Essence of Decision

Explaining the Cuban Missile Crisis

by Graham T. Allison, Philip Zelikow

Cover of Essence of Decision
complete 440 pages
History
Added: June 18, 2025 Updated: June 18, 2025

Reading Questions

Do the authors succeed in questioning the rational actor, Model I, paradigm?

As the authors stated, their goal was to question the RAM, but to supplement it. When information is limited, the RAM can still provide some reasoning or prediction for future actions, but it is not complete on its own. Unfortunately, models II and III require significantly more intelligence and time, so they are not always applicable until after a conflict or decision.

Do Models II and III explain more than Model I?

They do, while building on Model I, but both have their limitations. Model II explains how organizations influence decision-making, and Model III illustrates how competing personalities impact decisions.

What makes an action “rational”?

An action is rational if it is utility-maximizing (may be perceived) and aligns with the actor’s goals or preferences.

Is there a single “rationality” by which all actions/actors can be judged?

No, what is rational is defined by the actor making the decision, so rationality is independent of every person and can further change based on the situation and available information.

What are the interrelationships between the three models? Are they mutually exclusive?

All three models build on each other (loose coupling) and are not mutually exclusive. While it is always preferable to use the highest model possible, frequently analysts can’t use more than the RAM due to time and intel.

What is the utility of the “Models I-II-III Questions” summarized by the authors on pages 389-390?

The model questions provide a quick reference for analysts/scholars to apply the models to a scenario. They are not comprehensive approaches, but sufficient to quickly identify which model can be used in a situation with a given dataset.

Does the Cuban Missile Crisis offer any strategic lessons?

  • The impacts of public communication in deal-making.
  • The impact of vague orders during points of crisis (U-2 shoot down, camouflage, nuclear fighters, etc)
  • The criticality of time for an opponent to see, think, and blink.
  • Organizations and players constrain options.

WWCLD (What Would Curtis LeMay Do)?

  • Kill ‘em all.
  • He would have been selling the advantage of air power and potentially shaped the conversation, influencing Kennedy to use airstrikes instead of a blockade.

Online Description

This classic book has been substantially rewritten to reflect the significant new American and Soviet archival sources now available to researchers. Using the central case of the Cuban Missile Crisis as a basic frame of reference, The Essence of Decision teaches readers how to compare and contrast perspectives on foreign affairs.

🔫 Author Background

🔍 Author’s Main Issue / Thesis

  • The Authors major argument is to understand the Cubin Missile Crisis or any other major geopolitical event, analysis must move beyond the Rational Actor Model (RAM) (Model I), and analyze decision with the Organizational Behavior Model (Model II) and the Governmental Politics Model (Model III).

📒 Sections

Introduction

  • Book is written for Colleagues, Students, General Audience and spouses (so everyone)
  • “Although the Rational Actor Model has proved useful for many purposes, there is powerful evidence that it must be supplemented by frames of reference that focus on the governmental machine—the organizations and political actors involved in the policy process.”

Key Points

  • *Chapter 1: Model I: The Rational Actor

    • RAM can be helpful because it is still provides significant explanatory power.
    • The RAM is based on the Economic Man, or the idea that states or individuals are self-interested, rational decision-makers who will make decisions based on known or perceived outcomes.
    • “Rationality Theorem”: There exists no pattern of activity for which an imaginative analyst cannot write a large number of objective functions such that the pattern of activity maximizes each function. Put another way, with a simple and small data set like that used in the RAM, an analyst can make the data match any outcome they want.
    • Classical Realism: 1) Unitary states are the key actors in international affairs 2) States act rationally 3) The world is a Jungle (Hobbies) 4) States pursue security and power
    • Neorealism (Structural Realism): More “scientific”, and stresses system-level variables (international balance of power). Neorealists believe states act poorly when put in poor positions, not because the state in inherently bad.
    • International Institutionalism: Steps beyond neorealism and argues that institutions can enable effective allegiances / reduce conflict, by making the transaction cost of cooperation lower, and providing additional information to states.
    • Liberalism: The structure of the state matters because it affects the values and views of citizens and the state.
      • Conditions for peace in Liberalism (Kant): 1) Must be a republic 2) Market Economies aimed at improving citizens’ well-being
      • Traditions: 1) Social Value of Leaders and respect for rights 2) Commercial cooperation 3) Republics.
      • “The central proposition: “Societal ideas, interests, and institutions influence state behavior by shaping state preferences, that is, the fundamental social purposes underlying the strategic calculations of governments.""
    • Schelling’s Strategy of Conflict: “strategy analyzes and explains the maze of national actions and reactions as more or less advantageous moves in a game of interdependent conflict.”
      • Limited war is more probable when surprise is harder (global ISR). Limited war requires limits (Norms), explicit statements and actions of nations constitute strategic signals.
    • Sir Geoffrey Vickers key questions: How does the actor “select, derive, and represent its information about the ‘state of the system’? How does it derive the standards by which this information is evaluated? How does it select and initiate a response? “We can think of these three sets of judgments, value, reality, and instrumental—coming together in a triangle of mingled beliefs.""

  • Chapter 2: The Cuban Missile Crisis: A First Cut

  • “Hypothesis 1: Cuban Defense: An analyst who knew nothing about the Soviet Union except that it was a powerful country and that one of its important allies, Cuba, feared attack by a large, threatening neighbor, might infer that the powerful country would come to the aid of its weak friend.” In Model I / RAM, the countries almost don’t matter; the USSR has a weak ally that needs support, and for minimal risk, it can help further its own interests.

    • Counters:
      • No evidence that Khrushchev was just there to support Cuba
      • If the USSR wanted to deter an invasion, ground troops would have been a better choice
      • Tactical nukes would have deterred an invasion more than strategic nukes, and if the USSR needed to range all of the US, MRBMs would have made more sense than ICBMs.

  • Hypothesis 2: Cold war politics: The missile deployment was a strong probe of American intentions, aimed at strengthening the communist world while dealing a blow to America.

    • Counters:
      • Why probe after Berlin?
      • Size and scale were larger than just a probe.
      • Too low a chance of an afait accompli
      • Why 1962?
      • Why Cuban, one of the most disadvantaged locations for the USSR?

  • Hypothesis 3: Missile Power: The USSR didn’t have confidence in their ICBMs and felt overmatched by the US. Moving missiles to Cuba allowed the USSR to utilize shorter-range nukes to attempt to reach nuclear parity with the United States.

    • Counters:
      • No real military advantage to missiles in Cuba
      • Extreme Risk (unless afait accompli)
      • SAM was directed to deploy before the missile decision was made.

  • Hypothesis 4: Berlin — Win, Trade, or Trap: The USSR was attempting to force the US into making concessions regarding Berlin. If the US wanted the USSR to leave Cuba, then the US would have to leave Berlin. Khrushchev had few other options to gain a positive outcome in Berlin, so they forced the issue in Cuba.

  • Other observations about all 4 hypotheses: There was a grand plan, of IADs then missiles, but the decisions were separated. The goal of the USSR was surprise, but they used almost no camouflage, and the USSR deployed missiles even though Kennedy was very clear that Cuba was off the table.

  • US Options

    • Do Nothing, unacceptable to Kennedy
    • Diplomatic Pressures, Kennedy thought it would be a weakness because he had said publicly multiple times that Cuba was off the table for offensive weapons.
    • Secretly Approach Castro, could offer to Castro to split from the USSR or be invaded, but the US didn’t think Castro would split.
    • Invasion could also remove Castro, but the expense was significant and would weaken the US presence in other parts of the world.
    • Air Strike was Kennedy’s preferred option, but it had to be small and focused. Also worried about killing Russians and a surprise attack, looking like Pearl Harbor.
    • Blockade, a Hostile act, but limited response actions in Berlin. Significant issues were whether the Navy could form what, and whether Kennedy would authorize firing on USSR-flagged ships. Allowed the most greatest flexibility for the US, allowing a continued threat of invasion/strikes if Russia didn’t comply.

  • Left Khruschev with three options: keep boats out of the area, submit to being stopped and searched, or provoke the US further.

    • The first private offer was to remove the missiles for a pledge not to invade Cuba.
    • The second offer was public and identical to the first, except for the removal of missiles from Turkey.
    • Ultimately, Khruschev accepted a public pledge not to invade Cuba, and a private commitment to work on the missiles in Turkey for the removal of Russian missiles from Cuba.

  • Chapter 3: Model II: Organizational Behavior

    • The Organizational Behavior Model (Model II) examines states’ decisions as outcomes of their bureaucratic structures. This can be seen as the organization following the same processes they have always followed, regardless of whether they are still rational or efficient.
    • This model is exaggerated in governments because they aren’t profit-driven, have limited control over organizational structure and goals, and are bound by both internal and external rules.
    • Model II also views managers and sub-organizations as critical roles, because they must distill directions from the government into actions, and will frequently shape their interpretation of guidance to meet their organization’s culture and structure.
    • When looking at an organization, an analyst can ask four questions:
        1. Where do organization derive their preferences?
        1. Why does organizational behavior contradict “rationality”?
        1. Why are organizational structures sometimes so peculiar?
        1. How do organizations relate to their environment?
    • Paradigm of efficiency: Organizations are generally more locally oriented and are focused on international optimizations.
    • Paradigm of culture: Organizations typically examine how they and their employees fit into the broader world (field/sector).
    • “What is now known about the behavior of organizations is enough to suggest significant limits and essential supplements to Model I for explaining and predicting governmental behavior.”
      • I. Basic Unit of Analysis: Government actions are organizational outcomes, organizational capacities, and structure-bound government COA generation. The only exception is that innovators inside the bureaucracy can stretch organizational capacity and structure.
      • II. Organizing Concepts:
        • A) The state is not a unitary actor, but a group of loosely allied government organizations that the executive has to lead.
        • B) Fractionated Power, because no one person or org can solve problems in a large government, all issues are required to be split between multiple organizations.
        • C) Organizational Mission, formal or informal, drives what an organization does (constrains).
        • D) Ops Objs, Special Capacities, and Culture all narrow the set of problems an organization will examine based on its beliefs and culture.
        • E) Actions as Org Output, the structure and culture of an org define the orgs outputs, characterized by the following:
            1. Objectives
            1. Sequential Attention to Objs
            1. SOPs
            1. Programs & Repertoires
            1. Uncertainty Avoidance
            1. Problem-directed Search (orgs will search for problems within their model and stop when it makes sense based on their culture.)
            1. Organizational Learning and Change (orgs are incremental, so all change will be minor and attempt to maintain norms.)
          • Things that can change organizational output: Prolonged Budgetary Famine, Budgetary Feast, and Dramatic Performance Failures.
        • F) Central Coordination/Control
        • G) Decisions of Governmental Leaders
      • III. Dominant Inference Pattern: Actions will only marginally change over time, constrained by organizational formation.
      • IV. General Propositions:
        • A) Existing Organized Capabilities Influence Government Choice.
        • B) Organizational Priorities Shape Organizational Implementations.
        • C) Implementation Reflects Previously Established Routines.
        • D) Leaders Neglect Calculations of Administrative Feasibility at Their Peril.
        • E) Limited Flexibility and Incremental Change
        • F) Long-range Planning.
        • G) Imperialism. (Orgs must grow or die)
        • H) Directed Change. (Strong leaders can direct change)
      • V. Specific Propositions
        • A) Deterrence
        • B) Force Poster
      • VI. Evidence: This paradigm’s stark statement of organizational tendencies constitutes a marked shift of perspective. Examination of government action in terms of these roughly formulated concepts and propositions can be fruitful.

  • Chapter 4: The Cuban Missile Crisis: A Second Cut

    • Russian organizational actions didn’t change just because the troops were deployed to Cuba.
      • U-2’s first overflight wasn’t interfered with. No one passed down for the troops to engage outside of an invasion.
      • Camouflage didn’t happen to the missile sites because they were never camouflaged in Russia.
      • Construction Pace was a regular business hour.
      • Missile Sites, which were all designed identically to sites in Russia.
    • The US only detected the missiles because of the capacities the country had created over the decades.
    • Months before the missiles were detected, the DOD created three OPLANs: one for air strikes alone, and two for the invasion of the island. When the missiles were detected, the organization had already imposed constraints on how to address the problem.
    • For the blockade, President Kennedy wanted to give the Russians time to see, think, and blink. To achieve this, he had to maintain positive control of the Navy, an organization that culturally favored engagement with the enemy as quickly as possible.
    • No Cities Doctrine was the idea that the US would only use nukes against military targets, hoping the Russians would do the same. However, at the same time, the US plan was to disperse nuclear bombers to civilian airports/cities.
    • The decision to broadcast the second Russian offer over the radio wasn’t a deliberate act by Russia, but rather the action of an organization seeking to convey the message to America as quickly as possible; they had not considered the additional political implications it would create for America.
    • When a U-2 accidentally flew into Russian airspace, the SOP was to scramble fighters to support. No one in government thought about the messaging of sending nuclear-armed fighters to support a U-2 that had flown into Russia. The organization just followed its processes.
    • When Kennedy instructed the State Department to discuss with Turkey the potential removal of the Jupiter missiles, the State Department fell back on the organization’s goals. Instead, it began developing a centralized missile plan in collaboration with other partners. It took direct action from the POTUS to get the state to overcome its organizational objectives and do what the president wanted.

  • Chapter 5: Model III: Governmental Politics

    • “Government behavior can thus be understood according to a third conceptual model, not as organizational outputs but as results of bargaining games. Outcomes are formed, and deformed, by the interaction of competing preferences.”
    • “In contrast with Model I, the Governmental Politics Model sees no unitary actor but rather many actors as players: players who focus not on a single strategic issue but on many diverse intranational problems as well; players who act in terms of no consistent set of strategic objectives but rather according to various conceptions of national, organizational, and personal goals; players who make government decisions not by a single, rational choice but by the pulling and hauling that is politics.”
      1. Separate Institutions Sharing Power (The constitution didn’t create separate powers, but shared powers)
      1. The Power to Persuade (Most presidential power is to persuade other officials of the government)
      1. Bargaining According to Processes (Bargains are struck within the framework of the governmental system, players are playing it)
      1. Power Equals Impact on Outcomes
      1. Intranational and International Relations (The model extends to relationships/politics inside and outside the government.)
    • Foreign participation (allies) affects how members play the game internally and externally to the country.
    • Group decision making (model III) leads to several outcomes:
      • Better Decisions
      • The “Agency Problem”
      • Participants: Who Plays?
      • Decision Rules
      • Framing Issues and Setting Agendas
      • Group Think
      • Complexity of Joint Action
    • I. Basic Unit of Analysis: Governmental Action as Political Resultant
    • II. Organizing Concepts:
      • Who plays?
      • What factors shape players’ perceptions, preferences, and stance on the issue?
      • What determines each player’s impact on results?
      • How does the game combine players’ stands, influence, and moves to yield governmental decisions and actions?
    • III. Dominant Inference Pattern: “Model III’s explanatory power is achieved by displaying the game-the action-channel, the positions, the players, their preferences, and the pulling and hauling that yielded, as a resultant, the action in question.”
    • IV. General Propositions:
      • A) Political Resultants
      • B) Action and Intention. Government actions do not presuppose government intention, but are generally aligned with organizational preference.
      • C) Problem & Solution. Players focus on the game (politics), so they are limited in their solution space and scope.
      • D) Where you stand depends on where you sit.
      • E) Chiefs and Indians.
      • F) 51/49 Principle. In any interaction, the player should attempt to give 51% to gain/maintain relationships.
      • G) International and Intranational Relations. It only matters to the degree that it gives an advantage or disadvantage to other players (internal or external).
      • H) The Face of the Issue Differs from Seat to Seat
      • I) Misexpectation. The pace of multiple games limits focus on any one game.
      • J) Miscommunication. Pace plus Noise makes communications hard
      • K) Reticence. Advantages of not showing your cards.
      • L) Style of Play. Elected officials and appointments will differ from civilians in terms of time horizons.
    • V. Specific Propositions
      • Use of Force in Crises. The more senior members are on board, the more likely it is that force will be used. Additionally, it is more likely if framed as an incremental action.
      • Military Action (short of nukes). It will be delayed in getting opponents on board, and it will be a group decision involving civilians and the military.
    • VI. Evidence: Model III can explain actions of government that don’t align with Model I or II.

  • Chapter 6: The Cuban Missile Crisis: A Third Cut

    • The Russian decision team was smaller, primarily focused on domestic governance, and was Russia’s normal government formation.
    • The timing shaped how American players viewed Cuba, especially for Kennedy after the Bay of Pigs.
    • The discovery of the missiles was shaped by politics, with McCone having to pull the organization to U-2 overflights of Cuba.
    • Kennedy felt he had to take forceful action both to show American strength to the Russians and to maintain power for his party internally.
    • JCSC wanted to invade; SecDef saw the situation as a political problem, while Rusk initially favored diplomacy. With other players supporting different ideas, the members’ seating arrangements and goals shaped the options they supported for the POTUS.
    • Unlike Kennedy’s hand-picked group, Khrushchev relied on his standard policy process. While little is known about the inner workings of this policy group, Khrushchev was closer to being a sole decision-maker.
    • Before Russia’s second public deal, Kennedy didn’t consult anyone outside of the XCOM (his small team); afterwards, he had to consult internal players (NATO), due to the Turkey issue. This brought additional players and politics into the situation.
    • Ultimately, the missile crisis was resolved through the interplay of competing ideas and agendas, as well as a strong executive who could drive both XCOM and NATO when necessary.

Conclusion

  • “The conceptual models are much more than simple angles of vision or approaches. Each conceptual framework consists of a cluster of assumptions and categories that influence what the analyst finds puzzling, how he formulates the question, where he looks for evidence, and what he produces as an answer. The three cuts at the missile crisis demonstrate both the complexity of the models and how each offers different explanations.”
  • Model I views the crisis as a power play between Berlin and Cuba, with Russia and the United States as the chess pieces.
  • Model II illustrates why Russia couldn’t hide its deployment (SOPs and organizational standards) and why Kennedy chose a blockade (utilizing preexisting capacities).
  • Model III reveals that decisions were made by one person or organization, but a complex web of multiple players is involved, jockeying for influence and power.
  • If Kennedy were in today’s world of leaks, there would have been huge pressure to act faster, potentially removing his ability to let Russia see, think, then blink.
  • “The information demanded by Model II and Model III exceeds that needed by Model I. An armchair strategist (in Washington or even Cambridge or Charlottesville) can produce accounts of U.S. or Soviet national costs and benefits. Understanding the value-maximizing choices of nations demands chiefly an analytic ability in vicarious problem solving. Analyses that concentrate on capacities and outputs of organizations, or on bargaining among individuals, demand more information. Some observers (particularly players in the game) rely on a version of Model III for their own government’s behavior, while retreating to a Model I analysis of other nations. Thus information costs account for some differences among explanations.”
  • “If economists have yielded to business strategy analysts in addressing these questions, can students of international politics reasonably expect to do less? A theory of foreign policy is thus an inherent and inescapable component of a theory of international politics; likewise, a theory of the international setting is an essential component of a theory of the behavior of states in such settings. Systematic identification of causal factors at both levels is necessary to explain and predict phenomena in international affairs.”

🥰 Who Would Like it?

  • SAASS People

☠️ Agree, Disagree, or Suspend

Agree, wonder how useful model II and III are as predictive measures, because the intel is so hard to gain.