Two of three reviews are favorable. The reviewers point out that the theoretical contribution, a method for pruning (merging) without requiring fine-tuning, is solid. The paper contains sufficient supporting empirical evidence for the efficacy of the method. The paper is therefore accepted.