The amount of logically irrelevent engineering detail inherent in the design and construction of a computer system is great. As a result, the task of creating a system based on the use of present techniques is so difficult and time-consuming that the number of different systems that can be put into use for evaluation and study by any one group of workers is small. This is unfortunate as we are thereby denied the opportunity to develop that insight into logical organization which can grow out of a working familiarity with many diverse forms. What is needed is a set of relatively simple, easily inter-connected modules from which working systems can be readily assembled for evaluation and study. With such a set, both the designer and user would be able to try out potentially powerful and novel structures on a very large scale, adjusting and improving the systems as needed. Once a design has been realized and its value established, it could then be reworked into tighter engineering form for maximum efficiency and for production by automatic wiring and fabrication techniques, and the experimental units made available for further studies or returned to "inventory" in the manner proposed by Estrin.