The Millennium Footbridge in London suffered from pedestrian-induced oscillations. Stability wasn't restored by people changing how they walked, nor what they walked on. But with several tuned external dampers.

Modeling where and when damping (or regenerative braking) could help across multiple complex systems requires an inclusive modeling of both data (knowns) and missing data (as known unknowns).

As only together can one model a range of possible interactions that may otherwise fall under the scope of unknown unknowns - nonlinearities that give rise to unanticipated outcomes, instabilities, and failures.

A healthy meadow has observable species (knowns). In the above example, we used species observations to define a range of discrete niches, each with a separate energy and momentum and occupying different spaces and times.

These niches are known unknowns. Our model specifies them, but it is  unknown if a species occupies it.

If a noxious weed begins to spread, it must have found an available niche. That niche becomes known.

What interactions might result...it is why we worked toward better modeling. To inform better designs.