The Natural Language Processing group at Microsoft Research is sharing the development-set results for the BioNLP Shared Task 2011 to help analysis of whether there is an upper bound on recall and precision in the trigger-detection process. It is interesting to consider whether there are annotations in the development set that cannot be predicted by participating systems.