Depending on the org structure, pair programming can sometimes be a clearly superior version of a code review policy. Particularly when the review latency is high and, as a consequence, developers work on several branches in parallel.