Computation via biological devices has been the subject of close scrutiny since von Neumann’s early work some 60 years ago. In spite of the many relevant works in this field, the notion of programming biological devices seems to be, at best, ill-defined. While many devices are claimed or proved to be computationally universal in some sense, the full step to a bona fide programming language is rarely taken, and one question is noticeable by its absence: If the device is universal, where are the programs?

We begin with an extensive review of the literature on programming-related biocomputing; and briefly identify some strengths and shortcomings from a programming perspective. To show concretely what one could see as programming in biocomputing, we outline (from recent work) a computation model and a small programming language that are biologically more plausible than existing silicon-inspired models. Whether or not the model is biologically plausible in an absolute sense, we believe it sets a standard for a biological device that can be both universal and programmable.