by Morris, Ian;
Fifty years ago the philosopher Karl Popper argued that progress in science is a matter of “conjectures and refutations,” following a zigzag course as one researcher throws out an idea and others scramble to disprove it, in the process coming up with better ideas. The same, I think, applies to history. I am confident that any index that stays close to the evidence will produce more or less the same pattern as mine, but if I am wrong, and if others find this scheme wanting, hopefully my failure will encourage them to uncover better answers. To quote Einstein one more time, “There could be no fairer destiny for any theory … than that it should point the way to a more comprehensive theory in which it lives on.”
WHEN AND WHERE TO MEASURE?
Two final technical issues. First, how often should we calculate the scores? If we wanted to, we could trace changes in social development from year to year or even month to month since the 1950s. I doubt that there would be much point, though. After all, we want to see the overall shape of history across very long periods, and for that—as I hope to show in what follows—taking the pulse of social development once every century seems to provide enough detail.
As we move back toward the end of the Ice Age, though, checking social development on a century-by-century basis is neither possible nor particularly desirable. We just can’t tell much difference between what was going on in 14,000 and the situation in 13,900 BCE (or 13,800 for that matter), partly because we don’t have enough good evidence and partly because change just happened very slowly. I therefore use a sliding scale. From 14,000 through 4000 BCE, I measure social development every thousand years. From 4000 through 2500 BCE the quality of evidence improves and change accelerates, so I measure every five hundred years. I reduce this to every 250 years between 2500 BCE and 1500 BCE, and finally measure every century from 1400 BCE through 2000 CE.
This has its risks, most obviously that the further back in time we go, the smoother and more gradual change will look. By calculating scores only every thousand or five hundred years we may well miss something interesting. The hard truth, though, is that only occasionally can we date our information much more precisely than the ranges I suggest. I do not want to dismiss this problem out of hand, and will try in the narrative in Chapters 4 through 10 to fill in as many of the gaps as possible, but the framework I use here does seem to me to offer the best balance between practicality and precision.
The other issue is where to measure. You may have been struck while reading the last section by my coyness about just what part of the world I was talking about when I generated numbers for “West” and “East.” I spoke at some points about the United States and at others about Britain; sometimes of China, sometimes of Japan. Back in Chapter 1 I described the historian Kenneth Pomeranz’s complaints about how comparative historians often skew analysis of why the West rules by sloppily comparing tiny England with enormous China and concluding that the West already led the East by 1750 CE. We must, he insisted, compare like-sized units. I spent Chapters 1 and 2 responding to this by defining West and East explicitly as the societies that have descended from the original Western and Eastern agricultural revolutions in the Hilly Flanks and the Yellow and Yangzi river valleys; now it is time to admit that that resolved only part of Pomeranz’s problem. In Chapter 2, I described the spectacular expansion of the Western and Eastern zones in the five thousand or so years after cultivation began and the differences in social development that often existed between core areas such as the Hilly Flanks or Yangzi Valley and peripheries such as northern Europe or Korea; so which parts of the East and West should we focus on when working out scores for the index of social development?
We could try looking at the whole of the Eastern and Western zones, although that would mean that the score for, say, 1900 CE would bundle together the smoking factories and rattling machine guns of industrialized Britain with Russia’s serfs, Mexico’s peons, Australia’s ranchers, and every other group in every corner of the vast Western zone. We would then have to concoct some sort of average development score for the whole Western region, then do it again for the East, and repeat the process for every earlier point in history. This would get so complicated as to become impractical, and I suspect it would be rather pointless anyway. When it comes to explaining why the West rules, the most important information normally comes from comparing the most highly developed parts of each region, the cores that were tied together by the densest political, economic, social, and cultural interactions. The index of social development needs to measure and compare changes within these cores.
As we will see in Chapters 4–10, though, the core areas have themselves shifted and changed across time. The Western core was geographically actually very stable from 11,000 BCE until about 1400 CE, remaining firmly at the eastern end of the Mediterranean Sea except for the five hundred years between about 250 BCE and 250 CE, when the Roman Empire drew it westward to include Italy. Otherwise, it always lay within a triangle formed by what are now Iraq, Egypt, and Greece. Since 1400 CE it has moved relentlessly north and west, first to northern Italy, then to Spain and France, then broadening to include Britain, Belgium, Holland, and Germany. By 1900 it straddled the Atlantic and by 2000 was firmly planted in North America. In the East the core remained in the original Yellow-Yangzi zone right up until 1800 CE, although its center of gravity shifted northward toward the Yellow River’s central plain after about 4000 BCE, back south to the Yangzi Valley after 500 CE, and gradually north again after 1400. It expanded to include Japan by 1900 and southeast China by 2000 (Figure 3.2). For now I just want to note that all the social development scores reflect the societies in these core areas; why the cores shifted will be one of our major concerns in Chapters 4 through 10.
THE PATTERN OF THE PAST
So much for the rules of the game; now for some results. Figure 3.3 shows the scores across the last sixteen thousand years, since things began warming up at the end of the Ice Age.
Figure 3.2. Shifting centers of power: the sometimes slow, sometimes rapid relocation of the most highly developed core within the Western and Eastern traditions since the end of the Ice Age
Figure 3.3. Keeping score: Eastern and Western social development since 14,000 BCE
After all this buildup, what do we see? Frankly, not much, unless your eyesight is a lot better than mine. The Eastern and Western lines run so close together that it is hard even to distinguish them, and they barely budge off the bottom of the graph until 3000 BCE. Even then, not much seems to happen until just a few centuries ago, when both lines abruptly take an almost ninety-degree turn and shoot straight up.
But this rather disappointing-looking graph in fact tells us two very important things. First, Eastern and Western social development have not differed very much; at the scale we are looking at, it is hard to tell them apart through most of history. Second, something profound happened in the last few centuries, by far the fastest and greatest transformation in history.
To get more information, we need to look at the scores in a different way. The trouble with Figure 3.3 is that the upward swing of the Eastern and Western lines in the twentieth century was so dramatic that to have the scale on the vertical axis go high enough to include the scores in 2000 CE (906.38 for the West and 565.44 for the East) we have to compress the much lower scores in earlier periods to the point that they are barely visible to the naked eye. This problem afflicts all graphs that try to show patterns where growth is accelerating, multiplying what has gone before, rather than simply adding to it. Fortunately there is a convenient way to solve the problem.
Imagine that I want a cup of coffee but have no money. I borrow a dollar from the local version of Tony Soprano (imagine, too, that this story is set back in the days when a dollar still bought a cup of coffee). He is, of course, my friend, so he won’t charge me interest so long as I pay him back within a week. If I miss the deadline, though, my debt will double every seven days. Needless to say, I fail to show up when the payment is due, so now I owe him two
dollars. Fiscal prudence not being my strength, I let another week pass, so I owe four dollars; then another week. Now his marker is worth eight dollars. I skip town and conveniently forget our arrangement.
Figure 3.4 shows what happens to my debt. Just like Figure 3.3, for a long time there is nothing much to see. The line charting the interest becomes visible only around week 14—by which time I owe a breathtaking $8,192. On week 16, when my debt has spiraled to $32,768, the line finally pulls free from the bottom of the graph. By week 24, when the mobsters track me down, I owe $8,260,608. That was one expensive cup of coffee.
By this standard, of course, the growth of my debt in the first few weeks—from one, to two, to four, to eight dollars—was indeed trivial. But imagine that I had bumped into one of the loan shark’s foot soldiers a month or so after my fateful coffee, when my debt stood at sixteen dollars. Let us also say that I didn’t have sixteen dollars, but did give him a five. Concerned for my health, I make four more weekly payments of five dollars each, but then drop off the map again and stop paying. The black line in Figure 3.5 shows what happened when I paid nothing, while the gray one shows how my debt grows after those five five-dollar payments. My coffee still ends up costing more than $3 million, but that is less than half what I owed without the payments. They were crucially important—yet they are invisible in the graph. There is no way to tell from Figure 3.5 why the gray line ends up so much lower than the black.
Figure 3.4. The $8 million cup of coffee: compound interest plotted on a conventional graph. Even though the cost of a cup of coffee spirals from $1 to $8,192 across fourteen weeks, the race to financial disaster remains invisible on the graph until week 17.
Figure 3.6 tells the story of my ruin in a different way. Statisticians call Figures 3.4 and 3.5 linear-linear graphs, because the scales on each axis grow by linear increments; that is, each week that passes occupies the same amount of space along the horizontal axis, each dollar of debt the same space on the vertical axis. Figure 3.6, by contrast, is what statisticians call log-linear. Time is still parceled out along the horizontal scale in linear units, but the vertical scale records my debt logarithmically, meaning that the space between the bottom axis of the graph and the first point on the vertical axis covers my debt’s tenfold growth from one to ten dollars; in the space between the first and second points it again expands tenfold, from ten to a hundred dollars; then tenfold more, from a hundred to a thousand; and so on to ten million at the top.
Politicians and advertisers have turned misleading us with statistics into a fine art. Already a century and a half ago the British prime minister Benjamin Disraeli felt moved to remark, “There are three kinds of lies: lies, damned lies, and statistics,” and Figure 3.6 may strike you as proving his point. But all it really does is highlight a different aspect of my debt than Figures 3.4 and 3.5. A linear-linear scale does a good job of showing just how bad my debt is; a log-linear scale does a good job of showing how things got to be so bad. In Figure 3.6 the black line runs smooth and straight, showing that without any payments the size of my debt accelerates steadily, doubling every week. The gray line shows how after four weeks of doubling, my series of five-dollar payments slow down, but do not cancel out, my debt’s rate of growth. When I stop paying, the gray line once again rises parallel to the black one, since my debt is once again doubling every week, but does not end up at quite such a dizzying height.
Figure 3.5. A poor way to represent poor planning: the black line shows the same spiral of debt as Figure 3.4, while the gray line shows what happens after small payments against the debt in weeks 5 through 9. on this conventional (linear-linear) graph, these crucial payments are invisible.
Neither politicians nor statistics always lie; it is just that there is no such thing as a completely neutral way to present either policies or numbers. Every press statement and every graph emphasizes some aspects of reality and downplays others. Thus Figure 3.7, showing social development scores from 14,000 BCE through 2000 CE on a log-linear scale, produces a wildly different impression than the linear-linear version of the same scores in Figure 3.3. There is much more going on here than met the eye in Figure 3.3. The leap in social development in recent centuries is very real and remains clear; no amount of fancy statistical footwork will ever make it go away. But Figure 3.7 shows that it did not drop out of a clear blue sky, the way it seemed to do in Figure 3.3. By the time the lines start shooting upward (around 1700 CE in the West and 1800 in the East) the scores in both regions were already about ten times higher than they were at the left-hand side of the graph—a difference that was barely visible in Figure 3.3.
Figure 3.6. Straight roads to ruin: the spiral of debt on a log-linear scale. The black line shows the steady doubling of the debt if no payments are made, while the gray shows the impact of the small payments in weeks 5 through 9 before it goes back to doubling when the payments stop.
Figure 3.7 shows that explaining why the West rules will mean answering several questions at once. We will need to know why social development leaped so suddenly after 1800 CE to reach a level (somewhere close to 100 points) where states could project their power globally. Before development reached such heights, even the strongest societies on earth could dominate only their own region, but the new technologies and institutions of the nineteenth century allowed them to turn local domination into worldwide rule. We will also, of course, need to figure out why the West was the first part of the world to reach this threshold. But to answer either of these questions we will also have to understand why development had already increased so much over the previous fourteen thousand years.
Figure 3.7. The growth of social development, 14,000 BCE–2000 CE, plotted on a log-linear scale. This may be the most useful way to present the scores, highlighting the relative rates of growth in East and West and the importance of the thousands of years of changes before 1800 CE.
Nor is that the end of what Figure 3.7 reveals. It also shows that the Eastern and Western scores were not in fact indistinguishable until just a few hundred years ago: Western scores have been higher than Eastern scores for more than 90 percent of the time since 14,000 BCE. This seems to be a real problem for short-term accident theories. The West’s lead since 1800 CE is a reversion to the long-term norm, not some weird anomaly.
Figure 3.7 does not necessarily disprove short-term accident theories, but it does mean that a successful short-term theory will need to be more sophisticated, explaining the long-term pattern going back to the end of the Ice Age as well as events since 1700 CE. But the patterns also show that long-term lock-in theorists should not rejoice too soon. Figure 3.7 reveals clearly that Western social development scores have not always been higher than Eastern. After converging through much of the first millennium BCE, the lines cross in 541 CE and the East then remains ahead until 1773. (These implausibly precise dates of course depend on the unlikely assumption that the social development scores I have calculated are absolutely accurate; the most sensible way to put things may be to say that the Eastern score rose above the Western in the mid sixth century CE and the West regained the lead in the late eighteenth.) The facts that Eastern and Western scores converged in ancient times and that the East then led the world in social development for twelve hundred years do not disprove long-term lock-in theories, any more than the fact that the West has led for nearly the whole time since the end of the Ice Age disproves short-term accident theories; but again, they mean that a successful theory will need to be rather more sophisticated and to take account of a wider range of evidence than those offered so far.
Before leaving the graphs, there are a couple more patterns worth pointing out. They are visible in Figure 3.7, but Figure 3.8 makes them clearer. This is a conventional linear-linear graph but covers just the three and a half millennia from 1600 BCE through 1900 CE. Cutting off the enormous scores for 2000 CE lets us stretch the vertical axis enough that we can actually see the scores from earlier periods, while shortening the time span lets us stretch t
he horizontal axis so the changes through time are clearer too.
Two things particularly strike me about this graph. The first is the peak in Western scores in the first century CE, around forty-three points, followed by a slow decline after 100 CE. If we look a little farther to the right, we see an Eastern peak just over forty-two points in 1100 CE, at the height of the Song dynasty’s power in China, then a similar decline. A little farther still to the right, around 1700 CE, Eastern and Western scores both return to the low forties but this time instead of stalling they accelerate; a hundred years later the Western line goes through the roof as the industrial revolution begins.
Figure 3.8. Lines through time and space: social development across the three and a half millennia between 1600 BCE and 1900 CE, represented on a linear-linear plot. Line A shows a possible threshold around 43 points, which may have blocked the continuing development of the West’s Roman Empire in the first centuries CE and China’s Song dynasty around 1100 CE, before East and West alike broke through it around 1700 CE. Line B shows a possible connection between declining scores in both East and West in the first centuries CE, and line C shows another possible East-West connection starting around 1300 CE.
Was there some kind of “low-forties threshold” that defeated Rome and Song China? I mentioned in the introduction that, in his book The Great Divergence, Kenneth Pomeranz argued that East and West alike ran into an ecological bottleneck in the eighteenth century that should, by rights, have caused their social development to stagnate and decline. Yet they did not, the reason being, Pomeranz suggested, that the British—more through luck than judgment—combined the fruits of plundering the New World with the energy of fossil fuels, blowing away traditional ecological constraints. Could it be that the Romans and Song ran into similar bottlenecks when social development reached the low forties but failed to open them? If so, maybe the dominant pattern in the last two thousand years of history has been one of long-term waves, with great empires clawing their way up toward the low-forties ceiling then falling back, until something special happened in the eighteenth century.