Magnetars produce beautiful fireworks in x-rays and gamma-rays: small, large, and stupendous flares, fast quasi-periodic oscillations (QPOs), as well as transient increases of their luminosities on timescales of months to years. How the release of magnetic energy leads to this rich variability is not precisely understood, but the processes responsible for it may well be hidden under the surfaces of the stars, in their solid crusts and super-fluid cores. The pulsars' magnetic dynamics is less dramatic, yet the evolution of the pulsars' magnetic fields with age is an important unsolved theoretical and observational problem. In this talk I will discuss a set of theoretical ideas for how the magnetic dynamics is likely to work. 1. I will demonstrate a new type of failure that is likely to occur in magnetars' crusts , a thermo-plastic wave. These do not have direct terrestrial analogues, and behave qualitatively like deflagration fronts. Coupling to Hall waves leads to avalanches of thermoplastic waves, and I will argue that these avalanches are responsible for magnetar outbursts. 2. I will show that the coupling between elastic waves in the crust and the Alfven waves in the core leads to rich dynamics and phenomenology of magnetar oscillations, but leave un-answered questions for how these relate to the observed QPOs. 3. I will show a series of numerical experiments on the magnetic field evolution in pulsars that model the coupling between the crust and the core and, with significantly more uncertainty, between the superfluid vortices and superconducting flux tubes in the core.