The exponentiation of a small seed of energetic electrons into a strong current of relativistic electrons during disruptions is a major threat to the ITER mission; the adequacy of the plan for the protection of ITER remains far from clear. The hope that issues involving relativistic electrons can be resolved in the non-nuclear phase of ITER operations has a fundamental flaw. In the nuclear phase two mechanisms exist for producing seed electrons: remnant and steady production. In the non-nuclear phase only the more easily avoidable remnant source exists. Whether the plasma current in ITER must be severely constrained to avoid unacceptable machine damage rests on issues discussed in this paper. An adequate understanding of fast magnetic reconnection is one of these issues. The loss of poloidal field energy during a disruption occurs through two distinct processes: (1) a quasi-ideal fast magnetic reconnection, which conserves magnetic helicity and accounts for large and sudden drops in the internal inductance, and (2) the resistive dissipation, which dissipates the helicity and is required to quench the current. To correctly separate these two processes requires spatial and temporal resolutions far beyond those of existing simulations. The assumption that stochastic magnetic field lines have a diffusive radial motion allows simple simulations of existing experiments and predictions for ITER that go far beyond existing analyses.