Uncertainty is far more than error bars, it is an instance of propagating distortion. In this descriptive and pedantic talk, I will use this metaphor of blurred images to examine some aspects of dealing with uncertainty in simulations. The approach starts with the dominant result of linear inverse theory defined by Backus and Gilbert and others that places uncertainty in its Heisenberg context as a trade-off between accuracy and resolution. This resolving kernel may be viewed as a lens through which the solution can be seen; in a simulation study with temporal and spatial components, the kernels provide a stabilization to the Kalman gain filters that properly place the problem in the realm of data assimilation or, as it's known to the engineers, control theory. I conclude this talk with some speculation about the role of visualization and computational steering in the context of uncertainty and proffer some suggestions on user interaction and metrics.