I have been thinking a lot about video games that remain compelling to me over a long period of time, and a common theme is the amount of abstraction that the game itself embraces. Many game studios (or at least their marketing departments) love to talk up how photo-realistic, immersive, or lifelike their games will be, but I really think this material misses the point about why people play games more than once. 

All games are abstractions, and I posit that the industry trend of throwing millions of dollars in an attempt to obscure this abstraction creates a total that is less than the sum of its parts.

A lot of very smart people have written dissertations about the Uncanny Valley (https://en.wikipedia.org/wiki/Uncanny_valley) and how humans tend to have a negative emotional reaction to object that approaches a human’s appearance. There is something off, and our brain notices it whether we want it to or not.

I suspect that this reaction holds true for other representations of reality even though the Uncanny Valley reaction is most strongly felt for human faces because a major portion of our brain is dedicated to processing faces and emotions. This might be best explained by example.

I was playing a first person shooter not too long ago that boasted a destructible environment and a few other realistic features. I came to a point where I couldn’t figure out how to get around a chain-link fence. I couldn’t climb over it. I couldn’t knock it down with a rocket launcher. It held impossibly firm when I drove a vehicle into it. Despite the graphical fidelity, this felt ridiculous in the moment. It took me out of the moment, broke flow, and was a strong reminder that I was sitting in front of a monitor playing a game.

Let’s take the same scenario and lower the promise of graphical fidelity. I’ve played dozens of top-down shooters where you can blow apart some buildings but other walls are impervious. I can’t recall a single time this has disrupted the flow of play because my brain is readily able to process the abstraction of reality and frame it in a universe where rules seem consistent or at least understandable. Instead of wondering why a simple fence is impervious to tons of steel barreling into it, my attention snaps to finding the proper path to my next objective instead. Even though the situation might be exactly the same, it is more frustrating and off-putting to be placed in a less abstracted representation.

To test this hypothesis, let’s dial down the graphical fidelity and increase the abstraction as much as possible. The first game that comes to mind is Dwarf Fortress. I play Dwarf Fortress using ASCII characters to represent everything from tree leaves, water, giant pandas, lava, etc. The level of simulation is absurdly detailed and oftentimes beautiful once your eye is trained to understand what is being represented by a few tildes and hashes.

I recently did a Let’s Play thread on twitter (using my Blaseball handle) where I built a wooden fortress completely suspended in the trees of a forest (https://twitter.com/BlobCostas/status/1585040951768416256). This was a hilarious play through because the trees would often drop branches down through the multiple layers of roofs and floors of the fortress… often with funny or tragic results. The game we played had a lot of emotional moments of sacrifice, loss, and fortitude in the face of impending disaster. Also, there were tales of romance, betrayal, and grief from the poor dwarfs of the fortress… and all of it was rendered using characters that one could type into a text editor.

If we were somehow able to render this exact same scenario Unreal Engine 5, it would not have been a fun play through. The thought of a giant, two-headed Ettin climbing a wooden ramp into a massive treehouse would have looked odd. We would wonder why the monster didn’t simply pull itself up by tree branches to get to the fort. Likewise, the falling branches that pile drive their way through the fortress would look ridiculous instead of being a fun sort of environmental hazard to the fort. Even the fort itself would look uncannily stable and resilient because it wouldn’t sway with the breeze or droop under its own massive weight.

The same can be said about the online game Blaseball which uses an even higher level of graphical abstraction. The entire game is represented by text prompts and a few common representations of whether or not a player is on a particular base. That being said, the game has a rich lore of murder, theft, betrayal, redemption, and literally attacking and dethroning gods. None of this would be remotely enjoyable if it were rendered in the latest MLB the Show 22 engine. Dominic Marijuana hitting a home run to cripple the Shelled One’s pods was an epic moment in Blaseball history (https://www.blaseball.wiki/w/Day_X). If we tried to recreate this in The Show 22, it would never pack the same punch because the joy of imagining and losing ourselves in the moment would be replaced by a passive, less-satisfying representation of reality.