I and Saska (a coworker from my day job) made this for the combined Global Game Jam (GGJ) / Finnish Game Jam (FGJ) of 2025. Before going into detail, you can and should play the game in your browser or download executables for Windows and Linux on the GGJ site. What you do when unprompted and reflection of that is key to the experience, so try to be a nasty little art hoe and do some analysis.
How it ended up
The project was probably my smoothest jam game up to this point and I think that was largely due to Saska. He’s a coworker at my day job who has a background in game dev / programming, but mostly does design for websites. He clearly had the vocabulary, drive and mindset to effectively work together on a team and the whole thing would be a hell of a lot less charming if he wasn’t there. I did do most of the coding, shaders and design, while he did design and sprites with a clear fingerprint in the design as well.
I will now proceed in roughly the following order:
- Whine and bitch about the jam
- Explain how the thing got made
- Give my take on what the message is
Whining about the jam
GGJ is a week long game jam, but the FGJ part is a 48h block, filled with lots of interruptions. I don’t want to complain too much as an outsider, but having done the Jam fully remotely, I don’t want to be notified whenever pizza is available for live jammers, nor do I really want a collaborative pitch session or a mid-check. I can see that these make a lot of sense for community building at a live event, but remotely they just seemed like a waste of time, which is annoying in a time critical setting like a game jam.
The information packages were dreadful. They sent two email PDFs which still had a fair few holes in them that needed to be plucked. The schedules in the PDFs were simply wrong the first time around for example. Still somehow after a revision, the PDF was missing or ambiguous about a lot of critical information, such as the team formation, pitching, mid-check and basically everything that happens live. I’m guessing the organizers have attended a few times themselves and take stuff like that for granted, but as a first time attendee this was a definite dampener on the experience.
Being a part of the GGJ also hurts FGJ in many ways. GGJ is a longer jam that starts earlier, so jammers can easily get spoiled on the shared theme accidentally, which happened to me. GGJ is also a staggeringly big organization and I don’t really think any of the five speakers that were stapled onto the theme reveal video had anything of interest to say. A lot of effort had clearly been put into something that could’ve been a single header tag on some website.
The GGJ website itself is also clunky on a good day and borderline unusable when there is a lot of traffic, such as when the submissions end. I’m still not a 100% convinced I managed to even submit correctly, as the instructions were were provided by FGJ contradict the GGJ website slightly. FGJ says to submit one zip where you have both the executables and source files, whereas the GGJ site has a separate file upload for both. The uploads say they have a max size of 5G, but whenever you upload, it likely breaks and the error message says there is a cap of 2M, which last time I checked was less than 5G. Despite the error and the message it told me to find in the browser dev console, I think I managed to upload the zips.
How the sausage got made
The thing itself was built in Godot, which I’ve gotten pretty comfortable with. Unless there is a reason to not use it, I think I’ll just go with Godot for all my future projects. It’s far from perfect, but the UX is worlds better than Bevy or Unity, which are the two other options I know. Unreal seems overkill for the types of projects I do, although I’ve tried it a few times. Smaller engines like löve seem to fall into the Bevy camp of good in theory but will probably be missing a lot of little things like a graphical shader editor. I didn’t really use Godot’s graphical shader editor for this project, but have done it in the past and through it was neat. I think I’ll probably use it for some kinds of effects where the intermediate previews are relevant.
I used this project to explore minimalism of file hierarchy. Usually I have a detailed folder structure, but creating and adjusting it takes time. I wanted to see how far I can go before having everything in the same folder became an issue. In the end, we submitted the project with two folders. One for art not in the Godot project and the Godot project itself. Everything used by Godot whether that be scripts, textures, scenes what have you is in the root of the godot project folder. Whenever I needed to scroll the resource view, I had the urge to make folders, but in the end it probably wouldn’t have been worth it. Godot is fantastic at letting you avoid scrolling through everything. If you type things out correctly, you can often use the quick select in the inspector to get a picker with a manageable amount of options. I think going forward I will change how I organize my projects. Not as spartan as here, but definitely more relaxed than my typical structures.
Some gripes I’ve started to notice about Godot are that occasionally it just needs to be restarted. Scenes are bleeding into each other and things are behaving strangely. Using git rebases probably has an effect on this, but I’ve found it happens in solo projects where I never rebase. Saska was on an M2 mac, which for some reason whenever he ran the game added some identifiers to three scene files, which my linux install then removed. I think we probably ping-ponged the same change over three or four times. When I scroll anything with the middle mouse button, it sometimes flings my camera in a random direction. This doesn’t happen when I use space to scroll but every now and then I forget. It only happens in Godot so I don’t think it’s a hardware issue.
Most annoyingly of all, I found that occasionally something works on desktop and
the game successfully builds for web, but that thing doesn’t work on the web.
I’ve had this happen with other things, but it’s most common with shaders. After
95% of the NPC beliefs align, the game ends. Desktop players get a smoothly
appearing blur of the game view, which keeps moving without sound. On the web
the blur simply doesn’t work and you get a black background. More importantly,
we have an indicator on the player’s hat that shows when they will next emit
messages and which colors they currently contain. This shader has a uniform
which takes in an array of sampler2D types, essentially textures. It then has
a loop to determine which of the segments the current fragment is in and samples
from the right texture. This works fine on desktop, but on the web, you can’t
index an array of samplers with a variable for whatever reason. The workaround
here was to sample all the textures into an intermediate array and then index
that instead. Thanks to the Godot discord
people for coming up with that one.
Oh and the name was originally meant to be the entire quote from Guilty Gear: Strive, as we are both let’s say consumers of that game, but sadly GitHub has a maximum character limit of a 100 in repository names, so I shortened it.
What it all means
Important note: I nor Saska agree with the message in full. This is my interpretation, which I’ve tried to create from a perspective of an outside consumer and not the author. I will not elaborate further.
The game is obviously a political statement on the polarizing effect of online echo chambers. Even if at the start there are people with no strong beliefs, simply existing in an environment with others will inevitably convert everyone to a strong believer in whatever is popular or have them reject the discourse, effectively refusing to play. Which idea ends up on top is completely random and largely irrelevant to the behavior of the system as a whole.
After a while, the divisions have been made. There are likely 2-3 bigger groups that flock together with perhaps a few stragglers that are trying to not get involved. Up to this point, the game essentially plays itself, and no player insight or action is required. The game features a timer, which is an implied objective in one of two ways. Either you should minimize or maximize it. Maybe the player goes for the later and tries to maintain a multitude of views for as long as possible, but eventually they notice the game is pretty much playing itself. No action is needed to keep the people with opposing views from interacting with each other, they seem to automatically keep their distance and most of the time won’t even interact with messages from “the other side”.
The only way to proceed is to convert. The player can enter one of the crowds to soak in ideals and then spew those onto others, acting as the propaganda arm for whatever ideology they chose. The only way forward is to regurgitate rhetoric until everyone is of one mind. It is a statement yearning to end all discourse and personality. One of idealized homogeneity. Until we act and convert others to whatever belief will work for us, we cannot move forward. People cannot be trusted to come to their own conclusions and this must be done by the player, a great man of history.
From observing others play, their typical first and only runs take about three to five minutes. You don’t have to play the game, but simply to see what happens next, most of my friends are willing if not eager to become dispensers of truth-irrelevant slop. I don’t know what to think of that one.
One could postulate that not playing is more ethical. Simply letting the discourse continue as it “naturally” would. If you do that, the game is nothing more than a fancy lava lamp. It loses meaning if you choose to not engage with it. What is the point of “playing” the game where you do nothing of significance nor of substance?
I tried to in a subtle way make the experience Train-esque, but fear it worked about as well as adding a pinch of salt to a soup. The point of Train to an extent is how it’s unbearably salty at a specific point, which is an effect greatly dampened by subtlety. In hindsight I probably should’ve made it more explicit.