PsyNet is build on top of the Dallinger framework. Dallinger is a framework for developing online network-based experiments, allowing researchers to run complex cultural evolution experiments and human-in-the-loop experiments online. Key features of Dallinger include sophisticated code for deploying online experiments onto Heroku webservers, an advanced system for representing network-based experiments as graph-based structures, and excellent integration with Amazon Mechanical Turk.
PsyNet provides several levels of abstractions above Dallinger that make it much more efficient to develop advanced experiments. One key feature is the timeline, through which the experimenter specifies the order of events within the experiment. Using familiar constructs from other programming languages (e.g. for loops, while loops, conditionals), the experimenter can construct very complex procedures in an intuitive and readable fashion. Moreover, timeline components can easily be wrapped into self-contained components (e.g. functions, classes) which can then be distributed and reused in other contexts.
Our long-term goal is that people should be able to design and run PsyNet experiments without knowing or worrying about the details of Dallinger. For now, however, the abstraction is still a little leaky, and you might find yourself running various Dallinger commands as part of your experiment implementation workflow. It is worth having a little look at the official Dallinger documentation to get a feel for this underlying framework.