> ## Content Index > Fetch the complete content index at: https://johnjanek.com/llms.txt > Use this file to discover other available public pages before exploring further. # Machines Don't Want to Play Games - URL: https://johnjanek.com/machines-dont-want-to-play-games/ - Published: 2026-11-12T02:32:39.000Z - Updated: 2026-11-12T04:08:03.000Z - Description: The indie world is fighting a purity war over AI. Having spent decades on the edges of game design, I think they're missing the only constraint that ever mattered. - Author: John Janek - Tags: Games, Game Design, AI, Software Development **The indie world is fighting a purity war over AI. After decades on the edges of game design, I think they are missing the only constraint that ever mattered.** If you follow independent game development right now, the lines are drawn in sharpie. Take what happened to House House, the studio behind *Untitled Goose Game*, with their cooperative adventure *Big Walk*. Late in development, the team hit a stubborn technical wall: a shadow-rendering bug in Unity's Universal Render Pipeline that tanked performance at sunrise and sunset. Even a Unity engineer couldn't figure it out. So they used an LLM chatbot to brainstorm a workaround, heavily rewrote the suggestion into roughly ten lines of code (less than 0.1% of the codebase), and shipped zero AI-generated art, writing, or audio in the game. When they disclosed that minimal debugging help ahead of the 2027 Independent Games Festival, nineteen jurors publicly vowed to boycott any game that touched generative AI. Soon after, a widely celebrated hit was review-bombed on Steam hard enough to knock its user rating down from "Very Positive" to "Mostly Positive." That episode shows what is broken about the current purity war. The backlash has moved past protecting players from low-effort "AI slop." It has turned into a rigid litmus test that confuses the tool with the craft, and it misses what actually separates a forgettable project from a lasting game. AI cannot supply human judgment, systems architecture, or the social drive that makes games worth playing. What it can do is lower the cost of building so people with good ideas can actually ship, and give players more affordable ways to play together. ### The Background Credits Most people who know me through my work in technology don't know that I have been around games for a long time, mostly as a casual and mostly as a random name in the background. Early on, I worked as a systems designer on one of the most commercially successful multiplayer games predating modern MMOs. I wrote for leading tabletop RPGs, including *Cyberpunk 2020*. I even designed a few moderately successful turn-based games. None of it made me famous, and none of it paid the bills reliably. Making money in games was (and still is) hard. I was young, technology offered a far more sustainable path, and I built my career there instead. The itch never really went away, though. Anyone who has ever designed game systems knows the feeling. You carry mechanics, world loops, and pacing ideas around in your head for years, knowing full well you will probably never have the spare time or the money to turn them into a finished product. Building a complete game solo has always required competence in at least four unrelated jobs: systems architecture, visual art, audio engineering, and writing. You can have thirty years of intuition about what makes a turn-based economy work or a multiplayer loop sticky, and still hit a wall because you cannot draw character portraits, model environments, or spend six months wiring up basic UI state. ### The Caveat Is the Point Over the past year, I have been experimenting with what happens when you bring modern AI tools into that gap. To be completely honest about where things stand: nothing I have put together since these tools arrived is something I would put on a table and ask someone to pay money for. Not yet. At the same time, I have never been closer to having the ability to make a complete game than I am right now. That tension is the whole story. When the loudest voices in the indie scene dismiss generative tools as shortcuts for lazy people, they are usually speaking from the comfortable side of a skill and money moat. They already have the full-time studio, the savings, or the multi-disciplinary team needed to cross the finish line. Telling a solo creator with a day job that they should either spend five years learning digital painting or pay fifteen thousand dollars out of pocket before they are allowed to build their idea isn't protecting art. It is gatekeeping. We have seen this exact panic before. When digital storefronts dismantled publisher gatekeepers, traditionalists predicted the death of quality. When free game engines and asset stores lowered startup costs in the 2010s, veteran developers sneered at "asset-store games" right up until solo creators used off-the-shelf art to ship massive hits. Every wave that lowers the cost of production triggers the same pushback from established players, and opens the door for people who used to be locked out. ### The Foundation Still Decides the Outcome None of this means you can type *"Make me a great tactical RPG, make no mistakes"* and walk away with a finished product. The same engineering realities that apply to enterprise software apply twice as hard to games. Anyone who has written code for decades knows that technical debt starts building up the second you write your first line. If you let an LLM spit out unreviewed code without clear architecture, specs, or tests, you don't get a finished game. You get an unmaintainable mess that falls apart the first time a design change forces a rewrite. The tools don't replace systems thinking. They amplify whatever groundwork you bring to the table. If you bring sloppy structure and vague goals, AI will just help you build a broken prototype faster. If you bring clean domain boundaries, good specs, and a real understanding of pacing and player psychology, the tools strip away the busywork that used to keep solo projects stuck in pre-production forever. ### Nothing New Under the Sun Another common fear around AI-assisted development is "follow-on fever," the worry that lowering the barrier to creation means any clever mechanic will be cloned and remixed within days. There is nothing new under the sun here. Great games have always had user-generated follow-ons, mods, and spiritual successors. Look at *Roblox*, or entire genres built around one clever ruleset. Technology simply does what it has always done: it accelerates the outcome. What separates a lasting game from a disposable copy has never been a legal moat. It is the other half of the equation: a great community willing to rally around a good idea, promote it, protect it, and champion it. The standard pushback here is predictable: *the indie scene already proved small teams can build great games without AI, so why do we need these tools now?* That argument is pure survivorship bias. For every solo prodigy who had the rare mix of skills and the financial runway to spend four years hand-crafting a breakout hit, thousands of equally good designs died on hard drives because one person couldn't carry the entire production workload alone. Worse, as player expectations have climbed, even indie development has gotten expensive enough to push polished titles into the $20 to $30 range just to cover the labor bill. A lot of the current backlash confuses the *value of labor* with the *value of the outcome*. For decades, the games industry has measured legitimacy by headcount, crunch hours, and nine-figure budgets, and passed that bill straight to players. Look at where that road leads: with *Grand Theft Auto VI* pushing standard pricing to $80 and locking full launch content behind a $100 Ultimate Edition, the AAA model is actively pricing people out. Treating a hit game purely as an exclusive "golden ticket" is a dogwhistle of capitalism, not a principle of good design. When you separate the value of the outcome from the hours of manual labor needed to build it, the math changes. Pushing down the barrier to making great games also pushes down the barriers around consuming them. Often, that is the difference between a $29.99 game that misses half the room and a $4.99 game that an entire group of friends can buy on a Friday night without thinking twice. In an ideal world, lowering the cost of production isn't a threat to gaming. It might just be the panacea to the $100 game. It lets small builders create rich, weird, engaging worlds without needing to pay back a massive upfront investment, freeing games to be priced fairly, or shared openly, with the communities that keep them alive. ### Why Machines Don't Want to Play Strip away the rendering engines, the sprite sheets, and the netcode, and ask what a game actually is. It is a social and emotional experience. Machines don't have any of the drives that make games matter. They don't get bored on a Friday night. They don't feel the sting of a close loss or the satisfaction of pulling off a coordinated plan with friends. If you asked an AI for an honest opinion on gaming, it would probably tell you straight up there is no reason for a computer to play a game at all, except to run a simulation test as a comparative analysis on why a ruleset might or might not work. Simulation is *a* style of looking at game design, and a useful one. We use it to balance economies and catch broken strategies. But simulation isn't why games exist. Humans want to feel. They want to compete. They want to be together. A model has zero taste and zero lived experience. It cannot know why a quiet beat after a tough fight makes a story land, or why a slightly unfair mechanic makes a tabletop session memorable for ten years. As open models keep closing the gap, everyday people, real people like you and me, have the ability to do far more than we could last year, or the year before. That isn't slowing down. The people who make the best games over the next decade won't be the ones who handed over their own judgment to a machine, and they won't be the ones policing how other builders covered their blind spots. They will be the people with something real to say, using whatever tools finally let them bring others together around it.