Replication is attractive for scaling databases up, as it does not require costly equipment and it enables fault tolerance. However, as the latency gap between local and remote accesses continues to widen, maintaining consistency between replicas remains a performance and complexity bottleneck. Optimistic replication (OR) addresses these problems. In OR, a database tentatively executes transactions against its local cache; databases reconcile a posteriori to agree on a common schedule of committed transactions. We present three OR protocols based on the deferred update scheme. The first two are representative of the state the art. The third is new; we describe it in detail. As all three protocols are expressed within a common formal framework, we are able to compare them, to identify similarities and differences, and to introduce common variants. We show that our protocol behaves better than the other two, with respect to latency, message cost and abort rate.