According to that post the permanence of change drives how we run our organizations,but it is against our human quest for stability. As far as Enterprise Architecture is concerned, he notes that the To-Be Architecture organizations trying to move to from current As-Is Architecture is a moving target: There will never be a stable Enterprise Architecture.
I do agree that Architecture is dynamic in nature, however we should look more deeply at the characteristics of that ever changing process.
Does Enterprise Architecture evolve linearly or Spiraly?
I use the term linear for describing any type of monotonic evolution, just because linear is simpler than other monotonic fuctions.
Yesterday, I encountered a SaaS example supporting my case.
I looked at an old Giga Information Group article dated 2002. The article written by Byron Miller title is: "ERP Software as a Service".
The issues and observations are similar to current ERP SaaS issues (described in many articles and Reaserch Notes including my post: Future Applications SaaS or Traditional).
The term SaaS in the old article does not refer to Cloud Computing but to Application Service Provider(ASP)model.
Is the As-Is to the To-Be Architecture approach a wrong approach?
I do think that it is a usefull approach. The fact that even if we complete the transformation from As-is to To-Be we will need a new To-Be, does not deny the value of reaching a better architectural state than the current state.
Prepetual change is against Human nature, but reaching a goal is not. It is easier for us to reach a goal(To-Be Architecture) and look afterwards for another goal (next To-Be Architecture), than to act in a chaotic ever changing environment without any sub-goals.
Why Architecture is doomed for change?
It is not only because of the Dynamic Business, The Technological changes and other organizational changes.
Another main reason for Enterprise Architecture inherent dynamics is its nature. EA is an abstract model describing artificats (Business artifacts, Technological IT artificats and Applicative IT artifacts) and the relations between them.
most abstract models does not fully correspond to the real entities they describe, so even if nothing is changed the model should be improved and changed.