To me, games have so much potential beyond the “fun”, dopamine feedings contraptions that may people play for.
They can mean something, they can express, they can contribute to greater ideas, they can challenge, they are art, but it seems to me the greater view of games is tainted by the goliaths who treat them merely as a vessel for profit.
I love everything that donkeyspace seems to be pointing towards, and I am fully behind this movement.
Reading your post sparked a connection for me. I've been thinking about what is going on with LLMs over the past year, and I had landed on gambling as a core metaphor. Not in a normative sense, but more like affectively, what does it feel like to work with tools that are fundamentally probablistic? It felt to me like it tickled some of the same parts of my brain as playing games. I hadn't thought of that metaphor arising out of my experiences with games, but perhaps it did.
Surely the gamic feeling of LLM coding is mostly about the fact that they create an illusion (or reality) of fantastically amplified effectance over familiar problems? (while also, as you touch on, having a nontrivial mastery curve)
Well, I might frame that as "variable reward" -- effective even without a feeling of developing mastery.
This rabbit hole goes deep, but just to pull on one small thread. The question of domain familiarity and LLMs is tricky. Experts in a domain can more rapidly validate the quality of results. For high drudgery tasks (large refactors, style changes, complex but routine algorithm implementations) the value can be there. In this mode, it's not about amplified effect really. Programmers I talk to report more of a sense of relief. "I'm glad I didn't have to manipulate all that text myself." It's do-able, it's just annoying.
On the other hand, having high expertise tends to also mean your satisfaction with LLM-quality outputs is low. You're getting like 60th percentile quality anything out of the model. If you can do 90th percentile work in that domain, that will never satisfy you.
So you're saying it doesn't count as increased effectance because it doesn't produce similar/acceptable effects as doing the work before? I've always assumed that's a solvable skill issue, solvable by like, reminding it to follow the style of the codebase, or knowing just the right hints to mention like "I know you're gonna do it this way, do it this other way". By the end of this you may not have offloaded any thinking, but you will have still offloaded a lot of typing.
Your blog is a great source of inspiration for me, and I hope to see more of it in 2026!
Are video games art? It's been a legitimate question for nearly half a century. For me, the answer has always been a self-evident "Yes," but the question still haunts us. I believe that interactivity will be the pivotal medium of the 21st century, as cinema was for the 20th. It will perhaps be a while longer until our Shakespeare and our Michelangelo are recognized, but I'm certain the day will come.
Now is the time to think it through and contribute to the theory of what games are, and what they can do!
I just finished The Beauty of Games and feel this post is definitely a good continuing of that read for me. Excited for the ongoing pursuit!
As someone trying to explore games as art and how that intersects with theology and religion, I'm always thankful to have more to read from you and your peers.
Just well said. I agree completely.
To me, games have so much potential beyond the “fun”, dopamine feedings contraptions that may people play for.
They can mean something, they can express, they can contribute to greater ideas, they can challenge, they are art, but it seems to me the greater view of games is tainted by the goliaths who treat them merely as a vessel for profit.
I love everything that donkeyspace seems to be pointing towards, and I am fully behind this movement.
Always down to chat about stuff like this.
Reading your post sparked a connection for me. I've been thinking about what is going on with LLMs over the past year, and I had landed on gambling as a core metaphor. Not in a normative sense, but more like affectively, what does it feel like to work with tools that are fundamentally probablistic? It felt to me like it tickled some of the same parts of my brain as playing games. I hadn't thought of that metaphor arising out of my experiences with games, but perhaps it did.
Old and self-promotional, but perhaps salient enough to be forgivable: https://medium.com/@drewwww/the-gambler-and-the-genie-08491d96aee6
Surely the gamic feeling of LLM coding is mostly about the fact that they create an illusion (or reality) of fantastically amplified effectance over familiar problems? (while also, as you touch on, having a nontrivial mastery curve)
Well, I might frame that as "variable reward" -- effective even without a feeling of developing mastery.
This rabbit hole goes deep, but just to pull on one small thread. The question of domain familiarity and LLMs is tricky. Experts in a domain can more rapidly validate the quality of results. For high drudgery tasks (large refactors, style changes, complex but routine algorithm implementations) the value can be there. In this mode, it's not about amplified effect really. Programmers I talk to report more of a sense of relief. "I'm glad I didn't have to manipulate all that text myself." It's do-able, it's just annoying.
On the other hand, having high expertise tends to also mean your satisfaction with LLM-quality outputs is low. You're getting like 60th percentile quality anything out of the model. If you can do 90th percentile work in that domain, that will never satisfy you.
So you're saying it doesn't count as increased effectance because it doesn't produce similar/acceptable effects as doing the work before? I've always assumed that's a solvable skill issue, solvable by like, reminding it to follow the style of the codebase, or knowing just the right hints to mention like "I know you're gonna do it this way, do it this other way". By the end of this you may not have offloaded any thinking, but you will have still offloaded a lot of typing.
Your blog is a great source of inspiration for me, and I hope to see more of it in 2026!
Are video games art? It's been a legitimate question for nearly half a century. For me, the answer has always been a self-evident "Yes," but the question still haunts us. I believe that interactivity will be the pivotal medium of the 21st century, as cinema was for the 20th. It will perhaps be a while longer until our Shakespeare and our Michelangelo are recognized, but I'm certain the day will come.
Now is the time to think it through and contribute to the theory of what games are, and what they can do!
I just finished The Beauty of Games and feel this post is definitely a good continuing of that read for me. Excited for the ongoing pursuit!
As someone trying to explore games as art and how that intersects with theology and religion, I'm always thankful to have more to read from you and your peers.