Devlog - Odyssos

Odyssos is the first game I’ve developed past any serious threshold and actually released to players. It’s a browser incremental, so there was no Steam release to organize. I just wanted it in front of as many eyes as possible.

It took the better part of a year, built with GPT-3.5 and GPT-4 plus a lot of manual coding. It’s one of the last projects where I hand-wrote a large share of it, before AI got good enough that doing that on a project like this stopped being necessary.

How it actually went

Three months of initial build. Six months away on other work and other projects. Three months back on it once I decided I wanted something published, because I’d never released a game. Then I felt myself losing steam, so I gave it one week of cleanup and put it out.

Reception

Mixed reviews. The negatives were almost entirely about balancing, which is correct, because the balancing is bad. The praise was about the gameplay, and it pulled real logged hours across a couple of incremental game sites. That’s the part I cared about.

Most of the feedback in my games Discord was some version of “this game explodes after level X.” I already knew. It was still useful, because now I have actual playtesting data instead of my own guesses about where the curve breaks.

What I got right

A lot of work went into the UI and the tutorial pacing, specifically not dumping a thousand menus and options on someone the moment they arrive. I hate that, and I believe most players hate it. A gently unfolding interface was the plan from day one and I think I pulled it off.

The other thing is that I finished it. For a first released game, and one of the first projects I’ve pushed past the twenty percent mark at all, I’m proud of that on its own.

What’s still wrong

The balancing problems aren’t only curve tuning. Some of them are in the actual calculations. A current model would annihilate that cleanup work if I sat down with it properly, which means the only real blocker is me setting aside the time. Another month of focused work and I think it’s near feature-complete and balanced.

There’s nothing especially notable about the architecture. This wasn’t the clever-systems project. This was the make-a-real-game-and-ship-it project.

Next

Do the real balancing pass, curves and calculations both. Polish until it’s well received across the board instead of split. Then port it to Android and monetize it a little, which would be nice, but isn’t the goal.

The goal was proof. I can make something that a lot of people genuinely enjoy, and I’m not crazy for thinking my ideas are fun.

You can play it at odyssos.io.