205B. A Blueprint for Extremely Successful GaaS

Making a Game as a Service is expensive and complex. Here I explain how GaaS works, literally going down to the subcellular level. Success is a science, not luck, and thus predictable.

This is a topic of grave importance and complexity. 2026 is a breaking point for the gaming industry. By necessity, this is a relatively massive paper. It gets more intense the further you read. If you are reading this by email or on LinkedIn, you may not be shown the entire document. That can be found as paper 205 on my substack (link in my profile on LI).

Efficiency

Current Games as a Service (GaaS) designs on the market only generate revenue ~20% of what they should relative to their budgets. Of course bigger budgets can generate higher revenues but also higher expenses. Thus the budget of the project is not so important. The efficiency of the design is what matters. When I use this term in this paper, I refer to the ability to translate developer spend into lifetime revenue.

Here the standard I set for 100% is what you could generate using technology I designed or deployed up until 2014. I will give verifiable examples. I will also provide real data to back up my statements. One could go significantly higher than 100% using post 2014 tech, but to do that we would have to go beyond the limitations of using dopamine as our primary reward chemical (See “Dual Pleasure Engines” in the last section on Pleasure Chemicals).

Oxytocin is an even more powerful pleasure chemical than dopamine, and both are made by the human body. Perhaps it is a good thing that such a powerful pleasure chemical has such a short (1 to 6 minutes) serum half-life. It also can’t be consumed orally. That means that other than by falling in love or similar difficult-to-reproduce social scenarios, interactive media is the only reliable way to stimulate its release.

Recent research also shows that dopamine receptors are reduced in sensitivity in the absence of oxytocin. I recently wrote a full article on the Dopamine-Oxytocin heterocomplex. Thus it is reasonable to assume that with a combined D2-OTR chemical stimulation, pleasure intensity could be three times as great or higher compared to existing products (explained in depth in the last section). In a world where we are under constant stimulation, consumers are both pleasure-addicted and desensitized. Thus the market for the body’s most powerful natural high could be outrageously profitable.

Accepting 20% as your standard, because that’s what you have been getting previously, leaves a tremendous amount of money on the table. 20% is very close to the break even point, so this may allow you to tread water for a time, but is not enough to secure your commercial future. As competition increases, and consumer demand evolves, the % required just to break even will go up, it is not static. Technological advancement is a necessity, not a luxury.

Fundamental Business Case for GaaS

Ryan Ward has been busy analysing data to explain what we may have missed in the recent data on our industry. Here I recommend this post of his, and I elaborate on his analysis here with his consent. He makes the following points:

• 41% of game time spent by players on Steam is on games deployed more than 7 years ago. These are the post Supremacy Goods games I referred to in the paper I just wrote explaining why I’m writing this paper.

• 44% of game time spent is on games deployed between 1 and 7 years ago. These aren’t innovative products, but merely iterations of the products that make up that other 41%.

• The remaining 14% of attention is going to games released in the last year. Huge budgets are being thrown at consumers, who react with a resounding yawn.

As budgets go up, either price has to go up, or purchases have to go up. The latter is not happening (hence the above 14%). If anything, consumers are making less purchases than at any time in modern history. Instead they are continuing to play these old legacy games which are almost all Free to Play (F2P) and thus don’t have to be purchased.

My view on this is that the industry got a tech boost from Supremacy Goods and rode it as long as they could. Why fix it if it isn’t broken? Well, the problem is that even the “winners” in the space, that captured that early market share, have reached perfect competition and aren’t generating profits anymore. If you wait until you are in this scenario before looking for new tech, you no longer have the money to develop new tech. You would have to borrow to advance. If that experiment fails, your company likely will fail.

“Call of Duty budgets tell the story. Black Ops 3 cost $450M. Modern Warfare hit $640M. Black Ops Cold War crossed $700M.

At those numbers, selling 10 million copies at $70 doesn't work. Investors need uncapped revenue potential to justify that risk. That's why multiplayer fragmented into every business model imaginable. Free-to-play. Battle passes.
Season passes. Subscriptions. Cosmetic stores.” — Ryan Ward

This “fragmentation” that Ryan refers to, I call that a “Shotgun” approach to design. A developer doesn’t know what to do, so they do everything. If you spent $700M and your best case scenario is you make $700M, you are already non-profitable. So you need to charge more. But consumers are buying less, and paying less per purchase.

The obvious solution is to charge repeatedly for the same product. You can do this with a subscription business model, or with a microtransaction model. These are very low tech solutions, with MTX being tech circa 2001 and subscriptions being a century old or older. As I explained in Supremacy Goods in 2012, you make more money by charging for less objects. So the solution that industry adopted was to bundle all those MTX into “battle passes” by 2013. That worked, but I see developers sliding back to 2001 tech because they didn’t know why the 2013 tech was adopted. Or, they didn’t even realize that was a tech.

This concept of making more money by selling less things was counter intuitive, and explaining it at Wargaming so that I could remove premium ammo from World of Warships took more time than it took me to actually design WoWs. So I can understand why people who don’t understand the underlying tech/theory have reverted to lower tech solutions that are easier to understand.

Being able to charge repeatedly for the same product requires the product to change over time, launch with years of single use content, or have content that repeats. Content that is only used once is inefficient. Thus the key to high efficiency is to make it desirable for consumers to repeat content.

Consumers don’t like that because they get less dopamine every time they repeat a task.

Modern games primarily rely on dopamine delivery to satisfy consumers, just like cigarettes do with nicotine and Coca-Cola used to when it had cocaine in it. Both of these chemicals stimulate dopamine release. Cigarette companies began optimizing nicotine delivery in the 1960s by hiring neuroscientists. I talked about that in my The Rise of Game Neuroeconomics paper (2014).

Note that dopamine and oxytocin are nearly impossible to legally regulate as they are created inside the body. Those chemicals belong there, though perhaps not at the levels we can achieve.

I explained in that paper and a lecture at SXSW that if game companies were going to get serious about being dopamine delivery companies, they would also have to hire neuroscientists to optimize this process. I warn developers what will happen if they just assume “more is better” in numerous articles. All the consequences I describe/predict are showing up in our engagement data.

As I assume you don’t have a neuroscientist, the only way you are going to get that information is reading my papers since I am the only public source of neuroscience advice tailored specifically to game design. Here in this document I will go very light on the neuroscience so that you don’t necessarily have to find a qualified neuroscientist, but this is with the understanding that higher levels of efficiency (above 40%) will be unavailable without those skill sets. Still, 40% is double 20%. That difference is the difference between being bankrupt and being very profitable.

Efficiency Examples

This is what 60% efficiency looks like. 9 years of Steam December average concurrent player rates are charted, from my World of Warships design:

2017: 3,082.7 2018: 3,758.6 2019: 6,909.1 2020: 7,361.1 2021: 7,478.3

2022: 7,801.9 2023: 8,906.9 2024: 8,747.1 2025: 26,307.2

Numbers will fluctuate based on the timing of new content roll outs and marketing campaigns. WoWs is multi platform so this is just a fraction of the total activity. The key takeaway is that the numbers go reliably up over time, for years. And not just slightly up. Engagement rates in 2025 are almost 9 times higher than in 2017.

This is what 20% efficiency looks like, here Marvel Rivals is shown over the last year. 60% of the user base is gone, 80% down on a dip.

Battlefield 6 lost 84% in under 4 months. I would consider anything that EA puts out as “sub 20%” efficiency. Nonetheless, both BF6 and Marvel Rivals got much acclaim for their gameplay. Unfortunately for them, efficiency comes from the meta game, not the gameplay. While gameplay tech has gone up considerably over the last decade, efficiency tech has actually regressed. Gameplay alone can get you to 40 hours, but not a lot further.

Helldivers 2 is probably my most complimented recent GaaS, and it’s participation rate has been stable over the last year and a half. You will see that it comes closest to my blueprint.

I expect that a smart reader will notice that Valve’s products have high efficiency rates, and don’t use my tech. A big part of this is that Valve owns the platform I am getting statistics from. That’s a huge home-court advantage as they can advertise essentially for free.

Further, Counter Strike has an open economy. I began pioneering the field of game economics in 2005 with the intention of optimizing and defending open game economies. By 2011 the industry had largely pivoted away from open economies, perhaps because they didn’t know about my tech or didn’t trust it, so I was forced to develop into another tech pathway (game neuroeconomics). Valve essentially built their own stock market. Perhaps more than 50% of that market was built by the community, but Valve was the main beneficiary. Thus they didn’t need any special technology as they had so much collective help. This is a unique situation that would be difficult to reproduce today, especially if this market is under regulatory threat for selling potentially illegal securities and facilitating children’s gambling.

DOTA was created by Blizzard and (as with CS) the community. Control of the property was transferred to Valve using lawyers. Again, this scenario seems difficult to reproduce and using lawyers to create games is beyond the scope of my expertise. Using lawyers to force privatization of open source community assets seems to be a major part of Valve’s business model, similar to what was done with Microsoft privatizing OpenAI (a 134 billion dollar lawsuit challenging that action from Elon Musk is pending).

The Optimization Blueprint

Here I go into actionable detail describing how you can make a dopamine-based GaaS with an efficiency of between 40% and 60%. I used this blueprint to design World of Warships in only 14 weeks in Saint Petersburg in 2014. Most of that time was spent convincing the team that we could make Asymmetrical Matchmaking work.

I will break the blueprint down into four categories: Purpose, Pacing, Trust and Fairness, and Asymmetrical Matchmaking:

Purpose

Players need a reason to exist long term in your New Reality. There must be something cool that they aspire to in the distance that keeps them going. Whether that’s unlocking hundreds of new ships (WoWs), unlocking a prestige class in an MMORPG (the “Jedi” playable class in the original Star Wars MMO), getting some far off economic goal (like a huge ship or star base in EVE Online), getting as many points as possible in a cyclical game, before the next cycle (Shattered Galaxy), or social/political progression (SG, Sid Meier’s Civilization Online), players need some reason to log in every day.

The cheapest content that will do that for you is other players. If you have a social layer that creates interdependency, then the other players in your game become essentially free UGC. A good design with these systems involved is complicated, so while you may get free UGC, designing these systems is usually very time consuming. Studios generally don’t create persistent relationships as part of the design, but at least allowing it was key to the success of early MMOs like World of Warcraft.

As studios moved away from open economies they also did a lot of damage to their social layers. “MOBA” games try to fake it with temporary co-op, grouping you with other players you will never see again. Good examples are WoWs and Marvel Rivals. I wanted to add social systems to WoWs but due to time constraints that was a post-launch feature. I was almost done when I left Wargaming in mid 2015, so those systems were never deployed. I left WG because someone tried to threaten me to steal the tech, and that person was a close friend of the CEO. This is why WoWs is only 60% efficient. [As the CEO is wanted in Russia for “extremism”, along with the head of the Lesta studio (that made WoWs), and Lesta has been nationalized, I don’t expect that the WoWs design will ever be completed.]

Mobile games sometimes just stretch out the content release rate to make top content aspirational. This is a bit ghetto and it shows. For instance, without spending money in Clash of Clans I calculated the time to reach the last tier of structures as being over 40 years. Back in 2014. It could be even more now. That’s a bit dishonest (see “Trust and Fairness” below).

Pacing

The idea here is that humans are happiest when they are getting enough stimulation, but not too much. I explain some of the neuroscience in the last section of this paper on Pleasure Chemicals.

Trying to optimize dopamine delivery without formal education in neuroscience runs the risk of doing more damage than good to your product. The easiest way to explain how to do it is you need a brief period of stimulation (no more than 10 to 15 minutes) followed by a period of restful activity that is under the control of the player. This is their “cooling off time”. In WoWs the garage fulfilled this purpose, giving the players something to strategize between battles.

Allowing the player to adjust their rest period is critical. They may have been eliminated early in their last battle and as a result had plenty of time to cool off already during the match. Thus they may skip the cool down entirely. If they had a very intense match, they may need extra cool down. Players will instinctively adjust this period even if they don’t know why.

The easiest and most common way to destroy your Pacing is with in-game timers:

These timers represent a primary source of metagame income for the player. Since their output is halted once the timer clicks down (until they are clicked and the timer is reset), if the player isn’t online at that moment, they suffer immediate opportunity cost. This will put the player under pressure to play on the schedule set by the developer, especially if the devs also put players in competition via a leader board of some sort.

Of course this doesn’t happen by accident. The developer is under the false assumption that more time played means more money spent, even if that time played is coerced. This interferes with the player’s health and real life activities. This is literally harmful to your customer. Thus it is considered a “player antagonistic mechanic” (see Trust, below). Players, especially “whales” are resentful of being disrespected in this way.

That button in the picture costs the player $15 USD if they click on it. You could buy an entire game for $15. Thus it causes even more Trust damage, just being there, even if the player never clicks on it.

Trust and Fairness

If you are hoping to generate $50 to $70 per player per year, the number of years a player plays your game is critical. These long term players are your “whales”, and they can also recruit their friends to your game. As I explained in my 2015 Whales Don’t Swim in the Desert paper, your whales wait on average 18 days before making their first spend. The more a person will spend, the longer they take before making their first spend. This is because they want to be sure your game will be a long term ride before going full in on the relationship.

That paper was widely read and accepted in 2015, but I recognise that with the turnover in the industry, many probably have not even heard of this paper. Nonetheless, any relationship is based on demonstration of value and trust. Games are no different. Any sign of disrespect to the player, or even the IP, is going to undermine that trust.

For instance, if you are working on a beloved IP and you feel like you want to change that IP, that’s going to break trust with long term fans. You might even brag that you hired writers that had no prior experience with the IP. I don’t understand how this could be conceived as a commercially successful approach, yet I observe this going on with some of the world’s most valuable IPs.

A super way to sour the relation with your players is to act like you think players are gullible. What do you think a player is going to think if you send them this within the first hour of play?:

If players are reluctant to spend $70 on the top AAA games, what player is going to see this and say “oh, this button is worth $500 to me!”. More likely, they are going to think you are lying to them and telling the player you think they are dumb. This is not a good way to start any relationship.

I realize you may be desperate to generate revenue. Showing desperation is not a way to get a sale. You are showing weakness. Play the long game. Be generous with your players. Learn how to get conversions without deception or coercion. If your team does not possess these skill sets, you should not be making GaaS.

Asymmetrical Matchmaking

High levels of efficiency will require you to ask your players to repeat content. With some creativity, you can easily make it feel a bit different each time. The easiest way is to involve other players as players are not as predictable as AI. The player behavior becomes User Generated Content (UGC). UGC, especially free UGC, is a great deal for you.

It’s possible to really spice things up by making battles having variable difficulty. Pay 2 Win did this, but the problem was that the unfairness broke Trust. Even spenders don’t like P2W because it makes outcomes too certain. An Asymmetrical battle occurs when one player or one side is intentionally stronger than others. The tricky part is making this “fair”.

Some definitions:

• Team Asymmetry is when one team is more powerful than another.

• Individual Asymmetry is when one player is more powerful than others.

• Content Asymmetry is when superiority is due to having a more powerful unit or avatar.

• Skill Asymmetry is when superiority is due to skill and play record.

When you have an open world, Asymmetry happens naturally. This was part of the excitement of early worlds like Everquest. Running for my life when I was Level 15 from a L28 Festering Hag in the Estate of Unrest was some of the best gaming moments of my life. When Nexon introduced Pay to Win in Maplestory in 2001 we got unfair Asymmetry. That’s an awful experience, and one that the developers didn’t learn before making EVE Online in 2003. I credit unfair Asymmetry with holding back EVE commercially. That’s an amazing game which would have just exploded in popularity and retention if they had Fair Asymmetry.

Fair Asymmetry is relatively complicated, especially content-Asymmetry. It had never been attempted before when I tried to convince Wargaming leadership to let me implement it in World of Warships in 2014. To my knowledge it’s never been deployed in another game since then. Since Asymmetry is inherently unfair, it requires a special matchmaker (I call this an Asymmetrical Matchmaker) in order to create two or more teams and to balance them all collectively. But the individual members don’t have to be balanced as long as the overall team composition is balanced. It is also possible to let the match be unbalanced, but make it fair by enhancing rewards for the underdog. This can dramatically speed up matchmaking as sometimes balance can be challenging, especially during periods of low player census.

I had been working on that tech since my time doing balance work on Shattered Galaxy (Nexon, 2001). I built a content-Asymmetrical Matchmaker for the End of Nations project (Trion Worlds) which was intended to be a successor to Shattered Galaxy. The team refused to engage with me for whatever reason despite the community (who knew me from SG) creating a petition asking them to work with me. The project eventually was killed, I assume once they realized they didn’t have the tech to do what they wanted. So that technology ended up in World of Warships instead. Since it was mostly pre-built, it was deployed rapidly.

In 2024 NetEase did internal research on a skill based Asymmetrical Matchmaker. This is a much less complicated affair. Their research, which I analysed here, proved a significant engagement increase versus a Symmetrical Matchmaker. While it was intuitively true, this was the first time it had been quantitatively proven. It is fair to assume that you will get even higher performance from a content-Asymmetrical Matchmaker if you have the tech to do so. Given the quality of their research, I assume it is only a matter of time before they make that breakthrough.

The Nuclear Reactor Complex Analogy

In 2012 I used the Sport Utility Vehicle (SUV) as an analogy to explain my Supremacy Goods microeconomic model. Here I will do something similar. I will use the much simpler Nuclear Reactor Complex as an analogy to explain how the neuroscience of GaaS works.

In a nuclear reactor complex you have a Reactor Island which produces Heat. This heat (in the form of steam) is then sent to the Turbine Island which converts that Heat into useable/transportable Energy (Electricity). Now imagine that your gameplay is the Reactor and it is producing Dopamine which goes to the D2 receptor. There are actually four types of dopamine receptors, and those are sometimes in High or Low states, so as complex as this is going to get, this is a massive simplification.

Now in a Nuclear Reactor Complex we use the Turbine Island to convert heat to Electricity. In GaaS we use the Metagame to convert Pleasure (from stimulation of the D2 receptor) into Money. That money is paid to us by the consumer we just provided Pleasure to. As you can imagine, that’s the more complex part of this system, and that’s the part that my research and design work is weighted towards.

As developers we generally assume that Dopamine is addictive (it can be), will increase engagement (not quite true), and that if we maximize dopamine delivery we will maximize revenue (dangerously false for both us and our players). The D2 receptor was discovered in 1979 and at that time we already understood that activating the D2 receptor with Dopamine would desensitize the receptor to additional Dopamine. The Dopamine stops providing Pleasure. The more Dopamine you send, the faster it stops working. Scientists and clinicians call this desensitization “Downregulation”.

In a nuclear reactor, if you let it get too hot you can have a meltdown or destroy your Turbine Island. Either way you are screwed. In game dev, if you let your player run too hot with too much Dopamine stimulation, they will Downregulate and stop enjoying your game. Then you (specifically the game dev) are screwed. Want to know what that looks like? Take another look at the graphs for BF6 and Marvel Rivals. They aren’t losing players because their gameplay is bad or boring. Their problem is that they tried to overstimulate their players and succeeded. Now they are screwed.

From Google AI as it did a good job:

Chronic exposure to high dopamine levels, whether from games, drugs, social media, or dating apps, causes semi-permanent damage in the form of changes in the number of D2 and related receptors. That means the longer your players have been exposed to this sort of content from any or all of those sources, the less Pleasure they will get from your game. If you released it 10 or even 5 years ago, it might have been a hit. You didn’t change, the consumer changed. But technically, that’s our fault not theirs. We did that to them by abusing their D2 receptors. Now they have Downregulated just like a sugar addict’s insulin receptors Downregulate as they develop Diabetes. That’s what causes Diabetes, insulin receptor Downregulation. So you can think of this as Brain Diabetes.

The Optimization Blueprint that makes up the majority of this paper explains how to repair/optimize your metadesign side, the part that converts Pleasure to revenue. That’s just my 2014 level tech as promised but that’s enough to double your revenues. We can raise efficiency much higher but to do that we will need to convert our games to run on two Pleasure chemicals, not one.

Advanced Pleasure Engines

I struggled with whether I was going to disclose this information. It’s all publicly available information, but it’s never been presented in commercially useful form like this. Ultimately the deciding factor wasn’t my desire to get rich or make other people rich. The conversion of our games from Single Pleasure engines to Dual Pleasure engines would have massive public health benefits as this is how the human body is naturally supposed to function. The way we are providing Pleasure currently is disease-promoting. We are also accelerating Demographic Collapse by disrupting how we as a species reproduce. So we will have no choice but to discover and apply these advanced technologies if we want to avoid Demographic and Economic Collapse in the long term.

Here I compare the natural internally produced pleasure chemicals that we can manipulate with interactive media to popular recreational drugs. Consumers will pay exponentially more for higher quality Pleasure chemicals. So adding two “+” signs in the Pleasure column represents at least 10 times higher price to buy it. Price goes up when it is also illegal, but that also lowers value.

“Stress” here mostly represents cardiovascular stress. This includes higher heart rate and blood pressure. Your current cardiovascular health determines how much additional stress you can endure before having a catastrophic event. This is progressive, the stresses listed here cause damage over time. Cocaine temporarily blocks dopamine Downregulation. So you can stick 5 cigarettes in your mouth and go to town. ~Fatal means this generally leads to what we call “Sudden Death”, it’s only a matter of time.

Because dopamine and oxytocin are produced in your body, they can’t normally be made illegal. Some delivery methods like gambling can be age restricted, and thus dopamine can be indirectly regulated.

“Intolerance” means your body Downregulates when exposed to these things. So they have less effect the more you use them. This causes both addiction and reduced value. Oxytocin resets Dopamine Intolerance, so it actually repairs the damage from Dopamine abuse. Even very poor sources of oxytocin like massage can command very high prices.

Because oxytocin can reverse Dopamine Intolerance, even a little bit generated in your game will reverse the revenue declines we are currently observing. I explain the mechanism for this in my paper on the Dopamine-Oxytocin Heterocomplex.

Releasing both dopamine and oxytocin is typically a prerequisite to making babies, so this also has very important implications for human Demographics. The reason economists world-wide are baffled by Demographic Collapse is because they don’t understand the reproductive process and are trying to treat it as an economic problem. As we promote gender division on social media, just to improve engagement, we are disrupting this physiological process. This is having very negative effects on reproductive rates. This is especially noticeable in countries that consume a lot of social media.

It’s for this reason that I include this table here. I don’t see research into this as optional. I’m going to be a bit reluctant to spill the beans on how to take full advantage of this commercially unless there is an agreement to open source the research results. Yes I know how that turned out with OpenAI.

Dual Pleasure engines will make Single Pleasure engines obsolete. So given a choice people will only continue to use Single Pleasure engine games if they are too poor to be able to afford Dual Pleasure engine games. We saw this happen in the tobacco industry after Phillip Morris hired a neuroscientist to figure out why people liked smoking. That scientist identified nicotine as the Pleasure chemical and was able to free the nicotine base using urea (essentially urine). Without some science, who would have thought to pee on their cigarettes? That one move made them around two trillion dollars over the last 60 years. I described how that all went down in my paper on the subject back in 2014.

Final Words

The gaming industry is where it is today because of three major themes:

1. We treat consumers as static objects like a box of cereal or a can of soda. While they are part of supply chains, humans are organic and will adapt to their environment. If you overstimulate them they will alter physiologically to Downregulate to stop enjoying what you are overstimulating them with.

2. That overstimulation is not accidental. We have been attempting to addict our consumers for our benefit in an Attention Economy. Commercially, the problem is we don’t employ people with expertise in addiction. [That has been my area of expertise going back to the 1980s, though I don’t advertise it] If we did, we’d know better than to go about this the way we do. With almost every media company (not just gaming) attempting to addict people, the entire population worldwide is Downregulating.

3. This is causing permanent changes in humans. Just as humans need to adapt or die, companies that depend on humans must adapt or die. The die-off in the gaming industry is mirrored by the collapse of reproductive rates. This is not coincidental. We have carelessly disrupted D2 receptors on a planetary scale.

Of course every problem also comes with an opportunity. The bigger the problem, the greater the opportunity. Once you understand the problem, you can craft the solution. In the 1960s, it was as simple as putting urine in cigarettes. Today’s problems are a bit more complicated, but potentially even more lucrative.