(A quick analogy which I’m writing mainly so that the idea will leave me alone.)

Each human working in a team can be considered as both an abstraction layer and as computational resources.

The abstraction layer aspect of someone works in two ways: 1) it hides information (the logic that goes inside) and only works with inputs and outputs. The human is the interface to the rest of the team concerning the information hidden. 2) it assumes responsibility for outcomes. This is equivalent to a set of assertions that tell the system that certain outcomes will be fulfilled. This, of course, also allows for escalation of problems when the outcomes cannot be fulfilled.

The computation layer aspect of someone is what happens behind their own interface. It’s the sheer energy and precision they employ to reach results. Curiously, a lot of their internal work will often require interacting with other human interfaces.