the habit of launching sequel titles, and then starting the expansion train up again, is a good way to deeply refactor the experience. (It's also a good way to refresh the graphics; you don't want an "upgraded" game to be unplayable on hardware that ran the pre-expansion version just fine, but people will buy new computers specifically to run brand new games.)

The asymptote that a well-run expansion strategy reaches is that it's practically a subscription product: as long as you're playing, you're paying. But payment has friction, and a player can be engaged enough to keep playing without being quite bought in enough to spring for an upgrade. Paradox is adjusting to this economic reality by giving players subscription options—they get all of the game's content as it's released, and pay every month.

The social dynamics of video games increasingly lead to more time spent per player on the most successful titles. For multiplayer games, there's a network effect; players socialize while they're gaming, or compete with one another, so the dominant title continues to win. For single-player games, sharing strategies and tips online (the subreddits for Paradox's marquee titles all have 200k-400k members). And, as games switch from one big release to continuous iteration, and monetization follows this, player time represents an implicit vote for more development on the same title. This does create some perverse incentives to make the game addictive rather than good, but they're offset by making it more lucrative. There isn't a strict contradiction between creating compelling episodic media and producing something of artistic value. Scheherazade and Charles Dickens both pulled it off. Paradox represents a broad theme in economics and finance: as markets get more efficient and as information gets more abundant, it's easier than ever for a business to set itself up so everyone pays in proportion to how much they like the product.

Get 20% off for 1 year

Elsewhere

Millennials: Not Poor, But Levered

Nick Maggiulli argues against the millennial poverty theory, noting that on an age-matched basis, millennials have similar real financial assets and net worths compared to previous generations. Some of the narrative is because of end date bias: the theory of broke millennials started getting popular in the early 2010s, but many measures of income and net worth were low then because of the financial crisis. (If there's a recession right around when you'd otherwise start saving money, and it prevents you from saving money, your savings are 100% lower than the average person in a previous generation; the younger you are, the more sensitive your net worth is to how much you save rather than the return you get). Nick notes that millennials do have substantially more debt than previous generations, especially if they went to college, and there's a natural explanation for that.

But it would be good to be cautious about comparing net worths over time, too: since interest rates have declined, the net worth required to "buy" a certain amount of annual consumption from dividends—or to buy a certain amount of housing—is much higher. And that same decline in interest rates benefited the people who did most of their savings before or while it happened—mostly Baby Boomers, and somewhat Gen X. So millennial net worths won't go as far as Baby Boomer and Generation X net worths did, even if in absolute terms they're similar.

Vaccine Markets

Supplying Covid-19 vaccines is a complicated economic problem. The goal is to get them to everyone, but "everyone" consists of people with asymmetric information, different preferences, different willingness or ability to use alternatives like social distancing, etc. Giving away vaccines and rationing them by age group and essential worker status has been a popular default approach, and worked reasonably well for the US after some early delays. But adding a monetary incentive worked even better.

This story from India is a darker look at vaccine incentives: the government opened up access to its vaccine booking service, and scrapers promptly reserved spots and started auctioning them off. One interesting element of this is that the first story in the article is about someone seeing an ad for such a service on Instagram, and then scheduling the vaccine (and setting up payment terms) over WhatsApp. Vaccine appointment scalpers are a monetary barrier to getting vaccinated, but a subsidy to promoting vaccination. It's entirely possible to have a similar set of positive incentives without the pathological downside of making appointments impossible to book through regular means; the government could provide a bounty on vaccination signups, which, given the human and economic cost of the vaccine, would be an excellent deal.

It's hard to spin up a system that can be used by everyone in a country with 1.3 billion people, but can't be abused by any of them. There are organizations that are pretty good at rate-limiting digital access to scarce resources, and at uniquely identifying people—like Google and Facebook. Operating at big tech scale, without getting thwarted by black-hat tactics, probably requires working directly with big tech companies.

Only Chamath can go to the SEC Headquarters

Chamath Palihapitiya has written an editorial calling for more SPAC regulation. This is not unprecedented; yesterday's post was all about an analogy to Mark Zuckerberg's requests for social media regulation. His argument is that SPAC sponsors should commit more capital, that SPACs should disclose more about the merger process, and that SPAC deal terms should be allowed to change the way PIPE terms can. SPACs do have a regulatory advantage over IPOs, since they can make projections that IPOing companies are reluctant to, but the most straightforward way to collapse that gap would be to allow more companies to make long-term estimates when they go public. Requiring a larger sponsor investment would also slow down some of the more egregious deals (and slow down the most aggressive promoters, since they'd have to exit one deal before committing to the next). Giving investors more rights to renegotiate a deal will probably not make a huge difference. SPAC holders have the right to redeem their shares at cost if they don't like a deal, and the fact that hedge funds do and retail investors mostly don't is one of the drivers of poor SPAC returns post-deal. Some markets are inherently riskier than others, and a market in early-stage companies will necessarily be more volatile than a market in more mature businesses. Aligning incentives is a good way to keep unscrupulous management from abusing their investors too much, but the underlying traits of the market limit how much risk can be eliminated.

Reigning Over Big Tech in China

One theme of yesterday's post was that companies sometimes ask to be regulated; the main theme was that monopolies can create beneficial R&D that's subsidized by their core business. Chinese Internet companies are being directed to do this:

Tencent Holdings Ltd., Alibaba Group Holding Ltd. and Meituan have all warned investors in recent weeks they’re prepared to open their coffers to expand in areas such as cloud computing, autonomous driving and artificial intelligence. The coming deluge promises to transform the internet landscape by funneling capital into fundamental technology and infrastructure -- not coincidentally priority areas for the Communist Party.

One of the challenges of state-directed investment is that it's very hard to pick winners. Politically-connected companies get easy access to capital, but the ones that are effective in part because they're bad at politics get left out. (Why was "prefer to invest in companies led by technical founders" such a good move by YC? Because of range-restriction: those founders were likely to be relatively bad at boardroom politics, so if you counter that selection effect you tend to get higher quality managers.) The CCP's approach seems to be that they'll bet on the winners, even if those winners come from different fields. Given how quickly China's consumer Internet companies launch and iterate on new products, it makes historical sense. Alibaba started out connecting factories to wholesale buyers, and ended up creating a spinoff that launched, among other things, the world's largest money-market fund ($, WSJ) (at least for a while; as it turns out, a money market fund that paid market interest interfered with the country's financial repression strategies).

Investment Boom

The Economist notes that corporate capital expenditures are recovering much faster from the Covid crash than they did from the financial crisis ($). Based on current projections, aggregate S&P 500 capital expenditures 10 quarters after the previous peak will be up about 20%, compared to +5% after the financial crisis. Or, put another way, it took corporate capital expenditures about three years post-crisis to recover as much as they have already recovered in this cycle. This is especially important because of the debate over low capacity as a driver of inflation. This economic cycle looks more like a reset in which the world will produce more, instead of the typical reset where it returns to the old steady state.

One driver of the difference is a change in which companies dominate the S&P's capital expenditures. Big tech has gotten much bigger since the last recession, and big tech companies have started spending far more than they used to. But that kind of mix shift is not an uncommon driver of changes in how corporations behave: US companies' spending decisions were very different, on average, when most market cap was railroads, and changed as industrial stocks got more representation. Increasing tech representation in broad market indices doesn't necessarily cause a change in how the average CEO thinks, but it causes a change in how the average dollar of available cash gets spent.

Archegos and SPACs

Complex market ecosystems with lots of leverage have a sort of "the Archduke was assassinated and it led to what!?" dynamic. Since financial leverage requires collateral, but the leverage happens at the portfolio level, and since many investment vehicles are diversified, a blowup in strategy A can have its biggest impact on strategy B. So, Archegos ran a concentrated portfolio of growth stocks, and its collapse has unwound a popular trade where funds bought diversified portfolios of SPACs ($, FT). The basic mechanics of the SPAC trade are:

Buy at the IPO, get SPAC units plus warrants for $10.
Think of a levered position in a SPAC as equivalent to a warrant: since the SPAC units can be redeemed at $10, owning a SPAC unit has capped downside—the cost of funds multiplied by how long it takes for a deal to close.
If the SPAC units rally, sell them; if they never rally, redeem at $10; the only cost is the opportunity cost of capital plus the (cheap) cost of financing.

This deal is incredibly attractive if available leverage is high and cheap, and if there are enough people doing it that SPACs never go below $10. But Archegos led to broad cuts in hedge fund leverage for equity trades. Since buying a SPAC at $10 is just about the safest equity trade imaginable, it's the most levered one by default, so SPACs bore the brunt of this delevering. Coincidentally, since SPACs typically target growth companies, this means that growth stocks on average have taken a hit due to Archegos—but only coincidentally, because high-variance growth stocks are disproportionately likely to be born as low-variance SPACs.