One of the main challenges of writing research has been its generative nature. As a production task, writing cannot be controlled as easily as other psychological laboratory tasks, such as reading. In this article, we present an analysis of 3 writing tasks, ranging from a simplified typing task to an ecological writing task. We show that multiple nonlinear metrics can be used to systematically characterize open, complex writing tasks. Moreover, we demonstrate that, when writing performance is quantified in such a way, a continuum of complexity can be identified on which increasingly complex writing performances can be placed.