Yeah totally. I think to go with the biological analogies Kay makes, having them be async objs running code they message-pass to each other makes more sense to

Yeah totally. I think to go with the biological analogies Kay makes, having them be async objs running code they message-pass to each other makes more sense to me than like methods on classes. In the latter there is still a hierarchical ontology of objects imposed top-down while in the former emergent behavior occurs bottom-up. "Notes on the Synthesis of Form" (from Kay's reading list on the Squeak website) has some interesting stuff about trying to take graphs of nodes like this and figure out a good hierarchical decomposition given emergent discovery of edges. Here's something I've been thinking about lately: what if the object edges were organized spatially like a cellular automaton, maybe we can use spatial intuitions then, and if we only allowed linear-typed resources to be passed between them we actually have conserved "flow" of these resources, allowing the use of tools from economics to study the flow. Then it ends up looking like https://mitpress.mit.edu/books/growing-artificial-societies https://mitpress.mit.edu/books/growing-artificial-societies except with programmable monadic behavior at each cell. The linear-typed resources will allow linear proofs locally at each object but the monadic programming of each will allow non-linear emergence.