To the degree this devlog is useful to anyone, I should also talk about when I drop a project, which is something I do a lot.

From checking the logs, I started working on Emoji Ball Battles on January 19th 2026 and stopped on February 26th 2026. That's 1 month of work.

I was pretty confident about this game early on, after I got the basics of ball bouncing working, so much that I said this on February 4th 2026:

This is by far the project I worked on that I’m most confident about. The question of whether I’ll finish it feels ridiculous, the game is already finished, it’s only a matter of time until it becomes reality.

Ah... my arrogance truly knows no bounds.

Well, it was the project I was most confident about ever, that much was true, but quickly after this point I became increasingly less confident in it.

This happened for two reasons. The first was that implementing each weapon increasingly felt like a lot of attention to detail that the robot couldn't really help with, especially juicing detail. This isn't an issue because I actually like doing this kind of work, but... given that we now have these bots that can do anything, it felt like a waste of time to use them for tasks that they specifically can't help much with when I could be using them for better tasks in games more suited to them.

And the second reason was that getting each weapon to feel "right" became hard. I ended up implementing 7 weapons I was happy with: dagger, gun, sword, bow, boomerang, kitchen knife and CD. Many of those took a lot of iteration until I was happy with them, and it really felt like for multiple of them I wouldn't be able to get them to feel right and balanced at all. And then I also implemented 5 weapons that I wasn't happy with and scraped: sword (v1), flute, lollipop, saw, candle (untracked on logs).

To implement 12 weapons, each taking a lot of work, and only end up with 7 good enough ones, and with the game already being pretty hard to balance as is, it became clear to me that this game would take a lot of iteration to get done if I wanted to add modifiers/items/passives on top of it all.

Additionally, because this uses the emoji style, I have to create new types of visual effects that match the style, and also hunt for sounds that match each emoji. This is an additional burden that the AI also can't help with and it's just a lot of work. Essentially, the more I worked on this game, the more it became clear that the benefit from the AI on it was fairly low (but definitely positive) but that the game's design also meant that I'd need a lot of design experimentation.

During any other time I'd be fine with this. But as things stand right now, I want to finish a game quickly. That's my primary goal. And so I decided this game was no more and I'd work on something else.

This something else ended up being a build-heavy game with SNKRX-like visuals. The visuals are simple, which means little effort when adding new things. The game is fantasy coded too, like SNKRX, which means little effort when finding sounds for each ability. And it's a build-heavy action game, where the AI really shines with helping. Currently I have 15 abilities and ~80 modifiers, and the AI is really good at keeping track of what should affect what, and when we add a new ability or modifiers, which previous abilities and modifiers it should affect or not. This is the kind of game that the AI was built to help with and so far, compared to Emoji Ball Battles, it has definitely helped waaaaaaaaaaay more.

As an example, a simple beam attack with mods spreading multiple 3, beam length++ and homing, creating the homing Brimstone/Technology-like effect:

And an orb build, with orb contact damage and orb movement control to a point:

These are two things I implemented today in a session, along with many others to make up these archetypes. In this game, per session I end up implementing entire archetypes, so multiple abilities and modifiers. Compare this to barely implementing a single weapon in the emoji game.... There's no comparison. It's just way more productive and fun here, so that's where I'll go.

No work is fully wasted, though. I do want to release an emoji visuals game in the future, so any work I did for EBB here can be reused for other projects, especially given that it's all logged now so future instances will have an even easier time extracting relevant information from the codebase.

It always feels kind of bad to drop a project, but it also doesn't. At least I only wasted 1 month this time. I just hope I can actually finish something this year, I really want to.

Also, apologies for the website being down a few weeks. Apparently the AI logs were ruining something about blot.im's indexing and the dev had to spend what I assume was a huge amount of effort fixing things. Apparently way more people than I expected are reading this blog because I got a few messages about it! I also got way more people than I expected that wanted to help with testing EBB, from when I asked in the previous post. Well, everyone who asked for EBB I assume will also be happy to help with Orblike xD As for anyone else, message me on Discord @ adn327ex if you'd like to help too!