Author: Studio Ganzheit

When making games, I always keep one principle in mind: minimize taking control away from the player. But last week, maybe because I was tired and drowsy, I did something I don’t quite understand myself. There was this frustrating issue in the very beginning where dialogue would get out of sequence if players didn’t follow the guide exactly, so I just blocked player input and forced all the dialogue to play through. I also added tutorials that freeze player control while teaching mechanics. I thought, “Ah, now it’s clean,” and forgot about it.
This week I noticed that player drop-off rates in that early section had shot up dramatically. Even with a small sample size, the abandonment was so severe that it snapped me awake. I realized I had momentarily ignored something I always keep in mind, all for the sake of convenience. So I rolled back the parts where I had taken control away from players from last week’s changes.
Working alone often feels like constantly having to observe myself. I find myself frequently ignoring rules I’ve set for myself or obvious principles, only to have feedback or numbers like these snap me back to reality.
Since releasing the demo, I’ve been stuck in this cycle of feedback, demo updates, and early-game fixes while working on Steam capsules and micro-trailers. I’d try to start on the next stage of the main game, but I’d always end up staring blankly, then drift off to other tasks. But now I think I’m finally finding my rhythm. I’ve started organizing the story roughly and placing it as events throughout the game.
Even though I don’t usually think about switching modes, changing pace just once is really exhausting. It’s draining.

I added pedestrian holograms in the preservation zone, based on the idea of installing them using “Street View from 100 years ago” as a reference. Technically, it’s nothing complicated, but having human figures versus not having them really makes a difference, so I put them in. I’ve hooked them up to the script, too, so they’ll have to play a role in the story and gameplay at some point.

I keep revising and polishing the early parts based on feedback, but it’s hard to tell if things are actually getting better or worse. It feels like I’m digging into the ground without knowing whether I’m digging up or down. Maybe it doesn’t even matter if I don’t know which way is up. Kind of like how, whether the future is predetermined or not, if I can’t change it anyway, does it really make a difference?

When I was making the prologue, I was just charging ahead blindly, excited to make things, and everything turned into a mess. This time, I’m trying to stay realistic, and the weight of the numbers coming in hits differently. The only thing I know for sure is that I’ll keep making this game as long as I still have HP left. No matter how much I look into things or have AI analyze it all, the conclusion is always simple: make it good. I can feel for certain now that creating is the only thing I can control. And I know it’s just an illusion to think I can control anything else. Maybe even creating is just me dredging things up from my unconscious, and maybe I can’t really control that either.

Writing these devlogs for years always makes me wonder—did early humans feel like this when they painted on cave walls?

I feel like I’ve roughly dealt with most of the functional issues in the early gameplay. I know more problems will pop up once I test again, but I’ve mainly been tweaking the flow and feel. I’ve added, removed, rewritten, and reinserted lines of dialogue. I’ve done the same with scenes—adding, cutting, fixing, repeating. I think this is the first time I’ve spent so much time just on the early section. I really hope it pays off. Some parts are inevitable as they are, but I just want what I’m trying to convey to actually reach the players. If I were to fix the beginning any further, I’d practically have to rebuild it from scratch—something I’d rather not do, though if I have to, so be it.

For the first time, one of my Shorts hit 27k views. It’s not an impressive number in itself, but for me, given what my Shorts have done before, it feels big. Aside from adding a voice-over and subtitles to fit what the platform wants, I didn’t change anything in the content. Of course, Shorts don’t bring in direct traffic on their own, but having some views is still better than zero.

Making a game alone feels less like just making a game and more like studying myself—seeing myself raw, bit by bit. Like I mentioned before, it’s a bit like alchemy: at first, the goal is to make gold or make a game, but in the end, it becomes about finding the gold within your own mind. I feel like I’m understanding parts of my personality and my issues more deeply than I did when I just vaguely sensed them. It’s like a journey where reality and the mind are tangled up together. Where it goes, I don’t know yet, but it doesn’t feel like a path I can simply avoid.

Maybe someday these devlogs will be unearthed like cave paintings, or maybe they’ll just get wiped out by some digital undertaker—who knows.

I’ve been working on improving the visual clarity of the boss fight based on feedback, and I’m also continuing to revise the early part of the game.
I used to get feedback about players getting lost early on, but I didn’t fully understand it at the time. Now I think it might happen when players don’t follow the tutorial exactly as intended. So I’ve been rebuilding the early section almost from scratch.
Some important systems—like “Contact”—had weak tutorials, so I’m reworking those as well. Contact is a core mechanic where players take risks in combat to recover health and generate resources. If players don’t notice this, the overall feel of combat can be very different.
Since I’m trying to stick to the original idea of limiting player movement as little as possible, it’s  labor-intensive to get everything to feel right. I’m currently using a simple, barebones in-house script, but I’m starting to wonder if the game might need a different kind of scripting—something conceptually and structurally distinct.

Whether the feedback is encouraging or frustrating, the process of sifting through it like panning for gold—and then making real improvements—is an indescribably rewarding feeling.

As always, I find it most effective to assume that what players feel is 100% right, and what they think is 100% wrong.The exposure penalty on Twitter, which had cut my impressions down to one-tenth, finally seems to be easing.
Twitter has been the best for driving traffic to Steam, and Reddit is best for feedback, so during the weeks of suppression, Steam traffic took a hit.
I’m hoping things will pick up again now.

I hit a wall while working on the Threshold, so I shifted to other tasks for a while. Previously, Pi’s lines were displayed like phone text messages, but many users found that distracting and hard to read. Recently, ElevenLabs added an emotional speech feature, so I took the opportunity to replace all those message lines while testing it out. My goal was simply to make it better than nothing, and I think I achieved that. Eventually, if I get the chance, I’d like to record proper voiceovers with real actors.
While generating the voices, I reviewed the English text and found a lot of errors. I fixed those, and there were also some bugs in the voice output system that I had to patch. I think I uploaded and fixed the build three or four times in total. Looking back, I believe I’ve at least reached the “better than nothing” goal, but I’ll have to wait and see how players respond over time.
Besides that, I started preparing a 7-minute dev video because I felt I needed to spread the word a bit more about the game. It wasn’t like making the short videos I’d done before—it felt more like writing an article over a looping video background. Anyway, I put together a rough script and showed it to both Claude and GPT, and they showered it with praise. But when I actually recorded the voice and stitched it together, just the audio for now, the result was a complete mess. Yeah, I totally got bamboozled. Remind me never to take AI advice seriously again.
Until I’m in the right state to return to Threshold work, I’ll continue with the video editing. Right now, it feels like all I can do is focus on what’s directly in front of me. It’s not easy, but somehow, I’ll keep going.

The sample size is small, but the dropout rate at the revised section has improved significantly. I just wish more players were trying the demo. 

Meanwhile, development on the full game continues. I’ve established the overall tone of the Threshold and roughly defined the feel of its most common enemies. The Beyond, which backs up reality, is designed to replicate it as closely as possible—so that the living feel no difference when they connect. That’s why there are scanners, whose job is to transfer the real world into the Beyond. Physical laws are strictly enforced there, except by system operators. Still, to avoid confusion with reality, the Beyond never has a sun, no matter how clear the weather.

But the Threshold I’m building now is a system cache between the Beyond and reality. Here, reality is often miscopied or distorted, and physical laws frequently break down. The sea floats above the sky, and the sky lies beneath it. Enemies behave in ways impossible in the real world—because AI has no reason to obey the laws of physics.

Kay’s suit programs function identically whether acquired in reality or in the Threshold, since they are, after all, just programs. Her goal is a structure called “Oryeon (兀然),” a tower that links the sea and sky within the Threshold.

I’ve made another major overhaul to the early part of the demo. I’ll need to monitor the dropout rate over the next week. There’s a particular spot where players were dropping off in large numbers, but after watching my daughter and her friend—both of whom have no gaming experience—get through it without any trouble, I figured the problem might not be that specific point after all. So I broadly revised the surrounding dialogue, quests, flow, and rewards. Players tend to continue much further once they pass that area. Hmm.

Players can now warp to new levels. It took some effort to keep things compatible with existing save data, but it’s working. With this, I can finally begin building the Threshold stage. I’m still thinking about how to make the enemies there feel fresh. Still, the core remains exploration and discovering new areas through acrobatic movement. While setting up level transitions, I found quite a few places I want to remake. For now, I’ve just left them as-is since they’re functional, but I’d love to revisit and refine them all if time allows.

Even when I think I’m done, I still end up spending half the day tweaking the demo build. While working on it, I finally completed the event where F6E’s consciousness chip gets transferred. Previously, I had temporarily handled it by giving the player a suit program. If I’d focused solely on this event, it would’ve been quicker, but juggling it alongside other tasks took time. Next, I’m planning to build the terminal room on Floor -3 of the Archetype Server to access the Threshold stage, and set up the transition into that area.

I recently got a DM from a dev I don’t know asking how I handle game marketing. I don’t really know what I’m doing either—it’s all trial and error—so I just shared what I’ve learned so far. After sending it, I figured I might as well document it here. Some of this might be specific to my situation, and a lot of it overlaps with advice you can find online.

• Open your Steam page early: If you already have materials ready, it helps to start bringing in traffic as soon as possible.
• Release the demo early and start collecting feedback early: If you already have feedback channels, it may not apply, but in my case, I had almost no opportunities for feedback. I’d heard this advice before but never fully understood it—now I do. For me, players are essentially producers, and the demo was almost the only way I could gather and apply their input.
• Post-demo “dead zone”: A few weeks after release, when the demo loses its freshness, traffic can stop. GPT warned me not to get discouraged at this point. It was disorienting at first, but I’ve learned to accept it as part of the system and now balance feedback implementation with main content development.
• Use Steam visibility tools: I gave GPT things like my click-through rate and traffic graph, and it recommended changes to tags and thumbnails. Nothing dramatic happened, but I was able to experiment and make small improvements—and at least understand how the system works.
• Reach a certain wishlist count before applying to Next Fest: Obviously, the higher the better, but GPT suggested a baseline number to aim for before applying.
• Grow wishlists through demo/Next Fest instead of pitching: I’ve learned that publishers use bots to track games with certain metrics, so it’s more effective to let them come to you through public visibility than by pitching directly. I used to assume publishers would see the “potential” in a half-finished game, but that was a mistake. Pitching might work for games that are either totally finished or wildly unique—but mine was neither.
• Reach out to sites like Lurkit or Keymailer around Next Fest: They seem to offer discounted indie support programs specifically for indie (solo?) devs using the service for the first time.
• Look at other games when morale drops: Everyone’s different, but seeing how difficult this process is for other indie devs helped me realize it’s not just me—it’s the system. (and it’s brutal)
• Use Featured Announcements wisely: I’ve heard you should only use them every 2–3 months. I’m not sure exactly how it worked, but I think I got a small bump in traffic from using one. I’m planning to write another in mid-August, when the new boss is added.
• Keep updating your community and social media:
• Steam seems to be unlikely to bring in traffic on its own unless something special happens—like sustained positive reactions or a Featured Announcement (and even then, nothing is guaranteed). So in my case, I’ve found that I need to drive traffic from the outside. That usually means posting devlogs regularly on social media and community platforms. A single post tends to stay visible for about a day—two at most. If possible, you could try posting once a week with a two-day gap between each post, across one social media platform and two communities. It felt important not to make people think I was spamming, so I tried to adjust the content to fit each space and avoided posting too frequently in any one place.
• Thinking about how to present your game: In my case, working solo meant there were clear limitations on visuals. And while I put a lot of effort into the story and worldbuilding, the game’s genre doesn’t allow those elements to be front and center during gameplay. But surprisingly, those story elements played a bigger role on social media than I expected. Even when I posted simple or unremarkable gameplay clips, if I added a short explanation about the story behind them, people showed much more interest.
• Ask AI often: I don’t know if ChatGPT or Claude can actually give me a strong, coherent plan, but when I fed them my current Steam data and development context, I got some surprisingly specific and useful suggestions. But lately, their responses have become overly accommodating and emotionally attuned, so I’ve had to filter that out manually—which can be a bit of a chore.

This is the first time I’ve actually tracked the numbers and consulted AI step by step. It made me reflect on how reckless I was last time.
While going over Steam with AI—and after spending the last 3–4 years using social media for the first time in my life—I’ve come to feel that exposure works a lot like bubbles rising underwater. No matter how large the bubble gets, nothing happens on the surface until it finally breaks above the water. And when it does, it usually gets noticed for just a day or two at most—then things go quiet again.

But underwater, there are all kinds of unknown obstacles—ones you can only vaguely guess at. So even if you manage to create a big bubble, it can disappear halfway up, or sometimes you can’t even tell whether it surfaced or not. For someone like me, who lacks social instincts and just knows how to work, this system feels especially difficult to navigate. Steam, at least, feels like one of the more fair and structured platforms.

And even when people do see the bubbles and try to check what’s behind them, if what’s down there isn’t interesting, it doesn’t matter how many or how big the bubbles are—nothing really happens. That’s how it feels to me.

I thought I had done everything I could with the demo, but that wasn’t the case. Even as I try to shift back to the main game, I keep needing to fix and polish areas where players drop off during the demo. I keep wondering—why didn’t I notice this before? Why did I release the demo in this state? I’ve probably fixed a bug that was hard to reproduce, but players are still dropping off at the same spot… It’s frustrating. Even with micro-level tracking using Game Analytics, I can’t stop thinking about why this part is so difficult. I wish I could just watch one video of someone giving up at that point. I might try asking for help on Discord.

For some reason, I’ve had a surprising number of visitors from Russian-speaking regions over the past few days. They left helpful bug reports and feedback—I’m very grateful for that. But now that the wave has passed, I feel strangely empty. Oddly enough, there’s also been a lot of traffic from Israel, though I’m not sure why. I still can’t properly support right-to-left languages like Hebrew, which is unfortunate. But I’ve gotten used to the process of adding new languages in general—if it’s not RTL, I can usually add support in about two days. Last time, there was a spike in visitors from Latin America, so I added both European and LATAM Spanish, but they haven’t returned since. Same goes for Germany and France. Maybe the shift in genre affected that.

In my mind, I’m already designing the next stage—Threshold—and imagining the events that will unfold there. But in reality, I haven’t even started finishing the current stage. I just keep staring at the graphs and adjusting things over and over. I really need to switch gears and move forward. 

The attached footage shows Node Group 1 and the ground level of the Archetype Server. I’ve been decorating these areas with a thought in mind: what if we looked at a massive computer from the 1970s the way we look at a towering old tree—what kind of feeling would that evoke? The servers on the ground floor of the Archetype Server follow the same idea. I think of it as a process of transferring the feeling one gets when standing before a giant tree. I always come back to the same image when I work on this game: a tree whose roots reach into hell, and whose branches stretch into heaven.