Working generatively can be a completely manual process of determining and specifying rules, and then exploring the possibility space they produce, or a heavily automated process where machine learning technologies take over these tasks, in turn creating a new role for the computational designer. However, more likely is that a generative process lies somewhere between these two extremes; designers pick up or discard digital tools as needed, always keeping in mind the agency that software and algorithms bring to the process. Generating is not a simple, cause-and-effect method with a ruleset deterministically leading to a collection of outputs, from which one optimized solution is selected. Rather, it is a complex feedback loop and negotiation between the designer, their tools and systems, and the data they create or provide. As shown through the Worldmaking case study, the process becomes even richer when multiple human designers work together in a dialectic conversation, generating new possibilities through productive (mis)communication. Working together, and with critical engagement with their tools at various levels of abstraction, designers may subtly and tangibly shape the holistic properties of systems.