Day Zero: Leg Day
The whole thing started on April 1, 2025. I was at the gym with Albert, my co-founder, and we were doing legs. Between sets we were complaining about a prior project we'd both worked on, the kind of conversation you have when you're pacing around with a barbell waiting for your heart rate to drop.
The frustration was specific. That earlier project would have actually worked if it had a shooter in it. It had everything else. It had a world. It had users. It just didn't have a reason to log in once you'd looked around. We were standing there saying, out loud, "if this thing had been a shooter it would have shipped."
That's the moment. April 1, between leg sets. No deck, no investor pitch, no whiteboard. Just two guys annoyed enough about a dead project to start a new one.
Day 1: The Pitch Deck
The first thing I did, the day after the gym, was sit down and make a pitch deck.
That sounds boring for a dev log, but it's the truth. I didn't open Unreal. I didn't sketch a level. I wrote out the ideas in slides. I needed to see if the concept survived being explained to other people before I asked Albert to spend his nights on a game engine instead of his architectural visualization work at AJ Studios.
The pitch deck was the first filter. If I couldn't make the case on a slide, the case wasn't real. It was real. We kept going.
The Hackathon Clock
Months of slow-build later, the actual 23 days started when we signed up for the Colosseum Cypherpunk Hackathon. Submission deadline: October 30, 2025. We had 23 days to take an idea on a slide and turn it into a build that other humans could log into.
The plan was simple. One map. One weapon. Mineable resources. Multiplayer. NPCs to make it feel populated while we got real players in. Extraction loop. Ship.
Then Epic broke our game.
Six Days We Did Not Have
An Unreal Engine update came down mid-sprint and put our entire multiplayer setup in the grave. Connection logic, replication, the lobby flow. Whatever we had working on a Tuesday was broken on the Wednesday. The build wouldn't host a session.
If you've never been on the wrong side of an engine update with a deadline in 17 days, here's what it looks like: every system you wired up is now interacting with a slightly different version of the thing it depends on, and the only way out is to chase the change through every layer of your code. Then test. Then watch a different thing break.
It ate 6 days. We had 23 and we lost 6 of them to a problem we didn't cause and couldn't have predicted. Roughly a quarter of the sprint, gone, scrambling to put back what was already working.
That was Albert in the chat at one point during the patching. I left the message in. He's a 10-year vet of game dev. He's seen builds break before. The Epic update was the worst stretch of the sprint and he said the quiet part out loud.
What We Got Wrong: Multiplayer
If I had to name the one thing we got wrong in the first 23 days, it was the multiplayer architecture itself. Not the design of it. The bet on the abstraction. We built it on top of pieces that the engine update was about to walk into. When that update landed, the bet looked obvious in hindsight: of course you don't put your most fragile system on top of the layer most likely to change underneath you.
We lived with the bad version for the entire window between the update and the rebuild. Six days of limping. The lesson wasn't "Unreal is bad." The lesson was: when you're sprinting, the parts of your stack you don't control deserve more pessimism than the parts you do.
Day 23: What We Actually Shipped
By the deadline, we had a build. It was super buggy. It was also, honestly, pretty great.
The numbers from the sprint:
Five lobbies running concurrently at the peak. One Sol-themed weapon, with mineable crystals scattered around the map. NPCs to fill the world out, and real players alongside them. The full extraction loop: drop in, mine, fight, get out with what you mined, or die and lose it.
Was it polished? Not even a little. Guns occasionally vanished mid-match. Things desynced. But the loop worked. People were genuinely playing it, not just opening the build to be polite. We placed 3rd in the Cape Town local bounty. Here's the video of the announcement.
We didn't make the global accelerator. The community feedback we got was the more important result. People liked how it felt. They liked that it didn't look like every other Web3 game. The concept landed with the people we wanted it to land with.
What 23 Days Actually Buys You
23 days does not buy you a finished game. It buys you something more useful: proof that the idea survives contact with reality. We learned the loop was fun. We learned what the engine would and wouldn't tolerate. We learned which parts of the stack we'd been too optimistic about. We learned that we could ship under pressure without the team breaking.
Most of what VS Protocol is now traces back to the decisions we made during that sprint. The single-gun discipline. The map-as-first-feature instinct. The multiplayer paranoia. The willingness to keep a build live with real players inside it even when nothing was finished. We didn't write those down as principles. We learned them by losing six days to Epic and shipping anyway.
If you want to know where this game came from, that's where. Two guys at the gym, a pitch deck the next day, and 23 days of sprinting through a sprint we didn't fully control.
// PAUL, CO-FOUNDER