Here I worry about whether an object should be extractable while in this note I speculate how to extract for willing objects. Yet again I feel forced to characterize those objects that we might extract.

A page key which I hold uniquely is an extractable object. If I know all this, it is easy to extract and move the page to another similar system. There may be a useful concept of a pure state bearing object. It stays put (has constant state) between invocations. This probably means that it has no internal running domains and that no one outside the object, besides me, has a non-sensory key to it or a part of it.

There may be no secret about the state, even though there are secrets about the representation of the state. An object is typically unable to disgorge its behavior accurately. Aside from the issue of extraction, the idea has been suggested several times that some objects manifest their behavior. Some shrink would be sent a key to the object and if the builder of the object were willing, the shrink could attest to the behavior. The most severe practical problem is the format of the response. Machine code is awkward. Higher levels are practical only if the language the object is written in is familiar to the requestor of the behavior. A strength of Keykos is as a place for programs in various languages to interact.

Already I stumble on the question of manifest extractability or mere extractability.