DuckWeights field notes
Original accounts of working with AI: the method, the mistakes, and what the results actually show.
Hardware lab · Field notes · September 22, 2026
Setting up a Flipper Zero on Ubuntu, building a local lab interface, and learning why simulated test results need a clear boundary from hardware evidence.
Read the field report →Etherbound · Development journal · Day 4 · September 8, 2026
Day 4 of the Etherbound development journal: two new playable Earth abilities, a broader Fire sweep, and the first paired whitewater-wave study, with each result kept to its current evidence.
Read the field report →Etherbound · Development journal · Day 3 · September 6, 2026
Day 3 of the Etherbound development journal: making three Earth abilities responsive as a combo, then defining a fair way to measure Water’s more demanding runtime work.
Read the field report →Etherbound · Development journal · Day 2 · September 5, 2026
Day 2 of the Etherbound development journal: moving a measured Blender study onto the existing Unreal mannequin, then testing two Earth abilities in one playable sandbox.
Read the field report →Etherbound · Development journal · September 5, 2026
Exploring AI-assisted animation through structured evidence, synchronized audio, and measurable constraints—with selected Python examples from the ideas behind Etherbound.
Read the field report →