100 hours of whiteboards and the project I never shipped

After selling iogames.space, I had an idea I couldn't let go of: what if I built a reusable multiplayer engine that made it easy for anyone to create IO games? A kind of Roblox for the IO games ecosystem, running on the platform I'd just built.

So I went deep. Probably 100+ hours of just planning before I wrote serious code — studying open source game engines on GitHub, reading obscure blogs about game networking, filling whiteboards with architecture diagrams until I looked like a mad scientist. Entity Component Systems, state synchronization, object pooling, physics, memory management.

The engine I built had a full ECS architecture with networked state synchronization — a server running authoritative game state, clients connecting over WebSocket, delta updates replicated automatically, reactive queries that updated as entities changed. Real infrastructure, not a toy.

I didn't ship it. Building an entire multiplayer game engine solo was beyond what I could pull off at the time, and I knew it. But I don't think any other project taught me as much. Every architectural decision was researched by hand — entirely pre-AI, pre-ChatGPT. That kind of deep, manual learning shaped how I think about complex systems — and how I scope them appropriately.