The germinal center reaction is the process by which low-affinity B cells evolve into potent, immunoglobulin-secreting plasma and memory B cells. Since the recycling hypothesis was created, experimental studies have both tracked movement of a small population of B cells from the light zone into the dark zone, supporting the recycling model, and parallel to the light zone-dark zone interface, indicating a one-way trajectory. We present a novel, sequence-based ab initio model of protein stability and protein interactions. Our model contains a dark zone region of clonal expansion and somatic hypermutation and a light zone site of antigenic selection. We show not only that a one-shot model is sufficient to achieve biologically-realistic rates of affinity growth, population dynamics, and silent:non-silent mutation ratios in the complementary determining region and framework region of antibodies, but also that a stochastic recycling program with or without realistic constraints on the structural stabilities of GC antibodies cannot produce biologically-observed affinity growth, population dynamics or silent:non-silent mutation profiles. The effect of recycling erases affinity gains made by potent antibodies cycling back from the light zone and causes B cells to pool in the dark zone under high replication rates.