Case Study 2: Military Strategy and Grocery Shopping -- Satisficing at Every Scale

"In preparing for battle I have always found that plans are useless, but planning is indispensable." -- Dwight D. Eisenhower

Two Decisions, One Pattern

At first glance, choosing a box of cereal and planning a military campaign have nothing in common. One is trivial, the other is life-and-death. One takes thirty seconds, the other takes months. One affects your breakfast, the other affects nations.

And yet both decisions share the same deep structure: a decision-maker facing a complex environment with incomplete information, limited time, too many options to evaluate exhaustively, and the need to commit to a course of action without knowing whether a better one exists. In both cases, the decision-maker satisfices. And in both cases, satisficing is not a compromise -- it is the rational strategy.

This case study traces the satisficing pattern at two radically different scales to show that the same structural logic operates whether the stakes are a few dollars or a few thousand lives.

Part I: The War That Could Not Be Optimized

The Problem of the Western Front

By late 1914, the Western Front of World War I had solidified into a continuous line of trenches stretching from the English Channel to the Swiss border. Both sides faced an identical problem: how to break through the enemy's fortified line, advance into open territory, and win the war.

The problem was, in principle, an optimization problem. There was a finite number of potential attack points along the front. Each point had measurable characteristics: the width of no-man's-land, the depth of the enemy's defenses, the quality of the terrain, the availability of roads for supply, the vulnerability to flanking. A sufficiently powerful analytical framework could, in theory, identify the optimal point of attack -- the place where a breakthrough was most likely at the lowest cost in lives and materiel.

The commanders tried. They studied maps, analyzed intelligence reports, surveyed terrain, and calculated force ratios. But the optimization problem they faced was vastly more complex than it appeared. The enemy's defenses were not static -- they shifted in response to intelligence about the attacker's preparations. The terrain's suitability depended on weather, which was unpredictable. The morale and combat effectiveness of troops varied in ways that could not be quantified. The interaction effects between artillery preparation, infantry assault timing, reserve deployment, and logistical support created a combinatorial explosion of possibilities that no staff could fully analyze.

The result was a series of offensives that failed catastrophically. The Battle of the Somme (1916), the Battle of Passchendaele (1917), the Nivelle Offensive (1917) -- each was planned with meticulous attention to detail, each represented the best thinking of experienced military professionals, and each produced horrifying casualties for minimal territorial gain. The plans were not failures of intelligence or effort. They were failures of optimization -- attempts to find the perfect plan in a domain where perfection was unattainable.

Von Moltke's Alternative

Helmuth von Moltke the Elder, who had commanded Prussian forces a generation earlier, would not have been surprised. Von Moltke had articulated a fundamentally different philosophy of military planning, one that recognized the limits of optimization and embraced satisficing as a deliberate strategy.

Von Moltke's approach rested on several principles:

Plan for the first contact, not for the whole campaign. Von Moltke recognized that detailed plans become obsolete the moment the battle begins, because the enemy's responses, the weather, and countless other factors introduce unpredictable variation. Rather than planning the entire campaign in detail (optimization), he planned the initial deployment and the first major engagement, then relied on his commanders to adapt (satisficing). The plan provided a good-enough starting position; adaptation provided the rest.

Specify objectives, not methods. Von Moltke told his subordinate commanders what to achieve, not how to achieve it. This is Auftragstaktik -- mission-type tactics. A corps commander might be told: "Seize the crossroads at Koniggratz by nightfall." He was not told which roads to use, which formation to employ, or how to deal with enemy resistance along the way. He was trusted to find a method that was good enough for the local conditions he actually faced, rather than being bound to a method that was optimal for conditions assumed in advance.

Cultivate judgment, not compliance. Von Moltke invested heavily in training officers to make independent decisions under uncertainty. He understood that a system designed for adaptation needs decision-makers who can satisfice effectively -- who can assess a situation rapidly, identify a workable course of action, and execute without waiting for orders from above. This is the military equivalent of Gigerenzer's adaptive toolbox: a repertoire of strategies that officers could select from based on their assessment of the situation.

Accept friction. Carl von Clausewitz, the great Prussian military theorist, had coined the concept of "friction" -- the accumulated effect of small uncertainties, miscommunications, equipment failures, and human errors that make real war fundamentally different from war on paper. Von Moltke absorbed this lesson. He did not try to eliminate friction (which would be optimization). He planned for it (which is satisficing). His plans included margins, reserves, and fallback positions -- the military equivalents of engineering tolerances.

The Modern Military Embrace of Satisficing

Von Moltke's philosophy has become the dominant paradigm in modern Western military doctrine, though it took another century and many painful lessons to get there.

The U.S. Marine Corps' doctrinal publication Warfighting (1989) states the philosophy explicitly: "We must be content to suggest the general nature and direction of the attack, avoid the sterile pursuit of certainty, and accept reasonable risk." The emphasis on accepting "reasonable risk" rather than minimizing all risk is pure satisficing. Minimizing all risk is an optimization problem with no feasible solution. Accepting reasonable risk is satisficing -- identifying a threshold of acceptable risk and planning accordingly.

The concept of the OODA loop -- Observe, Orient, Decide, Act -- developed by Air Force Colonel John Boyd, provides the procedural framework. Boyd argued that military advantage goes not to the side with the best plan but to the side that can cycle through the OODA loop fastest. Speed of decision beats quality of decision, because a good decision now is better than a perfect decision too late. This is satisficing elevated to a principle of competitive advantage: the satisficer who acts quickly defeats the optimizer who acts slowly, because the environment changes faster than the optimizer can compute.

Boyd's insight connects directly to the explore/exploit framework from Chapter 8. The OODA loop is a rapid alternation between exploration (Observe, Orient) and exploitation (Decide, Act). Each cycle through the loop is a miniature satisficing episode: observe the situation, identify a workable response, execute it, and then observe again. The speed of the loop matters more than the optimality of any individual decision because each decision is immediately followed by feedback that enables correction. Rapid satisficing with feedback is more effective than slow optimization without it.

A Modern Example: Mission Command in Afghanistan

The U.S. military's experience in Afghanistan illustrates both the power and the limits of satisficing in military operations. Early in the conflict, commanders in the field faced a deeply uncertain environment: an unfamiliar culture, an adaptive enemy, ambiguous intelligence, and rules of engagement that shifted with political winds.

The most effective units, by most accounts, were those that practiced mission command -- the modern descendant of Auftragstaktik. Platoon leaders and company commanders were given objectives ("secure this village," "disrupt insurgent logistics in this valley") and significant latitude in how to achieve them. They satisficed constantly: assessing the local situation, choosing a course of action that seemed workable, executing, observing the results, and adjusting. They did not wait for optimal intelligence before acting, because optimal intelligence never arrived. They did not seek the perfect plan, because the perfect plan did not exist in an environment where the enemy adapted daily and tribal alliances shifted weekly.

The failures, conversely, often came from attempts to optimize at higher levels. Centralized plans that specified detailed tactics for ground units based on intelligence that was days or weeks old repeatedly failed on contact with reality. Rules of engagement optimized for legal compliance sometimes prevented soldiers from responding effectively to threats. Metrics optimized for bureaucratic reporting (body counts, patrols conducted, meetings held) sometimes distorted behavior on the ground -- a classic instance of Goodhart's Law operating within a military bureaucracy.

Part II: The Supermarket as Decision Environment

The Scale of the Problem

Consider a routine grocery shopping trip. A typical American supermarket carries between 30,000 and 50,000 distinct products. A typical shopping trip involves selecting between 30 and 60 items. Each item selection involves choosing among dozens to hundreds of alternatives varying in brand, size, flavor, price, nutritional content, ingredient quality, and packaging.

The optimization problem is staggering. Even if each selection involves only 20 alternatives and only 5 relevant dimensions, a shopper making 40 selections would need to evaluate 800 options across 4,000 dimensions -- and that is before considering interactions (this cereal goes better with that milk, this sauce works with that pasta). A rigorous utility-maximizing analysis of a single grocery trip would take weeks, require information the shopper does not have (how will this unfamiliar brand actually taste?), and produce a result that is immediately obsolete (prices change, items go out of stock, preferences shift with mood).

Nobody optimizes grocery shopping. Everybody satisfices. The question is how.

How Shoppers Actually Decide

Decades of consumer behavior research have revealed the heuristics shoppers actually use. They are textbook examples of Gigerenzer's fast-and-frugal heuristics:

Habit. The most common grocery shopping heuristic is to buy the same things you bought last time. This eliminates the decision entirely -- no search, no comparison, no evaluation. It is the ultimate satisficing strategy: the previous choice was good enough, so repeat it. Research suggests that roughly 85 percent of grocery purchases are repeat purchases, meaning the vast majority of "decisions" in a grocery store are not decisions at all. They are habits.

Recognition. When shoppers encounter an unfamiliar category or have decided to try something new, they typically choose the brand they recognize. This is Gigerenzer's recognition heuristic in its natural habitat. Recognition is correlated with market share, which is correlated with at least adequate quality (brands with terrible products do not achieve wide recognition because they do not survive in the market). The recognition heuristic exploits this correlation: "I have heard of this brand, so it is probably acceptable." It ignores all other information -- price, nutrition, ingredients, reviews -- and it often produces good-enough results with near-zero cognitive effort.

Take-the-best on a single cue. Shoppers who do compare options typically use a single criterion: price, or brand loyalty, or a specific nutritional claim (organic, sugar-free, whole grain). They do not weigh multiple criteria simultaneously. This is the take-the-best heuristic: identify the most important dimension, compare options on that dimension alone, choose the winner, ignore everything else.

Social proof. Some shoppers use the heuristic "buy what other people buy" -- choosing the product with the most shelf space, the most prominent placement, or the one they have seen others put in their carts. This is satisficing by delegation: rather than computing your own utility function, you rely on the aggregated choices of other shoppers as a signal of adequate quality.

Why Satisficing Works in the Supermarket

These heuristics seem crude. A rational optimizer would be appalled. And yet they work, for precisely the reasons Gigerenzer's research predicts: the decision environment has a structure that the heuristics exploit.

Most options are adequate. In a competitive market, products that are genuinely terrible do not survive. The quality floor is high. Most cereals are acceptable. Most pasta sauces are fine. Most brands of milk are essentially identical. When the variation among options is small, the cost of choosing a suboptimal option is negligible, and the benefit of exhaustive search is correspondingly tiny.

The cost of error is low. If you buy a cereal you do not like, the consequence is a mildly disappointing breakfast. You do not buy it again. The error is self-correcting and low-cost. In environments where errors are cheap and reversible, satisficing is nearly costless. Optimization is valuable mainly when errors are expensive and irreversible -- which is why you satisfice on cereal but invest more effort in choosing a surgeon.

Time is scarce. The time spent optimizing one grocery selection is time not spent on the other 39 selections, or on the rest of your life. The satisficer who spends five seconds per item and gets through the store in twenty minutes has preserved fifty minutes compared to the optimizer who spends two minutes per item. Over a year of weekly shopping, this amounts to roughly forty hours -- an entire work week -- spent on marginal improvements in grocery selection quality.

Information is unreliable. Nutrition labels are confusing and sometimes misleading. Ingredient lists require expertise to interpret. Price comparisons are complicated by differing package sizes. Taste is subjective and unpredictable. In an environment where the available information is noisy and hard to interpret, the value of processing more information is low -- the additional information may be as likely to mislead as to inform. This is the less-is-more effect at work: the shopper who uses one cue (recognition) may make better choices than the shopper who attempts to integrate five cues, because the additional cues introduce noise without improving accuracy.

The Shared Structure

The military commander and the grocery shopper face problems that differ in stakes, time scale, complexity, and consequence. But the structural logic of their decision-making is identical:

The critical insight is that the same satisficing structure operates at both scales because the underlying constraints are the same: too many options, too little information, too little time, and the impossibility of identifying the objectively best choice before committing to action.

The Spectrum of Stakes

The comparison between military strategy and grocery shopping reveals something important about where satisficing falls on the spectrum of decision strategies. The appropriate level of satisficing -- how good "good enough" needs to be -- depends on the stakes.

For groceries, a very low threshold works. Almost any acceptable product will do, and the cost of a mistake is trivial. The appropriate heuristic is the fastest and frugalest: habit, recognition, or a single cue.

For military operations, the threshold must be higher. A "workable" plan must account for enemy capabilities, terrain, logistics, and a host of other factors. But it is still a threshold, not an optimization. The commander seeks a plan that is good enough to achieve the objective with acceptable risk -- not the theoretically optimal plan, which is uncomputable.

For the most consequential decisions -- whether to go to war, whether to employ nuclear weapons, whether to accept a peace settlement -- the threshold is highest of all. But even here, satisficing operates. Decision-makers cannot enumerate and evaluate all possible courses of action. They generate a small number of options, assess them against criteria of acceptability, and choose. The Cuban Missile Crisis of 1962, often cited as an example of careful deliberation, involved President Kennedy and his advisors considering only a handful of options (do nothing, diplomatic pressure, naval blockade, air strike, invasion) and assessing them against a threshold of acceptability ("Which of these avoids nuclear war while addressing the Soviet threat?"). They did not compute optimal game-theoretic strategies. They satisficed, under the most extreme stakes imaginable.

The lesson is not that all decisions should receive the same level of effort. The lesson is that all decisions involve satisficing -- but the threshold of "good enough" should be calibrated to the stakes. This is itself a satisficing principle: invest just enough effort in the decision to meet the threshold, and no more. Do not over-invest in low-stakes decisions (the cereal aisle). Do not under-invest in high-stakes decisions (the military campaign). But recognize that even the highest-stakes decisions are, at bottom, exercises in finding a course of action that is good enough given the constraints -- not exercises in computing the theoretically optimal course of action.

Questions for Discussion

1. The case study argues that the most effective military units in Afghanistan practiced satisficing (mission command) while the least effective were those that tried to optimize from above (centralized detailed planning). Does this pattern hold in other organizations you are familiar with? Can you identify a case where an organization failed because it tried to optimize rather than satisfice?

2. If 85 percent of grocery purchases are repeat purchases (habit), then most grocery "decisions" are not decisions at all. Is this a failure of rationality or a success of satisficing? What would happen if you actually deliberated over every item on every shopping trip?

3. The case study presents military command and grocery shopping as structurally identical forms of satisficing. What is the most important structural difference between the two, and how does it affect the appropriate satisficing strategy?

4. Boyd's OODA loop suggests that speed of decision beats quality of decision. Is this always true? Can you think of domains where it is better to decide slowly and well than quickly and approximately? What features of those domains make the difference?

5. The Cuban Missile Crisis is described as an exercise in satisficing under extreme stakes. How does this challenge the common assumption that more important decisions require more optimization? What would it have looked like for Kennedy's advisors to try to "optimize" their response to the crisis?