The development of alternative statistical models for flicker (1/f) noise in a framework broader than that of numerical probability. Such models are needed for flicker noise phenomena that occur in stable, time-invariant systems, yet exhibit power spectra that are not attributable to stationary processes in the usual sense of the term. Our experience with the area of interval-valued probability leads us to speculate that the sought models may be found there, and we put forward systematic procedures for searching that area. This is one of the rare instances in which the usual concept of numerical probability fails to yield satisfactory models, and an alternative concept seems more promising.
