Enterprise Architecture Objectives and Claims
There are different views (and debates) on the objectives of Enterprise Architecture (EA); is it about describing and specifying (‘enterprise engineering approach’), or is it about understanding and insight (‘decision support approach’) or is about managing (‘governance approach’). Also different modes of how EA is being practised can be observed: foundation architecture, where the focus is on the business-IT alignment, or extended architecture and embedded architecture with a more integrated approach of Business and IT architecture design .
Nevertheless the different stances, given the fact that enterprises are complex social-technical systems, the claim that Enterprise Architecture is making is big:
“Enterprise architecture is the process of translating business vision and strategy into effective enterprise change by creating, communicating and improving the key requirements, principles and models that describe the enterprise’s future state and enable its evolution” 
“Enterprise architecture is the inherent design and management approach essential for organizational coherence leading to alignment, agility and assurance” 
Based on the needs of the decision maker and the stakeholders in a formal and abstract way a complete EA process consist of three steps:
- Model and make an abstraction of a real world organization; the As Is Model.
- Model and make an abstraction of a wanted real world organization; the To Be Model.
- Model the transformation function that maps the current As Is model into the wanted To Be model; the Transition Plan.
These abstractions or cognitive simplifications are needed because of the complexity, dimensions and the size of real world organizations. For any real world organization there are likely to be multiple models but the success of a particular model stands or falls to its ability to capture the behavior of the real world organization and real world transition in a useful way.
The problem-solving strategies that are mostly being used in this EA process are:
- Abstraction; solving the problem in a model of the system before applying it to the real organization.
- Divide and conquer: breaking down a large, complex problem into smaller, solvable problems.
- Analogy: using a solution that solves an analogous problem.
But does EA really solves problems or is EA using all kinds of heuristics (experience-based techniques) for structuring, learning, discovery and giving direction to solution finding?
The big question is, how to ensure that the recommendations and architectural designs created in the EA process are objective, complete, rational solid, can be justified and can be implemented successfully? Or to put it another way, how to ensure the ability to infer causal relationships between recommendations, designs, actions and outcomes? A proper Architecture Design Rationale, providing argumentation-based structure to the EA process and outcomes, is the least EA should offer but to infer causal relationships is much harder.
For the Enterprise Architect the universe of discourse generally refers to the collection of objects and subjects that makes a particular organization. You can take different perspectives on this depending on how the organization boundary is defined. According to Santos , there are four different perspectives on organizational boundaries.
- Efficiency boundary: organizations are institutions where transactions are taking place with the lowest transaction costs, based on comparing internal transaction costs (organization) with external transaction costs (market). In short the organization objective is transaction costs minimization.
- Influence boundary: organizations are institutions that attempt to reduce uncertainty and exercise power to improve performance. In short the organization objective is autonomy maximization.
- Capability boundary: organizations can be seen as bundles of resources that create competitive advantages and unique value. In short the organization objective is capability maximization.
- Identity and Cognitive boundary: organizations can be seen as social forms for sense making and in this case the boundaries should be set to optimize coherence between the identity, cognitive and emotional characteristics of the organization and its activities. In short the organization objective is identity and cognitive coherence maximization.
Based on the historical background of Enterprise Architecture (Information Technology) the focus (perspective) is usually on efficiency and capabilities and less on the social-political perspective, identity and influence. Taking this bounded technocratic perspective will give a biased view and will influence the set of possible solutions one can choose.
To address organizational issues and to circumvent bias, a more holistic approach is needed. This is accomplished by combining the different perspectives as stated above. And of course also with an open eye to macro-environmental factors i.e. taking Political, Economical, Social, Technological, Ecological, Legal, Ethical, and Demographic factors into account.
The Enterprise Architect has to make decisions about the recommendations and the architectural designs he wants to make. Rational decision making requires that the decision maker (in this case the Enterprise Architect) has all of the information necessary and is knowledgeable about the outcome of possible decisions. In reality, the decision maker is limited by the amount of time and resources available to obtain information to make the recommendations and the architectural designs, so the rationality of these artifacts is bounded. A consequence of this is that the decision maker (the Enterprise Architect) can not seek for the best solution, but instead has to seek for a solution that he is satisfied with and he is not willing to take any action for better solutions given the limited resources that are available .
Dominant logic and Cognitive Boundaries
In most cases managers don’t approach organizational issues or actions as if they are totally unique and require systematic study. Instead, this is handled through already existing knowledge systems or schemas. A schema represents the way in which managers conceptualize the organization based on: “beliefs, theories and propositions that have developed over time based on the manager’s personal experiences” .
The cognitive composition of the top management team of an organization firm can lead to a dominant general management logic. The dominant logic is the way in which the management team makes critical decisions. One side of the coin is that dominant logic greatly simplifies the decision process. The other side, however, is the risk of cognitive bias when the issues at stake are different from what they appear to be at first glance, and decisions can thus lead to significant errors.
The mental models that are being used can constitute cognitive boundaries for the Enterprise Architect. The acceptance of recommendations and architectural designs of the Enterprise Architect who is introducing and importing mental models from other settings can be low. The dominant logic of the Enterprise Architect peer group is also something to take into consideration.
Subjective Knowledge Building
How should the Enterprise Architect find and extract useful information out of the real world? Using the knowledge model of Boisot  in a EA context, see figure below, you could state that the knowledge of the Enterprise Architect is fundamentally subjective.
The Enterprise Architect as a knowledge agent is using filters to convert incoming data, a set of distinguishable states of the organization that are discernable, into information. Then the Enterprise Architect convert some of this information into meaningful representations of the observed organization that he can subsequently act upon, that is knowledge. The knowledge model has a feedback loop where the Enterprise Architect deploy expectations derived from his prior knowledge and experience to tune the filters to extract information from the received data. The Enterprise Architect also act directly upon the organization and by doing this he is also changing the source of the incoming data. The merging of new Enterprise Architect-filtered information with the Enterprise Architect personal path-dependent prior knowledge and values creates the essentially personal and subjective nature of the Enterprise Architect knowledge.
Which Knowledge Are We Talking About?
Knowledge comes in several flavors. Based on the definition of knowledge as justified true belief we can distinguish several ‘knowledge worlds’ :
- Possible worlds: whatever gives rise to beliefs that persons are willing to act on, providing that it doesn’t contradict the laws of logic or of physics.
- Plausible worlds: beliefs that strike a person as being true based on a personal and thus subjective justification (coherence and correspondence with the facts).
- Probable worlds: beliefs that have to be justified to others, that is objectively demonstrable coherence and correspondence with the facts.
- Actual worlds: true, justified belief where justification to others is done by showing direct evidence in the real world.
Question is if the Enterprise Architect is acting in a ‘Possible World’ or if he is travelling along one of the two different knowledge discovery paths from Possible Worlds to Actual Worlds? As stated by Boisot, “is the action based on the coherence of experience, does it make sense?” (path 1) – or “is the action based on the robustness of the experience’s replication?” (path 2).
Complex Adaptive Social Systems
Although there is a common view in the Enterprise Architecture world that complexity is a big problem, the general idea of most of the Enterprise Architecture methods is that by getting the right simplifications in the EA model we will understand the real world organization. In other words if we understand the basics of the different objects and subjects of the organization we can simply apply this knowledge to understand the whole organization. But in reality most times the whole is greater than the sum of its parts. Relations, the interconnectivity, between objects and subjects are such that through various feedbacks, objects and subjects variations no longer cancel one another out but become reinforcing and creates emergent properties and behavior .
Basically with complicated problems, it is possible to identify and model the relationship between the objects and subjects, and the relationships between the objects and subjects can be reduced to clear, predictable interactions. However, with a complex problem the system has a behavior that cannot be predicted via linear relationships. This is because of the number of interacting objects and subjects, the interdependent connections between those objects and subjects, and the degree of diversity among those objects and subjects (heterogeneity). The background of simple problems are low numbers of interacting objects and subjects, almost no interdependency between those objects and subjects, and almost no diversity among those objects and subjects (homogeneity).
What kind of problems can the Enterprise Architect solve with the current EA methods? Because of the number of objects and subjects that are involved, the high interdependency and diversity of the objects and subjects and the fact that they can adapt and learn makes an organization a Complex Adaptive Social System. Do the current EA methods really help in solving problems of Complex Adaptive Social System or are these methods restricted to solving simple problems and do they limit the Enterprise Architect’s ability?
The Wicked Organization
In a lot of cases organizations are struggling with problems that are wicked. A wicked problem  is a problem where stakeholders are involved with differing perspectives, incomplete, contradictory, and changing requirements and complex interdependencies. There is no simple, clear definition of the problem and there is no simple right or optimal solution. The ‘solution’ strongly depends on how the problem is framed and is a ‘one-shot operation’. On top of this we sometimes see strategic behavior where stakeholders interpret policies, legislation or contracts for their own gain and act accordingly or where stakeholders are selective about sharing information with other stakeholders because they think there is a gain in so doing. This social-political dimension of problem solving can hinder the Enterprise Architect in his activities.
Bounded Enterprise Architecture?
So the question was, how to ensure that the recommendations and architectural designs created in the EA process are objective, complete, rational solid, can be justified and can be implemented successfully? Let’s dissect this sentence:
- Objective: the Enterprise Architect is working in an environment where he comes across dominant management logic, strategic behavior and stakeholders with differing perspectives, and incomplete and contradictory requirements (wicked problem). Also is the gaining of knowledge by the Enterprise Architect personal path-dependent. We could say, objectivity is under pressure.
- Complete: because of historical reasons the focus in Enterprise Architectural work is mostly on a tangible/technical perspective (efficiency and capability) and there is less focus on the the social-political perspective (identity and influence). Maybe a strong claim but see for example the ongoing debate on Enterprise Architecture versus Enterprise IT Architecture.
- Rational solid: the Enterprise Architect is limited by the amount of time and resources available to obtain information (bounded rationality). This limited information will also influence the process of gaining actionable knowledge. Therefore the rational solidity is bounded.
- Justified: the organization as a Complex Adaptive Social System and the ‘Wicked Organization’ makes it difficult for the Enterprise Architect to justify his recommendations and designs. The organization as a nonlinear system has emergent properties and behavior that we neither well understand nor master and has wicked problems that makes it difficult to define a frame of reference. Therefore to infer causal relationships between recommendations, designs, actions and outcomes is very hard.
To make the journey from Possible Worlds to Actual Worlds we have to rethink our current Enterprise Architecture methods and tools.
The current view of EA methods is based on the Machine Metaphor, the organization as functioning machine. Organizational problems are solved by a mechanistic, centralistic, command and control approach based on an analytical (functional) decomposition of the organizational ‘tangible’ components. The implicit principle that is being used is that the organization is a static linear system. This only works well in the case of simple problems.
Another approach is needed to ‘solve’ complex problems, problems of Complex Adaptive Social Systems. Insight into the behaviour of these types of systems makes it clear that what happens in organizations may only be controllable to a limited extent. But nevertheless the focus should be more on the relations and the interactions between the heterogeneous objects and subject and the dynamics. This new approach should also give more space to the notion of bounded rationality, the gaining of knowledge and the social-political aspects of decision making that limits the objectivity and rationality of design, recommendations and decisions.
 J. Gøtze et all, ‘Coherency Management’
 F.M. Santos, K.M. Eisenhardt, ’Organizational Boundaries: and Theories of Organization’
 H. Simon, ‘Administrative Behavior’
 C.K. Prahalad, R.A. Bettis. ‘The Dominant Logic: A New Linkage between Diversity and Performance’
 M. Boisot, ‘Explorations in Information Space’
 J.J. Miller, S.E. Page, ‘Complex Adaptive Systems’
 H. Rittel, M. Webber, ‘Dilemmas in a General Theory of Planning’