Very interesting! I've been reading through Bob's links in order to get an idea of what those things are about. Especially the Valueflows seems close to resource theories in spirit, and I can imagine resource theory to play a part in such a framework, probably together with other components like some kind of powerful higher-order [process calculus](https://johncarlosbaez.wordpress.com/2018/05/12/rchain/). However, it's very easy to promise too much of my own toy when I don't understand others' toys very well! One problem is that the multi-agent generalization that I described [above](https://forum.azimuthproject.org/discussion/comment/17995/#Comment_17995) does not take into account any potential interaction of the agents; it merely describes the capabilities of the two agents separately. But nobody has really yet worked out the details of this, so there might be much more to say.