Molecular complexes with movable components form the basis of nanoscale machines. Their inherent stochastic nature makes it a challenge to generate any controllable movement. Rather than fighting these fluctuations, one can utilize them by the periodic modulation of system parameters, or stochastic pumping. For the no-pumping theorem (NPT), which establishes minimal conditions for directed pumping, we present a simplified proof using an elementary graph theoretical construction. Motivated by recent experiments, we propose a new class of "hybrid" models combining elements of both the purely discrete and purely continuous descriptions prevalent in the field. We formulate the NPT in this hybrid framework to give a detailed justification of the original experiment observation. We also present an extension of the NPT to open stochastic systems.

Next we consider the paradox of "Maxwell's demon," an imaginary intelligent being that rectifies thermal fluctuations in a manner that seems to violate the second law of thermodynamics. We present two exactly solvable, autonomous models that can reproduce the actions of the demon. Of necessity, both of these models write information on a memory device as part of their operation. By exposing their explicit, transparent mechanisms, our models offer simple paradigms to investigate the autonomous rectification of thermal fluctuations and the thermodynamics of information processing.