This study reports findings from an experimental field study of scripted collaboration for net-based learning in the context of a one-semester university course on operating systems. In scripted collaboration, activities of learners are coordinated and guided according to particular rules, implemented via respective tools in the learning environment. Forty-two distributed groups of three students collaborated on five successive assignments employing the virtual learning environment CURE. Three collaborative tasks--brainstorming, clustering, and essay writing--were implemented as scripts with dedicated tools guiding the net-based collaborative process. Half of the groups collaborated via scripted task versions, and, as a control, half of the groups performed the tasks in a non-scripted manner. No general advantage of scripting was found concerning acquisition of knowledge; nor was overscripting observed. Collaborative scripting appears to be neither generally advantageous nor disadvantageous, but highly contingent on the particular content and task under consideration. Results suggest that scripting might be slightly more supportive in more complex tasks such as essay writing, in contrast to undemanding tasks such as brainstorming.