Local-first software must keep working when the network is down, so it has to be coordination-free by design. In this talk, we explore with the audience what “coordination-free” means, and show that it is fundamentally a property of problem domains, not programs. We then state the key result that coordination-free computation is exactly computation expressible in monotonic logic. From there, we walk through practical techniques for transforming a problem’s domain so that monotonic formulations become possible, and show how this perspective generalizes into Monotone-View Types (MVTs): a co-transactional model where state is observed through views that are revised as information arrives, while preserving invariants by construction. Finally, we ground the ideas in familiar systems (Bitcoin, Ethereum, and a tree-with-move CRDT), and argue that MVTs, like CRDTs, can be engineered with Byzantine fault tolerance, making them a promising foundation for authority decentralization while supporting substantially richer behavior than classic CRDT designs.