To resolve infrared problems with the de Sitter invariant vacuum, de S
itter symmetry must be broken. We discuss how the history of the de Si
tter phase is related to the infrared cutoff. We illustrate how either
(1) the diagonalization of the Hamiltonian for long-wavelength modes
or (2) an explicit modification of the metric in the distant past lead
s to natural infrared cutoffs. The former case resembles a bosonic sup
erconductor in which graviton-pairing occurs between non-adiabatic mod
es. While the dynamical equations respect de Sitter symmetry, the vacu
um is not de Sitter invariant because of the introduction of an initia
l condition at a finite time. In the latter case, we indicate just how
infrared cutoffs are associated with modifications of the metric. The
implications for the one-loop stress tenser and the production of par
ticles are also discussed.