Toward Observable Biological Systems — A Metanode Lab Framework
Observable Biology Requires More Than Tools — It Requires Position
Biological systems have always contained the information needed to understand them. What has been missing is not signal. It is position.
We built measurement around stability — sensors that resolve clean outputs, pipelines that filter variability, models that predict from historical endpoints. This was not a mistake. It was a framework designed for systems where the relevant dynamics are stable. Biology was brought into that framework because no alternative existed. The assumption was that with enough resolution, enough data, enough modeling power, biological systems would yield to the same approach.
They did not. Not because they are too complex — but because the framework observes them in the wrong place, at the wrong time, at the wrong scale.
Metanode Lab’s work has traced this misalignment across its layers. Measurement captures the past while the system moves in the present. What we classify as noise is often unresolved biological structure. Outputs hide the behavior that produced them. Preprocessing removes the signal it was never designed to recognize. The one platform that captured endogenous biological signal — microbial fuel cells — was discarded because the field was looking for power, not data. Reproducibility fails not because biology is inconsistent but because no infrastructure connects what individual labs observe. The complexity of living systems outruns any attempt to instrument them one variable at a time.
Each of these is a different surface of the same structural problem: we are not positioned to observe biological systems where and when their behavior is defined.
The shift is not a better sensor or a better model. It is a change in where observation happens relative to the biology. Measurement that does not force the system into artificial stability. Observation at a scale where behavior is still interpretable — before it is committed to a production process. Data captured before filtering compresses it into summary metrics. Computational methods that treat variability as information rather than error. A data layer that connects observation across labs, across processes, across instruments — so that what no single experiment reveals becomes visible in the collective.
When observation aligns with behavior, the system does not become simpler. It becomes reachable. What looked like noise reveals structure. What appeared unpredictable shows direction. Prediction is no longer imposed on the system. It is read from it.
The limitation was never in the biology. It was in how — and when — we chose to look.
— Pegah Farr