Your definition is of course an equally valid one that might be more appropriate in certain contexts - it just doesn't fit my mental model of computation equally well.
Likewise I see FP as a reactionary mechanism to write fewer and simpler protocols, because when the data is immutable the problem can usually be greatly simplified.
I believe CLOS would beg to differ.
Multiple dispatch is one of the things I really miss about Common Lisp. I do like OOP, but often the most intuitive way to think about something is multiple dispatch, and trying to shove it into a single dispatch model can produce some really ugly code. I'm pretty sure that is the main reason there are so many anti-OOP zealots in the world.
Also, I didn't realize Dylan had multiple dispatch. I should take a look at it sometime.
The cake pattern has as much to do with encapsulation as it does crosscut composition.
OOP is about subtype polymorphism [1] by means of runtime single-dispatch [2] and modularity, as in abstract modules or components that help with decoupling, a notion supported explicitly by SML, or exemplified beautifully in the Cake-pattern used in Scala.
[1] https://en.wikipedia.org/wiki/Subtyping
[2] https://en.wikipedia.org/wiki/Single_dispatch