@tetradian and @robert_phipps have recently floated the idea of Enterprise Architecture as Vectors.
Clearly the word "vector" is just a metaphor. But I agree that EA needs a more robust way of talking about change than simply assuming that change involves a jump from one state to another state. (Yesterday we were operating this process, tomorrow we shall switch to a new process.) Most organizational change requires an extended transition period, including the so-called learning curve. Capabilities develop and mature. Business relationships and trust take time to become established. And so on.
And EA needs to understand the complex interaction between many different changes. A typical enterprise is teeming with lots of different change initiatives, some of which may be formally constituted as "projects". Each one is based on a false or simplistic theory about the current state of the enterprise (AS-IS), and a naive or optimistic theory about the future state of the enterprise (TO-BE).
As the vector metaphor suggests, these projects and other change initiatives interact (compose themselves) in usually unpredicted and possibly unpredictable ways. Sometimes the initiatives merely cancel each other out, leaving the essential characteristics of the enterprise-as-system largely unaltered. (The ability of complex systems to preserve their identity and deep structure in the face of efforts to change them has been studied by many systems theorists, including Meadows, Beer and Maturana.)
However, the participants in these projects usually resist perceiving the full richness of what has happened: they try to convince themselves and their bosses that the intended outcomes have been more or less achieved, and that something else is now required. Thus the cycle begins again.
Since I posted this, Robert Phipps has posted a more detailed explanation of his idea: Part One. See also Linked-In discussion.