other crystallizing thought: ownership and trust should be more closely related to (or even direct functions of) identity and naming (which should be rooted in the same thing)

we say resource identifiers are typically "opaque", meaning you shouldn't extract any information from them. but in many cases they are actually not opaque strings: they directly encode multiple bits of information! http for example is more appropriately a tuple (scheme, host, resource) as i may have discussed before