General Modeling Considerations
Time and Data
Is it easier to model recent historical episodes, or ancient ones? For historical events that are far in the past, simpler technologies reduce the complexity of the situation. Moreover, we have have less of a vested interest in the interpretation of these events, making the essential power struggles more obvious. On the other hand, the farther back one goes in time, the less quantitative data is available to populate our power structure models, leaving us with a lot of guesswork as to what actually happened.
Historical situations can be characterized in various ways, making it debatable how particular individuals and coalitions should be grouped into abstract agents. This raises the question of the level of granularity at which the model should be applied. For example, in the Persian War (c. 500 BC), one might simplify the situation to be a case of two agents: Persia and Greece. Alternately, one might decompose Persia into its provinces, some of which rebelled during the conflict, and model the Greeks as being composed of a Spartan coalition and a Athenian one. One could go further and model each Greek city state as an agent. With more detail, one gets more nuanced and interesting political behavior, but one also begins to run into computational limits due to the exponential growth of the game tree.
Since models simplify reality, there are numerous ways to represent a real world situation within the model. A real world power structure, such as the United Nations, is composed of many interacting states and agencies. Representing each of these in an abstract power structure would be overly complex and burdensome. Decisions have to be made about which entities should be included and excluded from the model, and about how to group those entities into agglomerations. It is rarely clear cut how to do this, and the results of these choices may cause the outputs of the model to vary widely. This difficulty seems inherent to the endeavor, and can be managed only by being explicit about one’s choices and assumptions.
Perception versus Reality
Since states do not necessarily have objective knowledge of the power structure in which they exist, to explore their actions we must sometimes work from models based on how they perceived the situation. For example, in World War II, German military leaders greatly underestimated the strength of the Russian military, which they quickly realized soon after invading.
Streams versus Transactions
Another way that quantitative realist models are approximations is that flows of power in the model are streams, whereas in the real world power flows are composed of numerous discrete transactions.