One Idea: What Happens in One Evolution Step

"The matrix evolves" can sound like magic. It is a short, repeatable loop of ordinary arithmetic. The wider picture is in part 01.

Here is one step, in plain words, following the QxBin Connector white paper:

1. Compute the lean. Two values come from your settings: frac = biasⁿ and tail = (1 − bias)ᵐ.

2. Build direction. Those values generate coordinate vectors.

3. Make a new matrix. An outer product of the vectors creates a fresh grid of contributions.

4. Blend. The new grid is mixed into the current state of the Binary Probability Matrix.

5. Normalise. Everything is rescaled so the probability still sums to 1. Mass is conserved: when one region gains, another loses.

Repeat this for the number of steps you set. Then comes the measurement: the system selects the highest-probability combination of cells and assembles one brief from those atoms.

No hidden oracle sits in the middle. The lean, the blend and the normalise step all run on classical hardware, and because it is all classical arithmetic, every step can be logged and inspected.

Soft door: see what seven steps of this loop hand back on a real goal. Run Edge from Members.