Midplate volcanism, volcanic chains, diffuse boundaries and variable chemistry basalts are usually considered to be outside the plate tectonic hypothesis and to need separate explanations. This is true only for the instantaneous, steady state, kinematic and hard plate versions of the hypothesis. In the more general plate paradigm (with fewer restrictive adjectives), melting “anomalies”, seamount chains, and LIPs are by-products of plate tectonics. This assumes that the shallow mantle is close to the (variable) melting point and that athermal and episodic processes are important. Cooling of the surface generates forces that drive, break and reorganize plates; global reorganizations (including new plate boundaries) are intrinsic; regions of intense, long-lived magmatism and shallow tensile stress are (usually but not always) plate boundaries. Plates are regions of lateral compression. Plate boundaries have shallow extensional or strike-slip earthquakes; where the mantle is near the melting point the buoyancy of magma generates dikes and volcanoes. When compressional forces dominate, upwelling magmas pond beneath the plate until released by extensional stresses. Large melting anomalies are episodic and associated with changes in plate stress and new plate boundaries (often triple-junctions). Incipient boundaries can be extensional and volcanic, as can abandoned ones. Ridges, island arcs, seamount fields and chains, and reactivated and incipient boundaries, are part of a single process. The plate paradigm thereby reverses the assumptions of current geodynamic and geochemical reservoir models: Locations of volcanoes are controlled by lithospheric stress and fabric (not mantle temperature).The volumes of magma are controlled by lithospheric extension and shallow mantle fertility (not by conditions at the core mantle boundary). The stress-, fertility- and thermal-states are controlled by plate tectonics and upper mantle recycling (not by infusions from the deep mantle). The upper mantle has variable chemistry, melting point, potential temperature and is perisolidus everywhere (not just under hotspots). The upper mantle is not dry, refractory, isothermal, uniform, or well-stirred. The ends of volcanic chains are not “fixed”. One even reverses the motivating question, why are there not volcanoes everywhere (instead of why are there volcanoes some places)? Non-fixity, aberrant ages and directions, absence of uplift and thermal rejuvenation, low heat flow, “normal” magma temperatures, distributed volcanism, and seamount chemistry are not “anomalies” as they are in hotspot theory. The associations of LIPs with suture belts and cratons are no longer coincidences. Predictions: Midplate volcanism implies extensional stress, not necessarily high heat flow, uplift or thermally thinned plate. It need not be age-progressive, parallel or in chains. Basalts are variable in chemistry (central limit theorem) and involve the same recycled components. Recycled oceanic crust and gabbro cumulates are variably distributed in the shallow mantle. The scale of chemical heterogeneity corresponds to the scale of recycled components (crust, sediments, lithosphere) and arc segments. Volcanic chains are stress gages, not motion detectors. If the shallow mantle is perisolidus (near the melting point) then even the deepest slowest ridges will be magmatic; melting will occur under thick plates (not likely in thermal theories). Ponding and underplating should be common. In contrast to thermal theories, volcanism can initiate and terminate abruptly. The technique of testing a hypothesis by assuming the opposite, is reductio ad absurdum. A simpler more powerful theory emerges when we drop the adjectives and reverse the assumptions.