An alternative approach to stratigraphic analysis is proposed. It takes as its starting point the simple general characteristics of the stratigraphic record and the complex, chaotic behaviours of the systems that combine to generate this record. In this context, Smith’s Stratigraphy Machine (SM) concept is developed. The SM is a chaotic global system, combining the operations of the atmosphere, hydrosphere, lithosphere and asthenosphere, which everywhere tends to self‐organize towards a critical state in which continuous loading of the system with rock waste, over time, is balanced by the unloading that creates the record. The outputs of the SM are simple and nonscale dependent in character, with residence times in the record ranging from a second, or less, to 1 billion years. However, only those long‐lived in the human time‐frame are judged to be of stratigraphic significance; and the range of residence times entails that the stratigraphic record embraces a similar range of hiatuses, or gaps. The value of the SM, as developed here, is that it can be regarded as a meta‐stratigraphic concept that transcends the usual frameworks of stratigraphic classification and interpretation and thus allows a reappraisal of significant features of the record such as its perceived cycles and sequences. There are three outcomes of this meta‐stratigraphic analysis. First, the geometric units employed in sequence stratigraphy are components of a self‐similar series of essentially lenticular bodies, with no fundamental characteristics to distinguish them from larger and smaller bodies in the generally self‐similar series. Secondly, the interactions and feedbacks within the SM are so complex, and its outputs so repetitive in general character, that it is dangerous to assume that any perceived cyclicity provides an unambiguous chronicle of process cyclicity, whether eustatic, climatic or tectonic. The classic coarsening upward cyclicity of shallow marine facies can, for example, be viewed solely as the outcome of the lack of phasing between the chaotic behaviours tending to unload the SM and those creating accommodation space. Finally, the meta‐stratigraphic approach reinforces the view that, in the absence of demonstrable stratal continuity, chronostratigraphic correlation must continue to rely on sample‐based, mainly biostratigraphical, methods, rather than sequence stratigraphy.