78. Preventing The Fall of AAA and Web3: Principals of Monetisation Technology

In a previous paper I attempted to explain the core reason why AAA is failing: The use of monetisation tech they did not understand. The obvious next step is to explain how that technology works.

As my current project (REVENGE from Everreach Labs, an advanced social MMO built on a Web2/Web3 bridge economy) locks down key design elements and production ramps up, I’ve been spending a lot of time explaining how something so complex works to my teammates. Such an explanation turns out to be a complex task in itself.

[In my previous billion dollar revenue designs (World of Tanks Blitz, World of Warships) I generated the models and numbers in secret and my dev team would input them without knowing how they worked. They just worked, kind of like magic. The downside of this approach was that I became a single point of failure and when Wargaming failed to protect me from a hostile employee that threatened me because I was American, I exited the company and this terminated multiple current and future projects in development. The same person ultimately caused design legend Chris Taylor to leave the company, which compounded the problems for WG.

This occurred at a zenith of my design work/research but of course the usual NDAs, while perhaps beneficial for the companies applying them, prevent me from sharing my knowledge with the public (and other studios) and have a negative effect on the industry as a whole. Here I am trusting that after 10 years I am permitted to go into some detail. If that turns out not to be the case, I will likely retract or delete this paper. ]

The successful adoption of Web3 economies requires the deployment of much more advanced business models, and this in turn means much higher complexity. This in turn means higher development costs/time and enhanced risk. That means larger and more skill diverse teams. Gone are the days where I could enigmatically drop into a studio and deploy numbers that no one understood. Just explaining all those numbers to a diverse cast is in itself a technology. It is that technology that I am going to share here, not the numbers magic that makes a billion dollar Games as a Service project possible. But understanding how more advanced monetisation models work is a prerequisite for not only deploying them, but intelligently deciding whether they should be.

Monetisation Models are a Contract

When you deploy a monetisation model in a game, players decide if they will invest in your product or not based on the terms you offer them when you release the game. To me this is essentially a legal contract. Anything more than a minor change could be interpreted by consumers as a breach of that contract. Backlash can be significant and we have seen plenty of evidence of this recently, for example when Sony attempted to tamper with the business model of Helldivers 2. They were forced to roll back this breach.

This is the first in a long list of key business model concepts that AAA developers appear to not understand which I will address in this paper.

Generally one of the great advantages of an online Game as a Service (GaaS) is that it can be updated regularly and become better over time. This does not apply to the business model. That should stay as static as possible, unless it is changed in a way that favors consumers. That’s unlikely to be considered an option since the presumption is that the developer is surrendering power to consumers. This generally only happens in extreme emergencies.

So even as a game is getting new maps, content, skins, visual effects, marketing campaigns, or even entirely new engines, the business model should not be changing. Thus it can be argued that the business model is the most key technology deployed to a game as it will determine revenue generation rates (explained below) and game longevity (also explained). Thus, under even the best of outcomes, the business model will determine how much money you could make and how long your game will be in service before you retire it.

The choice of business model then is of paramount importance. Currently in AAA this means you pick the one with the highest Revenue Generation Rate. This is not the right approach, and not understanding why is why we have so many layoffs.

The goal in successfully selecting the best monetisation model for a project is to pick the one with the highest Revenue Generation Rate that you can successfully support with the necessary infrastructure and labor assets.

Those last seven words are really important. But if you don’t know how a monetisation model works, you have pretty close to 0% chance of organizing the associated assets. Even if you choose an advanced monetisation model, you are likely to perform even worse than if you used a simpler and weaker model if you cannot execute the advanced model properly. Likewise, the more advanced model you choose, the more labor assets will have to be applied to metagame design and production, and those people will need significantly more skill sets, requiring more years of education/experience, and costing more. They may also have a higher opinion of their self-worth, so they will expect better treatment.

This advice will likely fall on deaf ears, like a lot of my good advice to developers, because the tendency is for game developer leadership to use “whatever is best” without consideration to the cost, complexity, or consequences for misuse. Actually, using “whatever is best” is a misunderstanding that gamers often have, especially pro gamers. Only the most poorly designed games have a “best”, as this should be situational, requiring players to adapt to the opposition/environment.

So am I going to drop enough secret sauce to make this process manageable for leadership? I hope so, but this is a complicated topic and the answers may not be as convenient as my readers would like.

Classifying/Ranking Monetisation Models by (Potential) Revenue Generation Rate and Complexity.

[I had to replace the chart here with a photo of it since it refused to format properly on Substack]

Technically every monetisation category in this table is a Game as a Service model. It’s critical to not try to aim above your weight class when it comes to your metagame, which includes monetisation model, economy (if any), progression and reward systems, and social systems (if any). Sadly, even if you do hit the category that fits your team skill sets, that may not be enough if it’s an older model.

RGR: Revenue Generation Rate. This is the Group Budget Capture Percent (QC%) that I described in detail in my 2013 Group Monetisation paper, revisited. Since it was a complicated concept in 2013, and probably just as complicated in 2024, I urge serious readers to read that paper entirely. The basic concept is that players collectively have a budget (Q) they have allocated to playing their online game of choice, and how much of that total budget is actually collected from them is the QC%. The better the model, the higher the QC%. As described in that paper, if you make any of a long list of errors you will lower Q even as you are trying to raise QC%. Clearly the leaders of the largest game development studios didn’t read this paper because they did everything I warned against there and then got confused as to why their monetisation rates were so poor.

An RGR rate above 100% is only possible if your game is so good that it convinces players to raid their other budgets (like rent or food) in order to expand their gaming budget. This takes some time as it involves some potentially difficult choices on the part of consumers. It’s the reverse process of when consumers reduced their gaming budgets in response to the use of analytics and data “science” by developers as described in my 2017 Data Implosion prediction paper. Which… is how we got here. Games with RGR’s above 100% have not made it to market yet, but since that’s about to change, that’s why I feel it’s time to describe them in this paper.

These advanced models have been in design since 2011 (very primitive in 2011) but are difficult to bring to market because of their complexity and the high degree of risk aversion among game development leadership. Unfortunately for them, consumers demand new fresh entertainment content so risk aversion in this industry is a path to a long slow commercial death. And continual layoffs.

Category 5 and 6 games, at least the way I design them, lean heavily on neuroeconomics. Since game neuroeconomics is not taught in universities, this presents a problem for industry. Actually, even game economics (without the neuroscience) is still not taught in universities to my knowledge, 12 years after I was hired as the first game economist at Microsoft. I predicted this need and attempted to develop this academic field in 2009 but was rejected for “non academic reasons”, so I doubt that academia is going to be helpful to industry in this regard. Especially if those reasons (like race or autism status) are still held as more important than academic merit.

Categories 1 and 2 are now legacy categories with almost no chance of breaking even if deployed today. They were competitive in the eras when they were deployed (now 20+ years ago), but time has ground inexorably on, and consumers’ needs have adapted. Category 3 contains some of the world’s largest games that generally follow the best practice guidelines set out in my 2012 Supremacy Goods microeconomic model. But even Category 3 games are on their last legs. My internal information suggests that all of these games are now struggling to produce enough revenue to get to a net positive state when all overhead is accounted for.

World of Warships was the prototype for Category 4 models (described in the next section) so it might be properly described as Category 3.5. But it proved the efficacy of the key changes that were required for this category. It took a lot of trust for Wargaming to risk the livelihoods of their 4000 employees on these new models, but it definitely paid off for them. After 10 years it is still going strong enough to be profitable and pay for Arnold Schwarzenegger’s commercial appearances. The 2014 economic sanctions against Russia were devastating on the company, and the current sanctions are even stricter. But World of Warships was designed to be Western-centric and thus brings in critical USD to the company. 

So what am I up to with Categories 5 and 6? Is this shameless self promotion? Fear of being perceived that way has certainly reduced my enthusiasm for talking about them, especially without data to back up their claims of efficacy. I first published my intention to begin using neuroscience in game development back in 2014, so I’ve been quietly developing this tech for a decade now. 

Category 5 Games: LCS (Large Cohort Spectatorship)

All category 5 and 6 games strive to replace stressful dopamine reward and monetisation loops with healthier oxytocin loops. Oxytocin generating activities (like massage, sex, or “talking to someone who listens”) tend to be a lot less available to consumers and command very high prices, especially in the companionship starved West. The loneliness epidemic is real, and shows no sign of decelerating. 

This presents a commercial opportunity to apply advanced technology to provide these services more reliably, at any time of the day or night, and at a much lower price point. As dopamine is a stress hormone that does damage over time (especially for children), and essentially all games and social media are dopamine-optimized today, people are becoming much less healthy. This applies especially to younger generations that are more social media and game dependent. The great thing about oxytocin is that it acts as a counter chemical to dopamine. Where dopamine suppresses the immune system, oxytocin bolsters it. I go into a lot of detail on this in The Physiology of Gaming (2018). 

Because we can generate New Realities in game worlds (described also in The Physiology of Gaming), we can “trick” the body into thinking that it should release oxytocin, even in response to stimuli that was generated digitally. In Category 5 Games the approach is to create events with large and highly involved spectator crowds where everyone is gifting to each other several times a minute. Top performers or even savvy spectators can end up with more gifts than they gave, and turn a profit. But even those that do not will feel really good playing it because of the highly social nature of the environment. Because you don’t need to leave your house, and you can do it at any time you need a pick me up, and because it is so (relatively) inexpensive, it could quickly replace how people now spend their “socialization budgets”. Humans are social creatures. Social interaction is not a luxury. The body literally begins eating itself if you don’t get enough of it. Poor people, those that are socially awkward, and especially young people are taking it the worst. Inexpensive and deadly drug options to ease the pain are also devastating communities. 

REVENGE is the first game that will make it to market using an LCS monetisation model. Our public facing pre-alpha will launch in late 2024, and each stage of testing will have more complete oxytocin loops. To me at least, gathering the data to show that it works is going to be even more important than the exciting revenue generation opportunities. 

Category 6 Games: Persistent Social Networks

Category 6 games work by creating teams of people who need to cooperate daily to compete with other teams in cooperative play. What we now call “social networks” are often more anti-social networks and even Facebook’s own internal research (which they tried to hide) indicates that these anti-social networks are harmful to participate in, again especially for young people. The Guardian reported that research presented internally at Facebook in 2019 showed that a third of teenage girls using FB’s instagram product were being significantly harmed by its use. This of course was prior to the pandemic, where these effects surely were going to accelerate and Facebook leadership was of course benefiting from keeping that knowledge secret. 

Persistent Social Network games create smaller private groups of up to about 40 people (science indicates it’s hard to maintain more than that number of friendships) where there is interdependence built into the game play loops. You can’t advance without the help of your teammates. This incentivises social interaction and doing it in the context of gaming reduces the risk and anxiety of trying to talk to a stranger. As with Category 5 games, spending generally helps others more than yourself, but this tends to create reciprocation at conversion rates close to 100%. As the games are more enjoyable than regular competitive games, player budgets are larger and so are revenues. 

Stargarden, developed at Arrivant (I am a founder), began development under this model in 2020. It was an ambitious project that likely needed an 8 digit budget to complete but when the FTX scam wiped out many of our potential investors, that became an improbable goal. I think it’s inevitable that a Category 6 game will be built in the next decade, especially once it becomes clear that the technology already exists. Of course the most important thing required to make that happen is to get supporting data, even if it occurs with initial small populations. 

If Category 5 and 6 games are as popular with consumers as I believe they will be, then once one makes it to market it will be increasingly difficult to successfully launch any online game using Categories 1 through 4. You already see some of the most expensive games ever created being launched in the last year and being mothballed almost instantly. These were high 9 digit budget games, with some possibly even hitting 10 digit budgets once marketing was factored in. 

“How you can wish a company to fail simply because they do not cater to you or that the product does not please you is beyond me.” Stevy Chassard, Monetisation Director at Ubisoft. 

I completely believe that it is beyond him when the Director of Monetisation at Ubisoft publishes this sort of statement. In the environment where he was promoted and rewarded for these beliefs, this is likely a very normal perspective. But the goal of any successful company should be to serve its customers, not the other way around. 

Monetisation Performance by Category

There are a lot of factors that go into making a monetisation model work. At their core, initial advancements in monetisation focused on increasing revenue generation through advances in what us economists call Discriminatory Pricing. The idea here is that different consumers have different budgets but often need the same things. So we can in theory charge them more for the same product. The classic example is putting corn flakes into 3 different boxes in a grocery store. The fancier box costs more but has the same corn flakes as the other two. The cheapest box is directly aimed at the most price conscious consumers. 

This is what makes Category 1 models so poor performing. There is just one price for the subscription and while some players might be happy paying 10 times or even more for the service, they can’t. This results in only a fraction of the consumer budget being captured and thus a low QC%/RGR. When Blizzard realized that World of Warcraft was crippled by their Category 1 monetisation model they tried to put a limited F2P option in the front to try to get Category 2 performance. This didn’t really work. Then they added microtransactions, again with limited effectiveness, but in theory allowing for higher discretionary spending. In both cases they were violating the initial consumer contract and thus damaging their brand. 

Category 2 games tend to have consumer antagonistic elements in them to try to force players to spend. Coercive monetisation is not going to work long term, especially if the consumer has less hostile options available to them. Any F2P game that violates one or more of the major conditions explained in Supremacy Goods is going to end up in Category 2. That’s also why you saw a burst of games being released after Supremacy Goods that followed the rules and thus have enjoyed long and profitable runs in Category 3. 

Category 3 games avoid Pay to Win dynamics, and thus most of what they sell are cosmetics or other convenience items. They also go hard to promote various forms of user generated content. This can be player created skins, or even just their gameplay itself. In World of Tanks the player’s behavior adds variety to every match, essentially acting as UGC. I gave a talk in Russian in 2016 gushing about how fantastic the UGC was in a game called Tinder. This kept production costs low and the highly variable and potentially high value of the “loot” available on the Tinder app created fantastic low cost dopamine loops that stayed fresh indefinitely. Even people who got upset that they didn’t get a good reward who deleted the app would often get bored and reinstall it. 

But people who use Tinder are mostly seeking oxytocin, not dopamine. And when they get dopamine instead, they get upset and complain on (anti) “social media” about it. But Tinder was never designed to generate oxytocin. It is deceptively marketed as such, but it’s not that product. 

World of Tanks is barely a Category 3 game because it contains some pay to win elements, which I tried desperately to remove. The systems to remove the pay to win elements are all there, and I know they are because I created them. They could be implemented very easily but likely won’t ever be. This is one of the reasons why the World of Warships model is so much higher performing. Also, WoWs has a hidden 2nd tier subscription that was very challenging to implement. When you play a capital ship (battleship or aircraft carrier) the game is more expensive to play. But this is never explicitly disclosed. This is why WoWs is the first Category 4 game, and it gets that distinction by achieving better optimized discriminatory pricing without the introduction of pay to win elements. For months I was told such a model was impossible to build. By the time I was able to convince the Wargaming founders that it was, most of the total 11 weeks I spent designing the metagame had already elapsed. 

I never really discussed this publicly before, not only out of respect to Wargaming, but also because I appreciate their allowing me to test my research using their money. That was a lot of trust that I feel motivated to reward. This sort of research is very expensive and thus will never be done by any university. I have a number of friends in the neuroscience community that eagerly await the data on my current project, as they can’t really confirm their own research within the budgetary constraints they abide by. Of course the major platforms wait patiently to see how these things turn out, so that they can copy them without paying the research costs. 

I don’t really want to reward that behavior, so that is why I go “dark” and give those that have helped me time to earn a billion dollars or so before unfair competition steps in. 

Project Drakara was a social multi tier subscription game I began developing in 2011 on my own, and I took on a partner to help bring it to market after I left Wargaming. But it stalled in production due to difficulties with the developer in Ukraine after a modest spend in the 5 digit range. The design is complete and it just waits, maybe forever. It is basically a 10 tier (like World of Tanks/Warships) space combat game that plays on mobile with the ease of play of Tinder and recycling seasonal competitions like in Shattered Galaxy and that still secret 8 digit budget MMORTS that Chris Taylor was working on before he quit gaming. Two of the games in development at WG were based on my Drakara design, and had to be cancelled when I left.

While not necessary, the Category 5 (REVENGE, Everreach Labs) and Category 6 (Stargarden, Arrivant) games I have designed both use some blockchain elements. In games designed to be played for 10+ years, with assets that are hard to acquire and guaranteed to stay rare, it makes sense to include advanced player to player economies that benefit greatly from the security provided by blockchain technology. They also act as test cases to show how blockchain can be used to make games better for consumers, which should be the goal of any new technology if you want it to be adopted by consumers. Well that or reducing production costs and cost to the consumer, which is not the case here for dopamine delivery, but it is the case for oxytocin delivery.

Category 5 and 6 games can exceed 100% RGR because they essentially provide a new more valuable service (lasting safer social interaction) that can only be acquired by accident from games currently produced by the gaming industry. Thus players will expand their budgets for this new service that they likely will have a hard time doing without once they know it is available.

If this paper prevents the aborted development of even one more $800,000,000 Category 2 game, then enough investor money will be saved to create every game that I could possibly think of before I die. These failures affect me personally since they make it all that more difficult to raise money for scientifically designed games, and games that could benefit public health. This also means that these failures harm public health, so this is a much bigger topic than just preventing commercial catastrophes.