Diamond's fundamental move is to treat continents as cases in a quasi-experiment. You cannot run a true experiment on history (cannot rerun 1500 CE with different variables), but you can observe five independent domestication centers—Fertile Crescent/Eurasia, China, Mesoamerica, Andes, Sub-Saharan Africa—that developed independently, starting from similar conditions (Pleistocene hunter-gatherers with identical technology and intelligence), but diverging in their geographic circumstances. You can then ask: what geographic variables differ between these cases? How do those variables correlate with historical outcomes? If the same outcome appears across multiple cases wherever the same variable is present, that's evidence the variable caused the outcome. This transforms history from narrative (describing what happened) into science (explaining why it happened by identifying causal variables).1
The method requires treating historical cases as if they're experiments in progress: each continent is a case; each geographic variable is a condition that either is or isn't present; each historical outcome is a dependent variable that either appears or doesn't. Do this for multiple cases and you can identify patterns that single-case narrative cannot reveal. This is the method that enables Diamond's argument: not anecdotes about conquistadors or grand narratives about progress, but systematic comparison across cases showing that geographic variables predict outcomes.
What Makes a Variable Relevant
A relevant variable is one that differs between cases and plausibly affects outcomes. Continental axes (east-west vs. north-south) is relevant because: (1) it varies between continents, (2) it affects agricultural diffusion speed, and (3) diffusion speed plausibly affects technology accumulation. Domesticable animal availability is relevant because: (1) it varies by continent, (2) it affects agriculture and transport capability, and (3) those affect population density and state formation. Intelligence is not a relevant variable because: it doesn't meaningfully vary between continents (all humans have equal cognitive capacity), making it unsuitable for explaining continental differences.
What Constitutes a Valid Comparison
A valid comparison holds other factors constant while varying the target variable. Polynesian settlement is a valid comparison: Polynesians (constant) settled islands with different geographies (variable). Outcome differences (social complexity) correlate with geographic differences (island size, resources). But comparing Europe and China introduces confounds: both geography differs and cultural tradition differs and timeline differs. Which variable caused the outcome? Ambiguous. The best comparisons hold culture and initial conditions constant while varying geography—which is why Polynesian settlement, independent domestication centers, and similar-ecology regions with different climate are valuable cases.
The Challenge: Too Many Variables, Not Enough Cases
History offers only five independent domestication centers—five cases is small for identifying causal variables when many variables plausibly matter. This is the limitation of the case-study method: you're always working with small samples and potential confounds. The response is to look for consistent patterns across cases (if a variable appears in multiple cases and the outcome appears every time, that's tentative evidence) and to use natural experiments within cases (like Polynesia) to isolate variables.1
Eurasia: Horses, cattle, pigs, sheep, goats, camels, llamas
Americas: Llamas, alpacas (in Andes), no horses, no cattle
Sub-Saharan Africa: Elephants, Cape buffalo, zebras (not domesticable)
Correlation: Continents with domesticable large animals developed higher population density, more extensive states, and greater military capability. Continents without domesticable animals remained lower-density.1
Eurasia: East-west (9,000 miles wide)
Americas: North-south (9,000 miles tall)
Africa: Both dimensions roughly equal (5,000 x 5,000 miles)
Correlation: East-west continents developed superior agricultural technology through rapid diffusion. North-south continents remained more isolated in technology development.1
Eurasia: 10,000+ years of dense animal-human contact, creating epidemic diseases (smallpox, measles, plague, tuberculosis)
Americas: Different domestication history (llamas in Andes but not across continent), fewer epidemic diseases
Correlation: Eurasian contact with disease-naive American populations resulted in 90%+ mortality—not strategy, not technology, not genetics, but immunology.1
Fertile Crescent/Eurasia: Wheat, barley, peas, cattle, pigs, sheep, goats available in region → domestication
Mesoamerica: Maize available, but takes 7,000+ years to domesticate fully
New Guinea/Pacific Islands: Root crops available but lower caloric yield than grain
Correlation: Earlier domestication → earlier state formation → more time for technology accumulation → greater technological advantage by 1500 CE.1
A single variable (domesticable animals) triggers a cascade affecting multiple other variables:
Domesticable animals → Agriculture intensification → Population density increase → Surplus concentration → Administrative necessity → Writing emergence → States form → Taxation enables armies → Armies equipped with surplus-enabled metallurgy → Metallurgy produces weapons (steel, guns) → Armies with superior weapons conquer less-armed populations.
Each case that has domesticable animals enters this cascade. Each case that lacks domesticable animals doesn't enter the cascade (or enters it at lower intensity). By 1500 CE, after 13,000 years of cascade operation on Eurasian animals and 4,500 years on American animals, the gap in technology is vast. This gap is not due to intelligence differences, cultural differences, or moral differences. It's due to the cascade triggered by geographic distribution of domesticable animals—a distribution that has nothing to do with human choice or capability.
Tension 1: Animals vs. Axes vs. Disease
Do domesticable animals matter more than continental axes? Or does disease matter more? Diamond treats multiple variables as contributing to outcome, but the relative importance is unclear. If you remove domesticable animals but keep east-west axes, do civilizations still develop? The cases don't allow clean separation—Eurasia has all three advantages, Americas lack all three—making it hard to isolate individual variable contribution.
Tension 2: Variables as Necessary vs. Sufficient
Is each variable necessary for state formation, or just sufficient? Can a society reach state-level complexity without domesticable animals? The cases suggest no—but one case isn't proof. The Inca managed significant state-level complexity with only llamas and alpacas. Does this prove animals aren't necessary, or does it prove that absent animals, states require different structures (the Inca state was unique in relying on roads and runners rather than horses)?1
Tension 3: Variables as Determinative vs. Probabilistic
Do variables determine outcomes or merely increase probability? If an east-west continent makes technology diffusion likely but not certain, then cultures could still reject available technology (Japan with guns) or abandon superior technology (China with ships). The variables set the playing field; culture plays the game. But how much of outcome is field vs. play?
Diamond's case-study method is powerful but has a fundamental limitation: he compares five major domestication centers and several island systems. This is a small sample for identifying universal causal variables. He can show patterns (where the pattern holds, it's suggestive), but not prove causation rigorously. A skeptic could argue that correlation isn't causation—perhaps geography correlates with outcome through some third variable he hasn't identified. Diamond handles this by showing the mechanism: how domesticable animals drive the cascade, how axes affect diffusion, how disease creates selection pressure. Mechanism plus correlation plus multiple independent cases makes a strong (though not airtight) argument. But the method's limitation remains: you cannot run experiments on history, so all causal claims rest on suggestive evidence rather than proof.1
Comparative Ethnography and Functional Equivalence — Anthropologists use case-comparison to identify functional requirements independent of cultural implementation. For example, all societies need conflict resolution, but they implement it differently (councils, courts, ordeals, duels, etc.). The anthropological insight: you can identify universal needs by finding their different implementations across cultures. Diamond's method applies this to history: identify what Eurasia had that Americas didn't (domesticable animals, east-west axes), then explain why outcomes differed. The difference in implementation isn't cultural choice—it's structural response to different resource bases. The insight that transfers: comparing cases reveals functional requirements that no single case reveals. Looking at Japan alone doesn't reveal that hierarchical organization requires surplus; comparing Japan with small tribes does. Looking at Eurasia alone doesn't reveal that domesticable animals matter; comparing Eurasia with Americas does. Case comparison is how you find universal principles hiding in local variation.
Falsifiability in Historical Science — Karl Popper argued that only falsifiable claims are scientific. History seems unfalsifiable—we can't rerun the past to test claims. But Diamond's case method creates quasi-falsifiability: if domesticable animals are the primary driver of state formation, then every continent with domesticable animals should show state formation (or clear obstacles to it), and every continent without should not. This prediction is falsifiable—if a continent without domesticable animals develops states anyway, the theory is wrong. By proposing falsifiable predictions and showing they mostly hold, Diamond makes history more scientific than traditional narrative. The insight that transfers: you can make historical claims more or less scientific by grounding them in variables that can be identified across multiple cases and predicting outcomes. Narrative history describes what happened; scientific history predicts what would happen if conditions were different. This distinction enables testing historical claims against evidence more rigorously than narrative alone.
The Sharpest Implication
If continental case comparison is the valid method for historical science, then all single-region histories are incomplete—they're missing the control cases necessary to identify causes. European history looked like the story of European genius until you compare Europe with similar-sized continents and ask: why did Eurasia develop faster than Americas despite identical human intelligence? The answer comes from comparing cases, not from studying Europe alone. This means the entire discipline of history (which traditionally studies single regions in depth) has methodological limitations. To identify causes, you need comparison. Depth without breadth produces description, not explanation. The uncomfortable implication: most historical scholarship is pre-scientific because it describes within-region variation without comparing regions to identify cross-regional patterns. This doesn't make traditional history useless—deep understanding of one region matters—but it limits the ability to identify causal variables. Scientific history requires both: case studies for depth and case comparison for identifying universal principles.1
Generative Questions