As a game designer/developer I have a tendency to see games everywhere. Every trip to the supermarket? A fetch quest. Every package in the mail? An epic lootcrate. Within this view the question “what is a good life?” roughly translates to “what game do I want to play?” and I think this is a question well worth asking.
As a result of designing games, I have extremely selective taste in them and I get bored as soon as they become predictable. So, how much novelty is too much? How much challenge? How often should I “win”? Well, one of the core measures of a well-designed game (for a given player) is the felt experience of immersion, or flow. Specifically, an immersive game should hold the player in the flow channel at all times.
It turns out I have been iterating on my “life game” since I was a kid. I tend to bore easily and as a result my passions and hobbiesforce me to charge into the unknown, get confused and adapt on the fly. I prefer being in a little over my head with a topic I barely know about.
My current pet theory? All my interests are really just excuses to learn stuff. Specifically via autodidactic methods, and this really shouldn’t have been a surprise to me. My undergraduate thesis project Doublejump was a modular, personalised, adaptive self-education platform for creative coding and game development. That was ~9 years ago and I am still working this out about myself.
I’m not particularly interested in learning facts, filling my inventory with trivia has rapidly diminishing returns. It’s a closed, predictable game. Instead, I want to play an infinite game, one that continues to surprise me and defy a static solution. I want to understand everything (…and how it all fits together.)
Leading up to my work on Doublejump, I recall walking around the UQ campus, passing each building and thinking to myself:
“Why does each field of study act like it’s separate to the others? Aren’t all the schools studying the world? Surely all these areas connect somehow?”
And connect, they do.
There’s a deep satisfaction I get from assembling a model of the world that can account for everything I’ve ever seen. It’s kind of fun when new information challenges that model and I’m forced to integrate it. Of course sometimes it’s stressful to change your mind. That’s the other bound of the flow channel and it’s a signal to chill out.
While researching Doublejump, I stumbled into Connectivism which lead me to a foundational principle: perceiving possible connections between disparate fields and ideas is a skill. Everything is connected, if you can look at it from the right perspective. So the game I’m playing becomes: “look for every possible connection, between every thing and then update your model to make sense of it”. Let me talk about models for a sec, I promise it won’t take long.
Ok, so, every idea humans have is a model. Models are mental constructions that approximate elements of reality and are used to reason about casual chains and the future. They are not reality itself. The base level before we have any model at all is M0: actual ground reality.
If we construct a simplified slice of reality, we now have a model (M1):
We observe patterns in reality and codify them to create a model like the above, but looking at our diagram there are also patterns within the model. If we zoom out further we can create a meta-model, abstracting at a higher-level, landing us at the meta-model (M2):
What if we keep going? What is M3, the meta-meta-model?
Yep. That’s it. What about M4, M5, M6…. Mn?
In practice, the meta-meta-model is often self-describing. Somehow we’ve stumbled into the essence of both Category Theory and Knowledge Graphs at the same time! Any model, at its most basic, consists of only categories and the relationships between them. Why is any of this important? Because it’s an existence proof that any two models must have something in common.
Understanding comes from studying the relationships between categories within and between models, making understanding relative by nature. Analogy, not fact, is the core of cognition and in moments of insight we are adding a new analogy to our toolkit. In turn, this expands our adjacent possible by creating scaffolding for future analogies to attach to:
I define insight as the moment of “relevance realisation”; where you identify a new candidate edge for your internal knowledge graph.
If that insight is compatible with the rest of your mental model it might slot right in, or you might have to reshuffle some other beliefs to fit with this latest realisation.
Some insights cannot be integrated immediately and require more scaffolding before they will take hold.
There is no single fixed view from which everything can be understood, understanding is a constantly evolving feedback loop. All reasoning exists in relation to the axioms it is predicated upon. Perhaps, if you zoom out far enough, it’s all just nodes and edges on a hypergraph, it’s weird how much that sounds like Indra’s Net.
Next time, I want to talk about how optimal self-learning, happiness, creativity and software fit into this picture.
Thanks for reading to the end :)
Stuff I’ve been thinking about
📝 What Is ChatGPT Doing … and Why Does It Work?
📝 Experimenting with LLMs to Research, Reflect, and Plan
📝 How to use Free Recall to Learn More Effectively
📝 Think of language models like ChatGPT as a “calculator for words”
📝 Tradition is Truer than Truth
🎶 SCARING THE HOES - JPEGMAFIA, Danny Brown
hence my love for procedural generation, emergence, roguelikes etc.
e.g. philosophy, science, art, entrepreneurship, design, movement
more specifically, I want to know exactly what comes next or have no idea at all - I dislike the in-between
I can’t help but wonder how that project would go today with GPT-4 in the mix
this blew my damn mind the first time I saw it (YMMV). Also, can you believe they went with “OMG: Object Management Group” as the name?
I think we need the source code for doublejump. Looks awesome
So good. Resonating and encouraging words. Can't wait to read more.