All Over the Place

It's startling that a whole week has passed since I wrote the last entry. My sense of time is definitely wrecked lately. The reason is simply that I've been having hectic days. Let me write as things come to me, as usual. First, I recently registered a business. I'd meant to hold off, aiming for a pre-founder support program in the national funding schemes, but looking at the guidance for a program I've already been selected for, you can't double up on benefits with the same item anyway. So I hurried up and finished what I'd been putting off. Incidentally, Korea's startup support system is a little unusual. I gather there aren't many countries where the state provides this much in startup grants and programs. Whether that's good or bad, judge for yourself.

Since the registration went through yesterday, I ran around banks setting up accounts, including one for the grant money. Walk about ten minutes, ask them to open an account, wait about twenty minutes and they handle it all, which is convenient, but the going itself is a hassle. Also, for the grant account it was a bank I'd never used before, and the address on my ID differs from where I currently live, so they need a residence certificate. I've printed it out, and I'll probably drop by again when I have time. Such a hassle.

The competition I'd been looking forward to, I lost. It's the Lumen Prize, and they announced the finalist list yesterday, and I wasn't on it. Well, looking at the video quality and project description I submitted at the time, making it would have been stranger, but disappointment is disappointment. Once I've properly wrapped up the funding I'll apply to Ars Electronica and the like. Or various HCI conferences. My desire to get Patternflow into the art world or academia, one way or another, is the same as before.

Crowd Supply feels really really really a little very slow. But I know they're a small team managing a lot of projects, so it can't be helped. Still, I want to launch soon. Because I'm anxious. The project's subscriber count is around 270 now. It's been about three weeks since hitting 150, and it's grown by roughly that much again. Looking at the graph there are moments where it jumps, and the timing matches when a new pattern reel does well on Instagram. Obvious enough, but what was slightly interesting is that the conversion rate differs depending on the pattern's style. Or maybe some other factor was at play?

Anyway, you might wonder what I'm anxious about. There's no guarantee subscribers convert straight into buyers. I have no feel for it at all, and I think it comes from the experience that only about 10% of the people who filled out the pre-launch survey converted into subscribers. Whether people buy a lot or not, I just wish the numbers would come out clearly. If it's going to work, let it clearly work; if it's going to fail, let it clearly fail, so I don't build up pointless hope or fantasies and waste energy on them. That's just my own disposition.

That said, slow isn't all bad. Moving fast while leaning heavily on AI, there turned out to be a lot of details that were wrong or missing. While waiting for replies, I could keep revising and improving and tightening things up. But when the reply gap goes past five days, it really does sting. Something I keep feeling as I go: you really do have to do it yourself. Not expecting other people to work hard for you is what makes your mind easiest.

And yet there are people who do come through! First, Martin keeps making patterns and posting them to the community. Every one is new and fun to see. He also keeps leaving comments on Instagram, and he's been cheering me on since very early. Simone also keeps developing MQTT-related features. Along the way we DM a lot, and once we talked over video on Google Meet. Was it two hours? My English isn't good, so it was a bit hard. He's a senior with a lot of experience and accomplishments, and even so you can feel enormous enthusiasm from him, which is admirable. Not wanting to fall behind him, I find myself working harder too.

Something I keep feeling: I really do have to study English at some point. Honestly, starting now is the obvious move. I have channels and people to communicate with in English, the board is all set, really, and I'm putting it off with busyness as an excuse. Ah, whatever, but I really am busy. Maybe I'll memorize a few words a day at least. There's that proverb about how even a dog at a schoolhouse recites verses after three years, and I do get used to it bit by bit like that, so it feels like a little effort could get me comfortable pretty fast…

Recently someone even sent a PR adding a feature I needed. It puts Patternflow into sleep mode, a feature I genuinely needed but hadn't done out of laziness. And since I don't actually use it for IoT or leave it running, it's a feature I kept postponing even knowing it was needed. Having other people build and propose these parts is genuinely amazing and fun. Sure, properly integrating, testing, and reconciling it with what exists becomes new work, but it's far better than building alone! Maybe in the AI era they say you can do everything alone, but for that very reason, doing it all together like this might be better. Something got crossed in the integration and there are a few errors, so I'll fix it when I get home tonight.

Yesterday I fixed the part of the firmware that had been a particular headache! Here's what it was: the ESP32's internal heap runs very short, and because of it, Patternflow would go to pieces whenever I added a feature or a heavy pattern. Until now I thought it was an unfixable limit! At the same time I never got that impression from Simone's build (with identical code!), so I was just carrying around a slight sense of dissonance. Then yesterday, at the shared office in the Seoul Design Startup Center, I was playing with Simone's new feature and testing new patterns, and I figured I'd go home and load it onto the Patternflow there (one at the office, two at home) and shoot a reel. But when I got home and loaded it onto the device, it didn't work! It had been running perfectly fine at the office? I wondered if I'd uploaded the wrong code, but no. Absolutely. Way.

So, wondering what was going on, I asked Claude Code to properly compare Simone's against mine, and the problem and the fix came out. The Arduino core version was different. That one difference was enormously fatal. At the same time I found that the slight slowdown in WiFi transmission had been a bug too, and fixing it let me improve things into a version carrying only the strengths of both firmwares (mine and Simone's). This is a really, really, really enormous step forward for me. That internal heap business had been causing me so much stress. I'd even been anxious about other people adding features and eating into it. Now there's no reason for that! Fantastic. And I added a few more things around the math, so patterns that wouldn't run before can now be attempted. Very good.

Last week I also interviewed a friend who works in sound. It's an assignment from the incubating program I'm currently in at Seoul Art Space for Youth. Since the incubating is at the item-definition and market-discovery stage, it honestly feels like a stage I've already passed, but they're giving me 3 million won, so I'm doing it diligently. They told me to define a customer persona and go interview. I already have plenty of makers around and I'd surveyed them before, so I did someone from the sound side. I learned a lot of interesting new things doing it. Of course I only did one person so it's hard to judge from that, and honestly I don't think it'll change anything before the funding… but I'd like to make a new survey for the people currently on Discord and the ones who've already built it. Or, I don't know if I really want to. I always say it's needed and I'll do it soon, but the moment I actually sit down to make a survey it gets complicated and I give up. Too much of a hassle. And more than anything, asking other people things is still unfamiliar to me.

I'm doing what I can do alone as diligently as I can. Because of that, lately I've had no free time at all. Losing my sense of time is a bonus. My YouTube time has dropped enormously too. I don't know if I'm doing well, but I'm getting through the things that need doing, one at a time, putting them off as little as possible. As part of that I also bought new electronic components. Because I don't have parts, and shipping takes a long time to begin with, so I keep using that as an excuse to postpone hardware improvements and the like. Once the launch starts I'll have to work again, so this is preparing ahead.

There's still something giving me a headache around hardware and the launch. CE and FCC. Patternflow itself is such an unusual form of device that there aren't many precedents, and Crowd Supply doesn't help much either. Honestly, I feel Crowd Supply's care on the CE side is far too lacking. They'd have experience with similar products and surely have experts, but whatever I ask, they answer at a very shallow level and tell me to reread material that's already been compiled. If that were enough, I wouldn't be asking! It stings a lot. So I sent a lot of emails to other institutions and companies, and only one place replied. Truly too much.

I've been making calls and looking into it, and it isn't very optimistic. Apparently Chinese panels almost all fail. So you have to paint them or build shielding. And the cost of the engineering debugging that goes along with it gets added too. They said it runs into tens of millions of won. Seriously… ugh, if someone close to me said they were going to build and sell electronic hardware alone, I'd absolutely talk them out of it. Going back in time to talk myself out of it… probably not, though. I'd do it anyway even if I tried. Anyway, recalculating those costs, I've decided to raise the price to $249. Can't be helped. And if people who find it too expensive build it themselves instead, that's actually good!

For once, the writing feels like it's flowing. It probably isn't a well-written, easy-to-read piece, but it feels far easier than writing the recent blog entries. And accordingly, the length keeps growing. The difference is probably that the earlier ones were written at home at night, and this one I'm writing at the startup center during the day. In my case, home is effectively my studio. That's how I've lived, so I'd keep working and then write and go to sleep when a bit of spare time appeared. At the startup center I don't really do the making, so I think I just write pieces like this smoothly. There's nothing else to do there anyway, that kind of feeling. Home, school (cafeteria), and the startup center are laid out along a line, about twenty minutes of walking in total, so the route is pretty good. So I head out at lunch, eat at the cafeteria, stay at the center, leave around five, eat dinner at the cafeteria again, and go home. Though lately there have been far more days where I didn't leave the house at all because I was working.

This feels like a pretty scattered piece. Slightly ADHD-ish? But maybe going forward, this kind of scattered disposition will help a lot. AI has expanded what you can do, and you can set one task running and do other things while it goes. Those are people who find that comfortable to begin with. Not that I know whether I have ADHD. I've never actually been tested. And once something grabs me I focus intensely, so maybe not.

Next week I also have to decide about next semester. Take a leave of absence or not… it's a real problem. At my current pace, combining studies with this is absolutely impossible. But this is the busiest period, and since I've done things in advance, if it gets a bit easier ahead, maybe doing both wouldn't be bad. For now I've loaded all the classes and made a timetable. Maybe I'll try registration and just take leave if it goes badly. There are a few project-based major courses. Making a video on a free topic, or making immersive content. I think I'd tie both to Patternflow somehow. Otherwise it's impossible. I'd actually die doing it. Tie it together well and go about it smartly.

Lately I haven't been reading at all. Even when I go to the library, after a few pages my mind wanders and I can't read. And the star of those wandering thoughts is mostly Patternflow. I wonder if I should force myself to read. Or not…

Ah, right, I have to do sound too and I've basically dropped it entirely. I do have to. English to do, sound to do. Surveys to do. Scheduling and going in for CE testing to do. It's nothing but things to do! It's incredibly grueling and hard, but I don't want to quit. Thinking about why for a moment… it's because other people recognize that I'm working hard. I steadily get likes and views through Instagram, sometimes there are grateful people who improve things and propose them, and there's someone who keeps making patterns!

I just like being recognized. When the product of my effort is recognized, I feel recognized too, and that's why I do it. On days when views drop a lot and the response is poor, my mood turns immediately. I don't know if this is artistic. The common image of an artist is someone who doesn't care one bit about others' eyes and lives in their own world doing only what they want. I'm the exact opposite, so maybe this makes me a designer rather than an artist? Or closer to a businessperson? Or maybe those distinctions don't matter at all anymore. You just go with your disposition. I used to think a lot about making a pile of money through business and becoming an artist who ignores everyone and only does the projects they want. Now I see that's nonsense. Even if I made a pile of money and got rich, I'd surely throw myself only into projects other people recognize. And for now, that's Patternflow.