201A. Making Unreasonably Engaging Games
Here I explain the scientific process for optimizing engagement in interactive media. I also explain why this is so difficult for the game development industry.
I had intended to not interrupt The Fourth Tier of Game Development series with any other content. But I was inspired by this video on how the world’s most complex machine was made (despite it being “impossible”) and began pondering the parallels with what I envision for gaming. So I will finish Fourth Tier on my upcoming trip to Thailand.
Interactive media has the advantage over other media in that it can gather information from the user and then adapt the content to that user over time. In theory, this means that an ideal content quantity and quality can be achieved over time. Even if the person changes because they are upset or tired, the media can adapt to this.
In the gaming industry, the goal of the products we make is to deliver “entertainment”. That’s actually more of a scientific condition than you might imagine. It can also deliver educational content, which is the focus of The Third Tier of Game Development, and to a lesser extent The Fourth Tier also.
I define entertainment this way:
Entertainment: The process of washing the brain with neurotransmitters, reward chemicals, and pain relievers with the goal of providing pleasure while reversing the effects and damage caused by “stress”. As “stress” is the primary product of the sympathetic nervous system (SNS), activating the antagonistic parasympathetic nervous system (PNS) is a key element. Very long term sustained entertainment can involve alternately activating the sympathetic and parasympathetic nervous systems to keep the user enthralled indefinitely.
Note that this sort of cycling can cause damage to the user and thus has inherent risks which I addressed in my I’m Dying to Play paper from 2017. If bad actors try to apply this technology without understanding it, they can harm a lot of people. This is exactly what happened in the industry after I published Game Dosing in 2013 (written in 2012). Thus I will be vague enough in this paper to make it unintelligible to similar bad actors.
AAA developers have been focused on exploiting the mechanisms I described in Game Dosing since 2013. But this is century old tech going back to Dr. Skinner’s work with mice. By focusing only on dopamine delivery, which is an SNS stimulant, they are actually raising stress, not lowering it. For this reason it only really works on people who have addiction problems and thus the target (paying) demographic is only about ~5% of the population. Ignoring the ethical implications of targeting, harming, and creating addicts, it’s just a crappy business model because of the low conversion rates. Very recent research also suggests that the reward effects of dopamine turn off after a while and the harmful effects just escalate in the absence of oxytocin. This was always implied, but we didn’t understand the relationship between dopamine and oxytocin.
So what is the goal here? We can make games so engaging that the user will value our products quite highly and be very willing to pay a premium for them. The industry already uses the term “engagement” to measure how long a player is in a play session before stopping. This is a quantitative measure for something that should be qualitative. Since this is not approached scientifically, productivity (in terms of monetisation) in AAA is extremely low compared to users’ potential willingness to spend.
Here are my steps for creating productive (again, here I mean getting paid) interactive entertainment:
Get Consent: If you are going to use scientific methods (and this should include primitive Game Dosing methods used pervasively) then you should disclose this and get consent for what you are going to do with your user. The Digital Fairness Act prohibits using Game Dosing to target vulnerable groups. So “getting consent” will soon be a legal requirement for developers. This step will take out bad actors, which is over 90% of major game dev incumbents.
Induce Immersion: This is the process of creating a content layer/IP that is more attractive than Default Reality. Often times this means you have created a New Reality that is persistent. I discuss this briefly in The Physiology of Gaming from early 2018.
Adjust Threat Level: Dopamine rewards you for overcoming some threat or challenge. The more times you overcome a threat, the less dopamine you get. Many users come into your world already loaded with threat from Default Reality. So you need a mechanism for the user to choose their threat level. Bad actors don’t understand this and just try to maximize threat.
Inject Oxytocin Pulses: Oxytocin makes users more trusting and thus more willing to spend. It also relaxes them as it is the primary PNS reward chemical. It’s becoming increasingly hard to get in Default Reality so users are willing to spend thousands of dollars for this if they have it. Dopamine, which is hammered by AAA games and social media, stops working without Oxytocin. This leads to depression.
Get the User to Stop Playing: You need a system to off ramp your players before they become exhausted. If you are doing steps 2,3, and 4 correctly, step 5 is the hardest part. If you are struggling with “engagement”, then you have a problem in one or all of these steps.
The Scientific Process
There is a lot of science involved here, and the systems involved are typically too complex to be within the budgets of academic teams. The “test bed” games needed to do rigorous study can only be done in a science-supportive environment. When I was working with Microsoft and Take 2 in 2012/2013, their focus was just getting a product out, no interest in science.
That changed in 2013 when I was hired by Wargaming. My boss in Eastern Europe, Sergey Burkatovskiy (“Serb”), was the original designer of World of Tanks. He was a former nuclear physicist. So he had a strong scientific interest. The facility where I technically worked, in Austin Texas, was itself a think tank. Many of the employees there were top scientists already in game dev. After I applied my tech to World of Tanks, World of Tanks Blitz, and World of Warships, he authorized a full blown research program.
It was to be conducted at the Florida QA center, and we were going to test all aspects of physiologic response including dopamine and oxytocin blood draws. I had a team including a top neuroscientist green lit to go in 2014. While I have a knack for balancing these chemicals properly in my designs, to fully optimize levels we need that data. Also, humans only have limited lifespans so it doesn’t help if I’m the only one with that talent and I can’t transfer it to the next generation. If we had the data, schools could teach it decades after I’m gone.
The Crimean War was an unanticipated obstacle here. Rising nationalism among some of the WG founders created an East-West polarization. This led to my travel and research budgets getting locked. I was technically an American employee working for Wargaming America, not WG.net. So my budget was not under the control of Serb. Ultimately almost all of the Western employees in Austin were laid off by mid 2015.
Secrecy is very important in Russian culture. People only knew what they needed to know. I had a trust-less relationship with Serb. I knew how the designs worked, and he had the data on how they worked. Thus we were both screwed if either of us acted untrustworthy. But the Nationalists in the company, who controlled WG America, they didn’t even know about the research projects. They didn’t know there was a special tech involved in these products that were creating amazing revenues and lifespans.
They also didn’t know about the trust-less relationship that made it so that without me or Serb they could not be reproduced. They just saw an American was getting too much power in the company and didn’t understand why. It made them uncomfortable. So when they shut down WG Austin, they disrupted tens of millions of dollars of research, and likely billions of dollars of opportunity costs due to future inability to make new products. There were some attempts, but they were all stillborn. I imagine that Serb (who doesn’t speak English) was just told that I had quit. My Western boss, Chris Keeling, left WG in 2018. Serb himself was fired in 2022 (for being a nationalist of all things). There’s no one left at WG that knew about our research.
I was inspired to write this article after watching the video on the Extreme Ultraviolet Lithography machines that are the most important machines on Earth. They literally make AI possible. The tech that went into them was incredible and China is desperate to get their hands on them because they defy reverse engineering. Many of the world’s top experts laughed at the team that eventually achieved the tech, saying it was unreasonable to believe it could be done.
The tech I’m describing here is a lot less complicated than EUVL. But it requires a merging of multiple disciplines, human testing, and reinventing our perceptions that the human brain is a “black box”. Of course it will stay a black box if we are afraid to look inside it. Previous attempts by AAA at “science”, like EA’s EOMM “research”, were complete disasters.
There are very serious scientists ready to go to study how the human neuroendocrine systems respond to interactive media. The results could not only vastly improve the performance of games, but also their safety. Regulation would be much simpler if we knew what we were regulating and why. Right now it’s all very superstitious, like sitting around a campfire telling each other that Zeus is mad during a thunderstorm.