Ideology sets the menu. Interests help pick the meal. Costs distribute downward—and the meal fights back.
Three active lenses
Doctrine
Doctrine explains which actions appear legitimate, desirable, or imaginable. It establishes the policy menu, but does not by itself determine which option survives contact with competing interests and the actual decision-maker.
Interests and incidence
This lens asks two inseparable questions: who benefits, and who pays? Material interests influence selection, exemptions, timing, and implementation. Costs may be delayed, hidden, or shifted onto groups with less influence.
Personalism and decision dynamics
Planning is not deciding. Advisers compete to construct the choices a leader will see; loyalty, presentation, timing, and the last conversation can outweigh formal process at the moment of commitment.
The operating system
Patrimonialism describes how the lenses are processed: personal loyalty over impersonal procedure, institutions treated as extensions of the leader, and law approached as something to route around—or rewrite—when it constrains discretion.
The external-reality layer
Targets are not passive objects. Adversaries adapt, allies bargain, markets reprice risk, bureaucracies resist, and middle powers create off-ramps. The framework must account for agency after the decision, not merely the intentions that produced it.
Operational affordability
A policy does not require enormous strategic value if the immediate action appears cheap, dramatic, and externally payable. Target isolation, visible domestic cost, retaliatory capacity, and the availability of limited tools can determine which ideas become actionable.
Application workflow
- Pass the applicability gate. Use only the lenses that add explanatory tension; do not fill slots for symmetry.
- Declare the phase. Planning, decision, and consequence phases weight the lenses differently.
- Discount noise. Separate speeches and anonymous trial balloons from commitments of money, personnel, authority, or reputation.
- Map actors and incidence. Ask who bears each scenario’s cost before asking who wants it.
- Preregister scenarios. Record probabilities, resolution criteria, and falsifiers before the outcome is known.
- Keep score. Preserve misses and revise the model rather than attaching framework language after the fact.