A typical control loop consists of a plant, a few sensors taking measurements of observable variables of the plant, a controller which uses the gathered measurements to compute future inputs to the plant, and, one or more actuators which apply the inputs prescribed by the controller. Networked control systems have some data flow links with limits on how often data can be sent along those links. Such a situation arises in industrial PID control over wireless sensor networks. There, if we can design the control and communication protocols well, we can lessen wiring costs and ease up the placement and maintenance of sensors. These benefits might occur in relatively `slow´ systems found in process control, as well as in not so slow systems robot teams. Such practical uses aside, the problem of optimal, robust causal sampling might help us understand Delta modulation better, and might even give insight into how some parts of Biological neural networks work.