When Apollo returned the first moonrocks, a major surprise was that the lunar highlands are built of a single mineral - feldspar. Feldspar crystals floating to the top of a moon-magma ocean can explain this composition. If accepted as a key early stage in lunar evolution, it is a logical leap to infer that Earth must have had a similar – or larger – magma ocean during its early evolution. (The gravitational impact energy per unit mass that is released during a planetesimal’s accretion scales as GM/R.) The nagging problem with the inference that Earth passed through an early magma-ocean stage is that the oldest rocks on Earth show no direct signs of a magma ocean. Instead the petrology of the oldest preserved Earth rocks shows clear evidence that repeated events of small to medium degrees of partial melting and melt extraction, as opposed to pervasive fractional crystallization, has been the modus operandi of terrestrial differentiation. The big difficulty is how to effectively ‘remix’ the products of an early terrestrial magma ocean back into the quasi-uniform ‘primordial’ pyrolite/peridotite silicate lithology from which oceanic and continental crust are thought to have evolved by partial melting events. Here I propose that a partially molten silicate body is actually highly resistant to the formation of a magma ocean. Jing and Karato (2012)’s experiments imply that a silicate melt should absorb much more impact shock-energy than either a silicate solid or an iron solid/melt. In this case, impact energy will be heterogeneously added into the growing proto-Earth, with silicate partial melts being shock-compression-heated to their vaporization temperature before their surrounding silicate solids heat to their melting point. The growing partially molten planetary surface will tend to ‘explode’ during impact events, with each impact-induced-explosion using a relatively small mass of vaporized silicate partial melt to fragment and rework much larger masses of cold, shock-fractured overlying ‘lithosphere’. This explosive-armour-like mode for silicate planetary accretion will strongly resist the magma ocean-mode of planetary differentiation. A magma ocean would only tend to form in the planetary body created from the accreting debris ring of a giant impact event, a Moon.