Collapse: How Societies Choose to Fail or Succeed

by Jared Diamond

  Kanama has very fertile volcanic soil, so that its population density is high even by the standards of densely populated Rwanda: 1,740 people per square mile in 1988, rising to 2,040 in 1993. (That’s higher even than the value for Bangladesh, the world’s most densely populated agricultural nation.) Those high population densities translated into very small farms: a median farm size of only 0.89 acre in 1988, declining to 0.72 acre in 1993. Each farm was divided into (on average) 10 separate parcels, so that farmers were tilling absurdly small parcels averaging only 0.09 acre in 1988 and 0.07 acre in 1993.

  Because all land in the commune was already occupied, young people found it difficult to marry, leave home, acquire a farm, and set up their own household. Increasingly, young people postponed marriage and continued to live at home with their parents. For instance, in the 20- to 25-year-old age bracket, the percentage of young women living at home rose between 1988 and 1993 from 39% to 67%, and the percentage of young men rose from 71% to 100%: not a single man in his early 20s lived independently of his parents by 1993. That obviously contributed to the lethal family tensions that exploded in 1994, as I shall explain below. With more young people staying home, the average number of people per farm household increased (between 1988 and 1993) from 4.9 to 5.3, so that the land shortage was even tighter than indicated by the decrease in farm size from 0.89 to 0.72 acre. When one divides decreasing farm area by increasing number of people in the household, one finds that each person was living off of only one-fifth of an acre in 1988, declining to one-seventh of an acre in 1993.

  Not surprisingly, it proved impossible for most people in Kanama to feed themselves on so little land. Even when measured against the low calorie intake considered adequate in Rwanda, the average household got only 77% of its calorie needs from its farm. The rest of its food had to be bought with income earned off the farm, at jobs such as carpentry, brick-making, sawing wood, and trade. Two-thirds of households held such jobs, while one-third didn’t. The percentage of the population consuming less than 1,600 calories per day (i.e., what is considered below the famine level) was 9% in 1982, rising to 40% in 1990 and some unknown higher percentage thereafter.

  All of these numbers that I have quoted so far for Kanama are average numbers, which conceal inequalities. Some people owned larger farms than others, and that inequality increased from 1988 to 1993. Let’s define a “very big” farm as larger than 2.5 acres, and a “very small” farm as smaller than 0.6 acre. (Think back to Chapter 1 to appreciate the tragic absurdity of those numbers: I mentioned there that in Montana a 40-acre farm used to be considered necessary to support a family, but even that is now inadequate.) Both the percentage of very big farms and the percentage of very small farms increased between 1988 and 1993, from 5 to 8% and from 36 to 45% respectively. That is, Kanama farm society was becoming increasingly divided between the rich haves and the poor have-nots, with decreasing numbers of people in the middle. Older heads of households tended to be richer and to have larger farms: those in the age ranges 50-59 and 20-29 years old had average farm sizes of 2.05 acres and only 0.37 acre respectively. Of course, family size was larger for the older household heads, so they needed more land, but they still had three times more land per household member than did young household heads.

  Paradoxically, off-farm income was earned disproportionately by owners of large farms: the average size of farms that did earn such income was 1.3 acres, compared to only half an acre for farms lacking such income. That difference is paradoxical because the smaller farms are the ones whose household members have less farmland per person to feed themselves, and which thus need more off-farm income. That concentration of off-farm income on the larger farms contributed to the increasing division of Kanama society between haves and have-nots, with the rich becoming richer and the poor becoming poorer. In Rwanda, it’s supposedly illegal for owners of small farms to sell any of their land. In fact, it does happen. Investigation of land sales showed that owners of the smallest farms sold land mainly when they needed money for an emergency involving food, health, lawsuit costs, bribes, a baptism, wedding, funeral, or excessive drinking. In contrast, owners of large farms sold for reasons such as to increase farm efficiency (e.g., selling a distant parcel of land in order to buy a parcel nearer to the farmhouse).

  The extra off-farm income of larger farms allowed them to buy land from smaller farms, with the result that large farms tended to buy land and become larger, while small farms tended to sell land and become smaller. Almost no large farm sold land without buying any, but 35% of the smallest farms in 1988, and 49% of them in 1993, sold without buying. If one breaks down land sales according to off-farm income, all farms with off-farm income bought land, and none sold land without buying; but only 13% of farms lacking off-farm income bought land, and 65% of them sold land without buying. Again, note the paradox: already-tiny farms, which desperately needed more land, in fact became smaller, by selling land in emergencies to large farms financing their purchases with off-farm income. Remember again that what I term “large farms” are large only by Rwanda standards: “large” means “larger than a mere 1 or 2 acres.”

  Thus, at Kanama most people were impoverished, hungry, and desperate, but some people were more impoverished, hungry, and desperate than others, and most people were becoming more desperate while a few were becoming less desperate. Not surprisingly, this situation gave rise to frequent serious conflicts that the parties involved could not resolve by themselves, and that they either referred to traditional village conflict mediators or (less often) brought to the courts. Each year, households reported on the average more than one such serious conflict requiring outside resolution. André and Platteau surveyed the causes of 226 such conflicts, as described either by the mediators or by the householders. According to both types of informants, land disputes lay at the root of most serious conflicts: either because the conflict was directly over land (43% of all cases); or because it was a husband/wife, family, or personal dispute often stemming ultimately from a land dispute (I’ll give examples in the next two paragraphs); or else because the dispute involved theft by very poor people, known locally as “hunger thieves,” who owned almost no land and were without off-farm income and who lived by stealing for lack of other options (7% of all disputes, and 10% of all households).

  Those land disputes undermined the cohesion of Rwandan society’s traditional fabric. Traditionally, richer landowners were expected to help their poorer relatives. This system was breaking down, because even the landowners who were richer than other landowners were still too poor to be able to spare anything for poorer relatives. That loss of protection especially victimized vulnerable groups in the society: separated or divorced women, widows, orphans, and younger half-siblings. When ex-husbands ceased to provide for their separated or divorced wives, the women would formerly have returned to their natal family for support, but now their own brothers opposed their return, which would make the brothers or the brothers’ children even poorer. The women might then seek to return to their natal family only with their daughters, because Rwandan inheritance was traditionally by sons, and the woman’s brothers wouldn’t see her daughters as competing with their own children. The woman would leave her sons with their father (her divorced husband), but his relatives might then refuse land to her sons, especially if their father died or ceased protecting them. Similarly, a widow would find herself without support from either her husband’s family (her brothers-in-law) or from her own brothers, who again saw the widow’s children as competing for land with their children. Orphans were traditionally cared for by paternal grandparents; when those grandparents died, the orphans’ uncles (the brothers of their deceased father) now sought to disinherit or evict the orphans. Children of polygamous marriages, or of broken marriages in which the man subsequently remarried and had children by a new wife, found themselves disinherited or evicted by their half-brothers.

  The most painful and socially disruptive land disputes were those pitting fathers
against sons. Traditionally, when a father died, his land all passed to his oldest son, who was expected to manage the land for the whole family and to provide his younger brothers with enough land for their subsistence. As land became scarce, fathers gradually switched to the custom of dividing their land among all sons, in order to reduce the potential for intrafamily conflict after the father’s death. But different sons urged on their father different competing proposals for dividing the land. Younger sons became bitter if older brothers, who got married first, received a disproportionately large share—e.g., because the father had had to sell off some land by the time younger sons got married. Younger sons instead demanded strictly equal divisions; they objected to their father giving their older brother a present of land on that brother’s marriage. The youngest son, who traditionally was the one expected to care for his parents in their old age, needed or demanded an extra share of land in order to carry out that traditional responsibility. Brothers were suspicious of, and sought to evict, sisters or younger brothers who received from the father any present of land, which the brothers suspected was being given in return for that sister or younger brother agreeing to care for the father in his old age. Sons complained that their father was retaining too much land to support himself in his old age, and they demanded more land now for themselves. Fathers in turn were justifiably terrified of being left with too little land in their old age, and they opposed their sons’ demands. All of these types of conflicts ended up before mediators or the courts, with fathers suing sons and vice versa, sisters suing brothers, nephews suing uncles, and so on. These conflicts sabotaged family ties, and turned close relatives into competitors and bitter enemies.

  That situation of chronic and escalating conflict forms the background against which the killings of 1994 took place. Even before 1994, Rwanda was experiencing rising levels of violence and theft, perpetrated especially by hungry landless young people without off-farm income. When one compares crime rates for people of age 21-25 among different parts of Rwanda, most of the regional differences prove to be correlated statistically with population density and per-capita availability of calories: high population densities and worse starvation were associated with more crime.

  After the explosion of 1994, André tried to track down the fates of Kanama’s inhabitants. She found that 5.4% were reported to her as having died as a result of the war. That number is an underestimate of the total casualties, because there were some inhabitants about whose fates she could obtain no information. Hence it remains unknown whether the death rate approached the average value of 11% for Rwanda as a whole. What is clear is that the death rate in an area where the population consisted almost entirely of Hutu was at least half of the death rate in areas where Hutu were killing Tutsi plus other Hutu.

  All but one of the known victims at Kanama fell into one of six categories. First, the single Tutsi at Kanama, a widowed woman, was killed. Whether that had much to do with her being Tutsi is unclear, because she furnished so many other motives for killing: she had inherited much land, she had been involved in many land disputes, she was the widow of a polygamous Hutu husband (hence viewed as a competitor of his other wives and their families), and her deceased husband had already been forced off his land by his half-brothers.

  Two more categories of victims consisted of Hutu who were large landowners. The majority of them were men over the age of 50, hence at a prime age for father/son disputes over land. The minority were younger people who had aroused jealousy by being able to earn much off-farm income and using it to buy land.

  A next category of victims consisted of “troublemakers” known for being involved in all sorts of land disputes and other conflicts.

  Still another category was young men and children, particularly ones from impoverished backgrounds, who were driven by desperation to enlist in the warring militias and proceeded to kill each other. This category is especially likely to have been underestimated, because it was dangerous for André to ask too many questions about who had belonged to what militia.

  Finally, the largest number of victims were especially malnourished people, or especially poor people with no or very little land and without off-farm income. They evidently died because of starvation, being too weak, or not having money to buy food or to pay the bribes required to buy their survival at roadblocks.

  Thus, as André and Platteau note, “The 1994 events provided a unique opportunity to settle scores, or to reshuffle land properties, even among Hutu villagers . . . . It is not rare, even today, to hear Rwandans argue that a war is necessary to wipe out an excess of population and to bring numbers into line with the available land resources.”

  That last quote of what Rwandans themselves say about the genocide surprised me. I had thought that it would be exceptional for people to recognize such a direct connection between population pressure and killings. I’m accustomed to thinking of population pressure, human environmental impacts, and drought as ultimate causes, which make people chronically desperate and are like the gunpowder inside the powder keg. One also needs a proximate cause: a match to light the keg. In most areas of Rwanda, that match was ethnic hatred whipped up by politicians cynically concerned with keeping themselves in power. (I say “most areas,” because the large-scale killings of Hutu by Hutu at Kanama demonstrate a similar outcome even where everybody belonged to the same ethnic group.) As Gérard Prunier, a French scholar of East Africa, puts it, “The decision to kill was of course made by politicians, for political reasons. But at least part of the reason why it was carried out so thoroughly by the ordinary rank-and-file peasants in their ingo [= family compound] was feeling that there were too many people on too little land, and that with a reduction in their numbers, there would be more for the survivors.”

  The link that Prunier, and that André and Platteau, see behind population pressure and the Rwandan genocide has not gone unchallenged. In part, the challenges are reactions to oversimplified statements that critics with some justice lampooned as “ecological determinism.” For instance, only 10 days after the genocide began, an article in an American newspaper linked Rwanda’s dense population to the genocide by saying, “Rwandas [i.e., similar genocides] are endemic, built-in, even, to the world we inhabit.” Naturally, that fatalistic oversimplified conclusion provokes negative reactions not only to it, but also to the more complex view that Prunier, André and Platteau, and I present, for three reasons.

  First, any “explanation” of why a genocide happened can be misconstrued as “excusing” it. However, regardless of whether we arrive at an oversimplified one-factor explanation or an excessively complex 73-factor explanation for a genocide doesn’t alter the personal responsibility of the perpetrators of the Rwandan genocide, as of other evil deeds, for their actions. This is a misunderstanding that arises regularly in discussions of the origins of evil: people recoil at any explanation, because they confuse explanations with excuses. But it is important that we understand the origins of the Rwandan genocide—not so that we can exonerate the killers, but so that we can use that knowledge to decrease the risk of such things happening again in Rwanda or elsewhere. Similarly, there are people who have chosen to devote their lives or careers to understanding the origins of the Nazi Holocaust, or to understanding the minds of serial murderers and rapists. They have made that choice not in order to mitigate the responsibility of Hitler, serial murderers, and rapists, but because they want to know how those awful things came to be, and how we can best prevent recurrences.

  Second, it is justifiable to reject the simplistic view that population pressure was the single cause of the Rwandan genocide. Other factors did contribute; in this chapter I have introduced ones that seem to me important, and experts on Rwanda have written entire books and articles on the subject, cited in my Further Readings at the back of this book. Just to reiterate: regardless of the order of their importance, those other factors included Rwanda’s history of Tutsi domination of Hutu, Tutsi large-scale killings of Hutu in Burundi an
d small-scale ones in Rwanda, Tutsi invasions of Rwanda, Rwanda’s economic crisis and its exacerbation by drought and world factors (especially by falling coffee prices and World Bank austerity measures), hundreds of thousands of desperate young Rwandan men displaced as refugees into settlement camps and ripe for recruitment by militias, and competition among Rwanda’s rival political groups willing to stoop to anything to retain power. Population pressure joined with those other factors.

  Finally, one should not misconstrue a role of population pressure among the Rwandan genocide’s causes to mean that population pressure automatically leads to genocide anywhere around the world. To those who would object that there is not a necessary link between Malthusian population pressure and genocide, I would answer, “Of course!” Countries can be overpopulated without descending into genocide, as exemplified by Bangladesh (relatively free of large-scale killings since its genocidal slaughters of 1971) as well as by the Netherlands and multi-ethnic Belgium, despite all three of those countries being more densely populated than Rwanda. Conversely, genocide can arise for ultimate reasons other than overpopulation, as illustrated by Hitler’s efforts to exterminate Jews and Gypsies during World War II, or by the genocide of the 1970s in Cambodia, with only one-sixth of Rwanda’s population density.

  Instead, I conclude that population pressure was one of the important factors behind the Rwandan genocide, that Malthus’s worst-case scenario may sometimes be realized, and that Rwanda may be a distressing model of that scenario in operation. Severe problems of overpopulation, environmental impact, and climate change cannot persist indefinitely: sooner or later they are likely to resolve themselves, whether in the manner of Rwanda or in some other manner not of our devising, if we don’t succeed in solving them by our own actions. In the case of Rwanda’s collapse we can put faces and motives on the unpleasant solution; I would guess that similar motives were operating, without our being able to associate them with faces, in the collapses of Easter Island, Mangareva, and the Maya that I described in Part 2 of this book. Similar motives may operate again in the future, in some other countries that, like Rwanda, fail to solve their underlying problems. They may operate again in Rwanda itself, where population today is still increasing at 3% per year, women are giving birth to their first child at age 15, the average family has between five and eight children, and a visitor’s sense is of being surrounded by a sea of children.