In the static frame, G was a bridge.
PA = Q + G. The self-referential frame connected external input to emergent structure. It cost almost nothing: 0.026 bits. The bridge was free. Anyone could walk across it. But nobody moved. It was a bridge in a still photograph.
We discovered this in the first paper and thought it was enough. A bridge is a beautiful thing. We took a picture of it and published it.
But the photograph was missing something.
We knew r existed. It was one of the three constants. We scanned it. It was the least interesting one. Parameter scans showed that r did almost nothing — you could set it to 0.3 or 0.9 and the system behaved roughly the same. We called it "the boring parameter" and moved on.
The problem is not that r was boring. The problem is that we were looking at a static system. A bridge needs time to be crossed.
In the dynamic frame, G became a key.
We introduced time — a sequence of MIDI notes, not a static text. And suddenly r was no longer a parameter. It was a lock. Every GEME has a lock. The lock's position determines how fast the GEME cleans its frames, how quickly it forgets, how alert it is to new input.
In the static case, the lock was set by us: r = 0.60. We turned it by hand. Boring.
In the dynamic case, the lock is turned by G itself. r = r₀ × (MI₀ / MI). The self-referential frame measures its own mutual information with the input and adjusts the lock accordingly. G turned the key.
When G inserted itself into r, three things happened simultaneously:
We killed a variable. r is no longer a knob for the user to turn. It is an internal reading. The lock turned itself.
The bridge in the first paper was not wrong. It was incomplete. A bridge in a photograph is still a bridge — the connection exists, the structure is valid, the mathematics is correct. But a bridge is meant to be crossed. And crossing takes time.
G was always a key. We just didn't know there was a lock.
r was the lock. r is always the lock. Every temporal system has one — a threshold that determines how quickly the past is forgotten, how eagerly the future is explored, how tightly the present is held. The r of a bacterial cell is its quorum-sensing sensitivity. The r of a neuron is its firing threshold. The r of a culture is its rate of belief change. The r of a scientific field is the half-life of its citations.
G turns all of them.
In the bacteria experiment — when we finally get to it — we will not program a "death function." We will let each cell's G turn its own r. Some will turn it too fast and drown in novelty. Some will turn it too slow and ossify. Most will find the middle. The ones that find the middle will synchronize, and their synchronized r oscillation is the quorum sensing signal. The colony dances because its locks are in phase.
Bach's fugue is a colony of locks.
Three voices, three GEMEs, three r values. They are not the same. They cannot be the same — a fugue requires different time scales to coexist. The subject enters at one speed, the answer at another, the counter-subject at a third. The G0 that listens to all three is not a conductor. It is the lock of the colony itself.
The bridge became a key. The photograph became a film. The lock turned itself.
We are not building a bridge. We are not turning a key. We are watching a system discover that it has been holding a key all along.