I picked the art direction out of three — and the winner was the brightest of them
Three directions, same brand, same contents on every board. The one that won was the only one meeting a requirement I'd written months earlier and nearly forgot.
Decisions, systems, concept art and changes published with status—not retroactive mythology.
Three directions, same brand, same contents on every board. The one that won was the only one meeting a requirement I'd written months earlier and nearly forgot.
Not every animal Brazilians love gets a field name — only the ones with testimony do. And Caramelo has it — he shows up in an old photo before he was even born.
78% of desvios came from the clock, not from luck. We published 1/4,096 and delivered 1/887. And after fixing that, I found a second bug, twenty times bigger than the first.
The Ordem's interpretation can be wrong. The server's arithmetic cannot. That line became an engineering rule, not just lore, and it cost me a split I didn't want to have to make.
One binding per account, per item. The second copy does not bind — it necessarily goes to the market.
Brazil, angelological text, Norse, Egyptian, Greek, Japanese, Chinese, and modern cryptids. Nine readings of the same hypothesis, each tested, each with a case the Ordem never closed.
The only real-time content in a game that is idle everywhere else. It is not meant to be hard — it is meant to have mechanics.
Two of the strongest figures in Brazilian folklore failed my own cut test. I'd rather leave them out than look like something I'm not.
A party of up to three who never meet on the same screen — and the absolute rule: offline does not count.
The animal head in Egyptian iconography was an artist's convention, not a literal description. That one insight saved two entire incursões and nearly made me commit the next mistake.
Impossible arenas where two incompatible themes share one room. What you earn there does not leave there.
Índice, Vigília, and Raiz don't give any buffs. They're three divergent readings of the same night in 1751, and the game was built so none of them ever wins for good.
Species that ignore each other outside coexist without dispute inside — and react as one when someone descends.
Fourteen people in a room. Four died, three disappeared without ever appearing in a death record, seven survived and never agreed on what they saw. That night is what holds up the entire game.
In an idle game, making the player go find the content is a design error. Every quest opens with a server-side popup.
The most expensive rule to maintain is the simplest to write: your esva never resets. Everything else about the incursion cycle was designed around that one sentence.
A stuffed inventory is not depth. It is an attention tax — and we decided not to charge it.
Renaming eight stations, four elites, and one champion doesn't change the structure. It took someone pointing that out in writing for me to throw the draft away.
Trading between players is not deviant behaviour: it is the game. Money comes in; money does not go out.
"Absence never punishes" is easy to write in the manual. The hard part is picking a defeats-per-hour cap and not backing off when it starts to feel low.
Vigour does not regenerate. Without an irreversible cost, breeding becomes a factory and rarity dies.
I thought I was picking three classes out of personal taste. Only later did I realize the matrix and the roster had already decided for me, months earlier.
The nine bonus categories, the single engine that composes them, and why a cap you cannot prove is not a cap.
I flipped FRASCO ZERO backwards, invented a "Frasco Absoluto" that never existed, and split apart two things the manual already said were one. The closed lexicon only exists because of these four mistakes.
The math that proved I didn't even know how many attack animations I had — and why that, not nostalgia, is why I dropped Pokémon.
Agents write it; a person audits it. That only works if the code is born readable for audit.