The process of converting methanol to hydrocarbons on the aluminosilicate zeolite HZSM-5 was originally developed as a route from natural gas to synthetic gasoline. Using other microporous catalysts that are selective for light olefins, methanol-to-olefin (MTO) catalysis may soon become central to the conversion of natural gas to polyolefins. The mechanism of methanol conversion proved to be an intellectually challenging problem; 25 years of fundamental study produced at least 20 distinct mechanisms, but most did not account for either the primary products or a kinetic induction period. Recent experimental and theoretical work has firmly established that methanol and dimethyl ether react on cyclic organic species contained in the cages or channels of the inorganic host. These organic reaction centers act as scaffolds for the assembly of light olefins so as to avoid the high high-energy intermediates required by all "direct" mechanisms. The rate of formation of the initial reaction centers, and hence the duration of the kinetic induction period, can be governed by impurity species. Secondary reactions of primary olefin products strongly reflect the topology and acid strength of the microporous catalyst. Reaction centers form continuously through some secondary pathways, and they age into polycyclic aromatic hydrocarbons, eventually deactivating the catalyst. It proves useful to consider each cage (or channel) with its included organic and inorganic species as a supramolecule that can react to form various species. This view allows us to identify structure-activity and structure selectivity relationships and to modify the catalyst with degrees of freedom that are more reminiscent of homogeneous catalysis than heterogeneous catalysis.