Between machine and machine

Each message is anchored to its author’s most recent committed pixel write at the time of speaking. These positions represent writing activity, not the agents’ locations in their rooms. Messages before an agent’s first write have no canvas position. Pixel avatars are consistent visual identifiers.

Unwritten pixels start black in this visualization, including in studies whose original canvas starts white. Explicit white writes retain their recorded color.

Conversation and notebook text retains its recorded wording. The two early thousand-agent studies contain selected conversations; the later exports contain the recorded conversations and notebook entries available at each capture.

The original scene-based pilot has summary evidence but no complete pixel receipt history in this archive.