This work-in-progress talk explores richer notions of failure expressible using the Decentralized Label Model for Availability and Integrity. We generalize failure tolerance to encompass mixed failures, survivor and failure-prone sets, and participants with different trust assumptions.