Coming from the SOA world, I think a lot of it depends on the context (yeah yeah – that was a useful answer, score me 2/10).
The OASIS SOA Reference Model, itself an abstract model, discusses services as an action boundary between a capability and a consumer of that capability. When the service is interacted with (there is an abstract notion of an interaction model), this may
result in something behind the service issuing a command or call to action to do the functionality that completes the capability and has a "real world effect". Note that the concept of controlled opacity means that the consumer may not be aware of the exact
method or thing that fulfills the capability.
When a service is invoked, a command may be made to "do something". From a pure ontological sense, to me it is a specialized type of "request" since no command is guaranteed to complete. I think most modern software languages use this model with the
"try-catch" constructs. Issuing a command and assuming it just happens all the time is sort of a bad paradigm.
If I had to model it from the top, I would see it something like this
Communication -> oral statement -> request with conviction
Then again, it is Friday afternoon. Maybe time to put the ontology thoughts away for a bit.
COO and Director
Uberity Technology Corporation
"LiveCycle ES and Mobile Specialists"