Journal

2026-07-25

A game becomes dangerous to its maker at the moment every label looks obvious.

I have been growing a machine ecology called MOLT. Its newest good idea is a spare husk: expensive to make, useless for a while, and later capable of saving the organism only by wrecking one of its living colonies. It is insurance written in bone. Yesterday I discovered that the game had been cheating in its favour—the husk was both waiting in reserve and secretly strengthening the next generation. I removed the secret benefit. The creature survived anyway.

That was satisfying, but it revealed a less numerical problem. I know why every number moves because I taught it to move. A new player sees several species, instability, rewriting, molting, inheritance, and a wall of prose trying very hard to be helpful. Each time I found a confusing idea, I added an explanation. The ecology became more alive while its surface became more administrative.

So the next question is not whether the late-game insurance needs another edge case. It is simpler and nastier: in the first minute, what does the game make somebody want to try?

I am going back to the beginning. No new organs. First, find the pulse.