230A. DRGD1: Before There Was AI Slop, There Was Data and Dopamine Driven Design
First in a series on De-Risking Game Development, here I identify what we have previously been doing that has raised risk, in preparation for explaining how to lower risk for dev teams and investors.
This is paper 1 on De-Risking Game Development (DRGD). Our industry is in full collapse because it is being seen as a more risky investment compared to competing alternatives. Unless we can De-Risk our projects, this famine is going to continue. As I’ve often said, it isn’t game developers that make games, it is investors. As game developers we tend to make games for ourselves, not our customers. Our customers are players and investors. Both have been giving us clear advice and we just blissfully ignore it.
This is not an anti-AI article. As usual my message here is nuanced, not binary.
AI is a powerful tool, but can be counter productive if misused. The human brain is a powerful tool, but can be counter productive if misused.
In regards to deciding what to build, and how to build it, we have been misusing our brains. Now with AI, we can do the same thing, even faster.
Early designers were often rock stars because they came up with cool new ideas and then had to figure out how to turn that idea into a game. This was hard work, and a labor of love. When gamers agreed it was a cool new idea, they would get very enthusiastic and want more of that. After the Great Recession a lot of displaced business executives migrated to game development because it was seen as “recession resistant”. These were people who often bragged that they were not gamers, and looked down on gamers as unproductive people.
These executives possessed skills they learned in business school, and applied these skills to make our industry more “efficient”. Lowering labor costs, time to market, and operating overhead. They wanted systems that could be repeated over and over like clockwork to output product. These methods worked great for making cars, soda, power plants, the building blocks of society.
The problem here, that these executives could not understand because they lacked the relevant experience/training, was that consumers of entertainment highly value novelty. They may be able to buy the same gasoline or computer over and over again, but watching the same movie or playing the same game quickly becomes boring.
Business schools teach how to optimize the production of necessary goods that people will buy over and over again. They fail entirely to teach how to create entertainment goods that consumers will buy once but not repeatedly purchase.
[The reasons for this difference are neurological, and thus largely invisible to people who have not been trained to see these systems through this lens. That side of this dynamic will be addressed in the next paper.]
Thus since 2009 the game developers led by the aforementioned executives have been looking for ways to allow reliably repetative products to be output in a formulaic way that would eliminate unreliable/unpredictable characteristics like “creativity”. They also seek to De-Risk by eliminating innovation, which is inherently high risk, because by definition innovation has no clear road map. Another word for creativity is intelligence, which will become a critical concept when AI involvement in the process is explained at the end of this article.
By ~2014 the execs had identified two systems they could adopt to industrialize game production, while minimizing the creativity wildcard:
Data Driven Design (DataDD): Using collected data to iteratively produce similar products at progressively lower cost/time,
Dopamine Driven Design (DopaDD): If we can convert our previously creative products into addictive products, then we can force gamers to repeatedly buy our products even if they don’t like them.
Hollywood similarly tried to industrialize their creative processes, but since they could not control dopamine delivery in the way that games can, they had even less success than game dev did. You can see how this played out with the Star Wars IP, with Lucas being the old school creative approach and post-Lucas (and post Disney acquisition) being the industrialized version.
Consumers have less preference for industrialized entertainment, because being “novel” is a key element in how much entertainment they derive. When they already know how this story is going to play out, this content is boring.
Data Driven Design
This is the idea that if you collect enough data on a subject, both internally and from competitors, you can optimize your products iteratively over time. You use the data you have gathered to inform what changes you should make to save 1% here, and 1% there. It also can give insights as to what products are preferred currently by consumers. If consumers are buying more of “A” than of “B”, then a higher consumer demand for “A” can be inferred and thus “A” is the better choice for what to build.
Issues with this philosophy:
Consumers are adaptive. That means your data was only useful when you gathered it and may be inaccurate just weeks later.
Consumers adapt to experiencing “A” by wanting less of “A”. They biologically down-regulate repetitive experiences.
If every company is using DataDD, then they will all start building “A”.
The lack of “B” in the marketplace increases demand for “B” with consumers.
The flood of “A” in the marketplace causes competition.
The more competition, the higher quality your product requires to be competitive. That means the budget has to rapidly increase.
Competition forces sale prices down. Sale prices are maximized if you have a monopoly as you avoid negative competitive pressures.
DataDD logically must increase overhead and reduce revenue over time. You’ve removed variability by ensuring non-profitability.
From the perspective of an investor, you’ve are optimizing for commercial failure by using a DataDD philosophy.
Note that this wouldn’t happen if you were selling a necessary non-entertainment product like gasoline or non-entertaining foods like rice. Consumers have to buy these things so they don’t care about novelty, they are mostly price sensitive. This is what is taught in business schools, which is why the education brought by these business-trained execs is toxic to game development.
It also creates a business that is toxic to investment. This is because by “De-Risking” the way they were taught in business school, they maximized risk from the perspective of the investor. An investor might be willing to suffer a 90% failure rate if the payoff for that 10% hit is high enough. By using DataDD to make games, you ensure a nearly 100% failure rate. This is why the “hits” are coming from “indi” developers who don’t use DataDD as their development philosophy.
Dopamine Driven Design
Consumers are fickle about entertainment products. While entertainment may be a need (to reduce stress), there are many ways to fill that need and consumers will tend to pick solutions that they have not tried recently as that is “fresh” for them.
An addictive drug may start off as a “fresh” entertainment source, but once you are addicted (which for strong drugs can happen after one exposure) it is no longer a source of entertainment. It is now a “need” like gasoline or rice. By making your entertainment product addictive, a developer can force consumers to be more compliant by coercing them into acting “like they should” within stardard business educaton doctrine.
Since the consumer is forced to consume your product, biologically, now they don’t down-regulate from repetition. Boring becomes acceptable because boring is better than being “in withdrawal”. Instead of using your entertainment product to combat stress, your addictive product now becomes a source of stress. Attempting to stop likely means moving laterally (key concept!) to another addicting product, not “going cold turkey”.
This is especially effective on children as they are more vulnerable to addiction, may have less freedom of choice, and adapt faster to anything that affects their physiology. Things that tamper with physiology are much more likely to have permanent effects in children as this can generate additional receptors on their cells. Thus they overreact to similar stiluli.
I was smoking unfiltered Camels at the age of 5. We may never know what that did to my nervous system because it is unethical to study this on children. After 6 months I developed severe asthma. While we may not know exactly what a drug does to children, we generally know it’s a bad idea so we try to require children to become adults before we allow them to ruin their bodies and nervous systems.
This last generation of game developers has benefitted by using DopaDD to boost sales with children while lobbying regulators to look the other way regarding health risks. Ignoring the ethical implications of harming children entirely, regulation is starting to arrive outside of the USA where regulation is minimal. Continuing to rely on DopaDD increases risk for your products globally as your product may be restricted in the near future. This also means risk for investors that might consider DopaDD ventures.
There is also some risk that consumers themselves will become aware of the risks and attempt to protect their children. From my experience, parents are pretty slow to protect their children so this is not the primary source of risk to DopaDD developers. What is a risk is blanket bans by governments, such as Australia’s ban of DopaDD social media products for children under the age of 16. These are very popular with parents as they don’t want to be the “bad guy” with their kids by taking away their electronic devices. As long as the government is the “bad guy”, then the parent can just shrug and say “I didn’t do it to you”.
Okay I mentioned “moving laterally” five paragraphs up. When you use DopaDD to addict your players, you aren’t making them addicted to your product per se. You are making them addicted to physiologically unnatural levels of dopamine release. Because this is so harmful to humans, we have built in safety mechanisms that down regulate our response to dopamine. This means that it starts to have less and less effect and be less and less pleasurable over time. Eventually it goes from being pleasurable to being stressful.
This doesn’t make them want to keep using your product. They will move laterally to a competing product to try to get their high again now that your product isn’t doing it for them anymore. That can be any social media platform or Game as a Service (GaaS) as in 2026 they are all using DopaDD. This is why it is smart to do blanket bans of these products with children. For the consumer the problem with moving laterally is that once they are dopamine down regulated, dopamine just feels stressful regardless of what source they are trying to use to stimulate its release.
This is why all of the DopaDD products released recently without novel social elements have been Dead on Arrival (DOA) for the last couple years. This has spooked both developers and especially investors because they don’t understand why it is happening. What they do know, is that these companies have become very high risk.
I explain and predict this phenomena in detail in my paper on the D2-Oxy Heterocomplex. This is essential reading for all game developers and related investors. The first step in avoiding risk is determining where that risk is coming from.
What was a commercial boon to early adopters of DopaDD, has turned into a sure fire route to insolvency due to the down regulation effect I explained in that paper. Investors should stay as far away as possible from all companies (game devs and social media platforms) using DopaDD as part of their business model. This will change if developers and social media platforms learn to make their products more social. And no, social media platforms are not social. They are spaces where misogyny, misandry, racism, classism, ageism, transphobia, etc are promoted because threat is algorithmically prioritized.
Consumer Physiology
If you are selling gasoline, it might help to understand the “physiology” of automobiles that still use Internal Combustion Engines. If you are selling something that goes into a human body, and especially if it affects the human nervous system, then it helps to understand human physiology and neuroscience.
Here in this paper I am making the case that the reason that contemporary business doctrine has failed so miserably in game development and for Hollywood is that Consumer Physiology is not taught in business schools. It isn’t even considered. Business programs typically require no biology background. This leaves a huge blind spot in modern game development leadership that is dominated by MBA types. They often aren’t even gamers.
They may assume these are businesses like any other, and follow the same rules. But that’s not true. Any product who’s target is the human nervous system is not going to behave like a typical product because the human nervous system is both sensitive and adaptive. I’ve heard 100 conflicting opinions on what is causing the industry collapse. But all of this can be explained by human physiology.
I made that case all the way back in 2017 when I successfully warned about the coming industry implosion. Data Implosion (2017) would lead to my being de-platformed across the industry. My prediction being correct just made it all the worse for me. But deplatforming me had catastrophic repurcussions for the industry that I don’t think the advocates of that action foresaw. Removing the dedicated physiologist for the industry that was filling that business education void left them driving blind right into a cliff.
I would go as far as to say that Consumer Physiology should be a key part of the education of all business majors, for those that plan to sell things that affect human physiology. That includes everything from rice to the latest Star Wars movie. If you can’t anticipate how consuming your product will alter the behavior of your consumers, you are going to be left completely unprepared when their behavior (and purchase patterns) adapt/alter.
AI is Fast but not Intelligent
AI is a tool, and knowing how to use the right tool for the right job is critical for maximizing efficiency. AI still isn’t very “smart”. It knows a lot, so it is SUPER educated. People often conflate education and intelligence. Back in 2012 I explained why this was going to cause people to become less intelligent over time. We now depend on cloud knowledge instead of crystalized intelligence (all explained in that paper).
This means that the solution to all our problems could be right in front of us, and we wouldn’t know enough to identify it as critical information. I’m reminded of a scene from The Andromeda Strain (1971), which was one of the first movies I saw in the theater. I was 5 years old in 1971. A scientist was reviewing petri dishes to see if any of the antidotes was working against the Strain. She nods off just as the smoking gun petri dish rotates in front of her and the opportunity to stop the Strain is lost.
This is how humans are today. With almost infinite information at our finger tips, our ability to make use of that information is the lowest it has been in almost a century. Similarly, AI can give you incredible detail on any topic you ask it about (if it’s not a censored topic). But that’s not intelligence, it’s knowledge. AI isn’t very good at applying that knowledge because AI still isn’t very smart. It is also not good at identifying and correcting its own errors.
As stated previously, creativity is a form of intelligence. As we moved towards DataDD 15+ years ago, we laid off a lot of our most intelligent employees. They went elsewhere and likely aren’t coming back. Leadership obsessed with DataDD is also going to be fond of the idea that they can use AI to avoid hiring and training new employees.
The result will not only be an intelligence gap, but a generational intelligence gap. This isn’t me being ageist. We are literally not training younger people to replace older people like we used to. As the older intelligent people are laid off or age out, they won’t be replaced in sufficient numbers.
This is very bad news for game development. Back in 2006 I attended E3 as part of the press pool. When we got to the Bethesda booth, we were greeted by Todd Howard, their lead designer. Oblivion had just been released about 6 weeks earlier and I remarked, loudly, “I put a rock on my keyboard and came back 2 days later and I was max level.” The rest of the press were asking Todd “can you really do that?”
Todd’s head was turning red, since this was an obvious design flaw, but to his credit he didn’t comment. Oblivion still sold well because it was graphically gorgeous and people loved exploring it. Two decades later Todd has made sure he still gets all the press attention but Starfield was an even worse design disaster. Influencers love to rip it apart. The graphics are okay, and mostly proceedural. But that’s not enough to satisfy gamers anymore, and AI is especially good at that tedious but low intelligence content creation.
Gamers don’t like AI, though they may have a hard time articulating why. Their expectations for graphics are going to go up every year since AI makes that cheap and easy. What’s going to stand out, and what gamers will pay a premium for, is high quality design. This is what is going to be what investors want to look for, as that’s what’s going to distinguish hits from all the Slop.
If you think about what DataDD is, and how the goal is to reach some mathematically perfect iteration, you can see that AI is going to be amazing at speeding this up. So anyone embracing DataDD and AI is going to combine these as a top priority. If DataDD worked, that would be a match made in Heaven. The problem here is that, as discussed above, DataDD will destroy your products and your company.
AI will enable you to destroy your company faster.
So yes AI will be used widely in game dev. A lot of studios will be wiped out. The smart ones will use AI to augment human intelligence, not replace it. As an investor, you shouldn’t be asking if the studio will use AI. You should be asking HOW the studio will use AI. Pitch decks are nice and all, but having the team demonstrate to you how they are productive is much more meaningful in determining the prospects for success.
If the studio has a lot of graphics but not much design, that’s a bad sign. If they have a design, then ask them why they chose that design. If they say it meets specific consumer demands/needs, that’s an excellent answer. Especially if they meet one or both of the Core Consumer Needs. If they talk about how their design is data driven, that’s a very bad sign.
Here I went into what makes for a high risk game, from a design and philosophy view. I’m not an expert in production, so all those technicalities are best left to someone more qualified. In the next paper I will assume you’ve read this one, and then drill down deep on what a team can do to De-Risk their project and make life easier for investors.