My problem with events is the
definition of boundaries, the identification of the "pattern" of an
event for the purpose of recognition. Although I see events as a sequence of
temporaly identifieable entities and not as a chain of cause of effect, because
there are many concurrent causes leading to an event (see Petri nets), I still
can see how an event remains very vague and arbitrarily identified as changes
take place at all sorts of scale and places before they add up to a situation
that is no longer identical with the one a second ago. So this approach does
not solve the problem of showing the trajectory of learning about things,
recording it a manner that serves as a good (and standardized) interface
beween FO components.
RC> Agreed. Tiny changes make big
differences in chaos theories (butterfly wings “cause” hurricanes).
If we start with just the two items – existence detectors and identifiers
we still have to define identity. The only place to go with that is FCA, but
that skips over the question of how you identify the objects and properties
BEFORE you start clustering them. There are missing pieces in the math here,
or at least the realities are not fully modeled yet.