Vegetation is understood as the joint occurrence of plant species within any given site. Species combinations tend to reoccur at different points in space and time suggesting the existence of assembly rules. The issue of this essay is the vegetation process, reflecting the fact that vegetation changes permanently. Finding a universal prediction theory of vegetation dynamics is considered the ultimate goal in vegetation science. The essay follows a traditional scientific scheme discussing the questions of "what" can be assessed, "how" processes operate and whether there is an explanation to "why" the systems behave as observed. We argue that sampling, apart from statistical considerations, is very much a question of choice: choice of the variables, the temporal and spatial frame. For finding the appropriate sample size, however, there are methods available to find efficient solutions. Concerning the understanding of processes, a combination of determinism and randomness has to be taken into account. Some deterministic processes, however, show chaoticity. This is difficult to distinguish from stochasticity. In this context the role of linear processes is discussed as well. Identifying causes is often beyond the reach of ordinary investigations and one of the reasons is the requirement to access interacting scales. The essay underlines the relevance of flexibility and empiricism in a holistic approach, whereas a reductionist approach will only reveal isolated properties of the vegetation process.