Karhin’s Blog
Apps, Design and Music
It's me!

Hello hello

I'm Mikalaj Karhin

I create apps and music, and write about both.

Telegram Instagram

AI effect

I remember how fascinated I was by the first version of DALL·E, back when it was still invite-only, and the shock it gave me. The world had changed! It seemed to me that it was for the better.

Over these past few years, generative models have become much better and more natural. The results are now almost indistinguishable from real photographs. And as for drawn images forget it. Try now to tell apart hours of human effort from something generated in a second.

I don't know if this is good or bad, but psychologically it's easier for me to perceive and trust something made by a person. There's value in it: human labor, effort, creativity. And all of that was devalued in an instant. I would rather computers take away monotonous, mindless work from people, not their creativity.

In short, I'm biased against all creative content produced by AI. And it's starting to play a cruel trick on me, when I mistakenly take human work for a machine's output and devalue it.

The other day I was rewatching Spider-Man from 1994. It's one of the most classic Spider-Man cartoons. Back then, such graphics weren't possible, the standards for visuals were lower, and sometimes the artists saved time where they could.

– You're AI. No, you're AI. In reality, this is a frame from the 1968 cartoon, but I couldn't resist inserting it here.

And the joke is that if you showed these frames today to someone who hadn’t seen that cartoon, they would most likely say it was drawn by some cheap, low-budget model. A classic was devalued in an instant.

Cartoons have already been taken from us. Ordinary videos are almost gone. Music will be taken tomorrow.

New Design

Wow! You're now looking at the updated design of my personal website™! I haven't changed anything here in a long time. Thanks, as always, to Timur for helping with a bit of redesign.

There was one problem with the previous design. It was too tied to graphics, and I got tired of making fun pictures. I even thought about adding something like a microblog here to make posting easier, but I changed my mind in time 😁

This design is now part of history, unless I suddenly decide to roll it back.

As part of the redesign, the look of the main feed has changed (now posts are displayed in full), the tag feed layout was updated, fonts were changed, and some other small details too. I think more changes will come along the way.

Lately, I've been writing a lot on Telegram, but they cut off web access to my channel a long time ago. On top of that, they can delete or block an account at any time (and there's no such thing as support there). I wouldn't be surprised if they keep moving further toward restricting access to the platform.

And the truth is, Telegram isn't really used in the English-speaking world. And if someone does use it... Well, you know what they're buying there. Besides, Telegram is designed so that it's basically impossible to discover content.

At least on the regular web, you can still compete in search with throwaway websites stuffed with LLM-generated content. Though I don't even know if anyone still goes out onto the regular internet. We're living in some kind of cyberpunk world with a Blackwall made of social networks that don't let anything out.

Social networks in general have gone downhill. It's all dopamine, clickbait, and content that becomes outdated before you even hit "publish". But you already know all that yourselves.

And there's a lot of work ahead. I need to transfer all my old posts from Telegram here.

A small unethical experiment with (un)expected consequences

I had a discussion with an old friend. Making even simple good music isn't enough, and no one listens to good music just like that, because there's simply too much of it. Btw, I'm pretty sure that the real quality of music can only be determined over time. And one of the most important metrics is relistenability. Not how hard the bass hits, how cool the piano part is, lofi/hifi, all of this is only artistic characteristics, but that's a topic for another time.

My initial hypothesis was that without some sort of "bait", no one will appreciate or listen to a track in 2025. Even your acquaintances and friends don't care about it. No matter how good it is, how much soul you put into it, etc.

You know, I'm something of a scientist myself, and I prefer to test hypotheses with numbers, not just "that's how I feel", so I decided to either confirm or disprove it.

I love posting demos on Instagram (this is still the main social media, right?). I've long noticed that people just don't listen to them. Nobody cares, even though you worked so hard. That can even make you feel sad, honestly!

At this point someone will say "бля, Калян зноў крынжа наваліў" or "well, your music is shit, that's why no one listens to it". You can't really argue with that. At some point you might even start to believe it. I probably would too, if I didn't have the numbers.

So, to test the hypothesis that music itself is secondary to whether people even give it a first listen, I decided to do something unethical. I simply took a track by a successful producer with a million plays in a similar style, and... posted it as my demo. I specifically chose a track that they most likely hadn't heard before.

What happened next? Nothing. (Honestly, I was hoping there'd be some kind of reaction). If we take the Instagram Stories "like" slider as a metric, it was a complete failure. The track got fewer reactions than my own demos. By that metric, this demo shouldn't have gone anywhere.

So what's the conclusion? Nobody cares that you wrote a "hit". Basically, making good music (or a product in a broader sense) means nothing. Presentation and marketing are just as important, if not more so. If you're starting something new, always keep this in mind. In 2025, nothing takes off on its own. Either you’re very lucky (statistically, your chances are slim, sorry).

Unreadable Fun

I can barely remember the last time I actually drew or designed a cover, instead of just using a photo with a logo. Maybe around ten years ago.

Usually, covers are supposed to show the title and the artist's name, but whenever I try that it just looks boring. So I've been sticking to only using the logo. To be honest, most covers these days look boring anyway. And it got even worse once people started making them with ChatGPT. You can tell right away, and it feels kind of cheap.

This time I wanted to make something fun. A cover you can look at for a bit longer and actually read. Some people might say "who even reads that stuff"? Well, that's exactly the point! I wrote it in Belarusian Latin, so only a few people could even read it. And maybe they'll smile, the same way I did when I came up with it. It would also be cool if someone from another country and culture decided to find out what's actually written here and what language it is! I like things like this, little puzzles. If I enjoy them, then someone else out there in the world will find them fun too 😁

Of course I should also add that more people from Australia or Indonesia listen to my music than from Belarus. Sometimes I think about that and it makes me a little sad. What do you even call it when you feel like you "don't care", but at the same time it still kind of hurts? Ah, whatever, screw it.

Here are some covers I made almost ten years ago (2015-2016). A bit naive in places, but fun to remember.

Apple’s New Glass Design

It seems I've figured out what the design transition between iOS 6 and iOS 7 felt like. The only difference now is that there isn't a global paradigm shift, there's no need to remove shadows and textures and replace them with solid colors. And I think that's where the big problem will be.

The end of an era

If you want to do it properly and care about backward compatibility, you still have to duplicate and rework all components for the new iOS version. Glass design, after all, implies turning pretty much all major interactive elements into cards and circles. There are no longer components that blend into the edges of the screen (mainly navigation elements).

Old components were easy to create yourself because of the minimalism. With the new design it's no longer such a simple task because of all the tricky special effects.

If it's a non-native app, achieving the glass effect and a "native" feel will be a very difficult task. In some cases, impossible.

So it turns out that nothing has really changed globally, and there's very little point in reworking anything, especially considering how hard it's to support both the old and the new. Particularly in a situation where development and design are being done on a tight budget.

I have a bad feeling that most apps will stay in their current design for a very, very long time.

Earlier ↓