Demic Atlas

GeoCurrents Break and Renovation

Dear Readers,

GeoCurrents will be taking an end-of-the-year break; regular postings will resume in the second week of January. During the break, plans will be made for renovating and expanding the site. In January, blog postings will increase from the current two or three per week to four or five per week. New features will also be added, focused on providing brief coverage of geographically significant news stories from around the world that are neglected by the mainstream press. According to current plans, the fully refurbished site should come on-line by April 2012.

The GeoCurrents team will also expand. In addition to myself and Kevin Morton, Asya Pereltsvaig, frequent commentator and author of the Languages of the WorldWorld Map of Per Capita GDP, with Large Countries Divided weblog, will be joining the project. As the site grows, other contributors may join the team as well.

During the end-of-the-year break, periodic “housekeeping” posts may appear on the site. Today, for example, I have posted an additional map from the Demic Atlas project that was carried out this summer by myself, Jake Coolidge, and Anne Fredell. This map shows nominal per capita GDP (2009, in US$) for all countries with fewer than 100 million inhabitants and for all first-order subdivisions of countries with more than 100 million inhabitants. Some patterns that are invisible in both the standard state-based framework and the demic framework are apparent in this map. Russia, in particular, takes on a distinctive appearance. Here the lightly populated but oil- and gas-rich districts of western Siberia fall into the highest category, whereas the North Caucasus, as well as parts of European Russia and eastern Siberia, fall into the second-lowest category. In Indonesia, the resource-rich province of East Kalimantan stands out, contrasting sharply with the much poorer southeastern reaches of the country. In Nigeria, a distinct north/south division is visible.



A Global North/South Division in the Demic Framework?

As has been argued previously on GeoCurrents, the commonplace notion that the world is starkly divided between a prosperous and powerful “global north” and an impoverished and underdeveloped “global south” (with Australia and New Zealand forming southern outposts of the north) receives little support from world maps of socio-economic development. As can be seen in the state-based map of per capita GDP (PPP) posted below, the so-called global south is a diverse zone, containing relatively wealthy as well as extremely poor countries. As a result, the south fails to cohere as a region; in terms of virtually all development indicators, a well-off southern country such Chile is much closer to such northern regions as southeastern Europe than it is to central Africa. Most socio-economic maps constructed in the demic framework are even less supportive of the notion of a fundamental latitudinal divide. As the demic map of per capita GDP posted below indicates, the real zone of poverty is focused in tropical Africa and the northern part of southern Asia, rather than in a hypothetical globe-spanning south.

Such maps, however, do not tell the whole story. Statistical information can be mapped in many different ways depending on how the data are categorized. By playing with different modes of division, one can generate a diverse set of maps, some of which might uncover patterns that otherwise remain hidden. As it turns out, the database used in the Demic Atlas can indeed produce maps replicating the north-south global divide. Such a pattern is not pronounced, as it takes careful manipulation of the underlying information to expose it. Yet its presence is notable, and hence deserves consideration.

Most of the maps posted in GeoCurrents last week were based on the division of the data into quantiles. Quantiles break up a dataset to yield subsets of equal size; as the same number of units is slotted into each category on quantile maps, relatively well-balanced depictions generally result. In the Demic Atlas, seven such divisions were employed for most maps, as experimental evidence suggests that most people cannot readily differentiate more than seven values in a color spectrum. Although the seven-fold quantile design seems to be the most useful general vehicle for mapping global development, it is not without its problems. Most significantly, it under-represents the discrepancy between the poor and wealthy parts of the world. In absolute terms, the per capita GDP figures of areas placed in the middle categories on the quantile maps are closer to those of the countries on the low end of the spectrum than they are to those of the countries on the high end.

One alternative method of dividing data is that based on “equal intervals.” In this system, the value ranges of each category are of equal numerical scope, set at the same regular intervals regardless of how many units fall into each category. If, for example, the lowest figure in a five-category dataset is 1 and the highest is 100, the first category will cover the numerical range between 1 and 20, the second between 21 and 40, and so on. If most elements of the dataset are clustered in one part of the range, the resulting map may appear quite unbalanced, with several of its categories forming null sets, with no members.

Due to such data clumping, depicting global per capita GDP in equal intervals in a state-based framework yields an uninstructive map. As can be seen in the image to the left, almost the entire world appears impoverished in a seven-fold equal-interval scheme, with most countries falling into the lowest category. Although unsuitable for general purposes, this map does illustrate one important aspect of the global distribution of wealth among polities: a small handful of small countries and dependencies are vastly wealthier than the rest, at least in terms of per capita GDP. Qatar in particular, with its small population and vast reserves of oil and natural gas, towers above most of the world’s richest countries, with a per capita GDP that the CIA regards as almost four times greater than that of the United States. As the detailed map of Europe and the greater Middle East shows, Qatar, Luxembourg, and Lichtenstein monopolize the high-end categories in such a portrayal.

As the demic framework eliminates small countries by merging them with their neighbors, it allows much more effective equal-interval mapping. In the demic equal-interval depiction of per capita GDP (PPP) placed at the top of the post, large expanses of land fall into the upper categories, but even larger areas end up in the lowest grouping. The middle categories, in contrast, are sparsely occupied.

As a quick comparison with the small inset map indicates, the resulting categorization scheme conforms well with the conventional north/south global division. Whereas almost all regions to the south of the heavy black line fall into the bottom two categories (the only exceptions being Regions 19 [Arabian Peninsula] and 47 [Jiangsu and Shanghai]), none of the regions to the north of the line do. The line itself, moreover, falls almost exactly along the north/south division as it is conventionally mapped. The only differences here are the demic map’s placement of southeastern Europe and Israel in the global south and of Kazakhstan and Mongolia in the global north.

Although a north/south global split can be derived from the demic database, it would be unwise to read more into this division than is warranted. As explained above, the global south remains invisible in almost all other methods of mapping the data, whether in the state-based or the demic framework. More important, most of the north is currently exhibiting slow rates of economic growth, and has been for several years, whereas many southern countries, particularly China and India, are moving rapidly forward. If current trends continue, the north/south divide will vanish even on the equal-interval demic map within the coming decade.

Concluding Posts on the Demic Atlas

Dear Readers,

Map of Island Assignments in the Demic FrameworkThe Demic Atlas project will conclude at the end of this week; next week’s posts will return to the standard GeoCurrents model, examining local issues of geographical significance. Today’s map merely shows which island groups are associated with which regions in the demic framework. As the map is self-explanatory, no further comment is provided. Tomorrow I will respond to the comments that have accumulated over the past few days—my apologies for not having done so already.

As originally envisaged, the Demic Atlas would have contained a number of maps showing the spatial patterning of a wide variety of development indicators. Depictions of literacy, average age of schooling, fertility, mortality, sex ratio and so on would have been included. As we have discovered, however, such information is simply not available in comparable form for most of the sub-national units used to construct the demic framework. Even for the indicators that were mapped (GDP and HDI), problems of data comparability compromised the project in several ways. Yet such difficulties were instructive in their own right, reinforcing the central thesis underwriting the entire project: sovereign states (and their dependencies) so dominate the realm of global data collection and collation that they systematically distort our view of the world. One can gather relatively solid information at the provincial level for most large countries, but only if one does so in isolation from the rest of the world, thwarting the comparison of units of like size. Although the states of India and the provinces and other first-order divisions of China are country-sized units, they cannot readily be contrasted with each other in a single framework. Outside of Europe, choropleth maps integrating data for a number of countries at the sub-state level are, to say the least, challenging to produce.

Further mapping of developmental indicators within the demic framework will therefore not be forthcoming. The remaining atlas posts will take up another issue, that of the variable methods one can use to construct maps out of a single set of data. Quantitative information can be broken down in many different ways, producing divergent images when translated into cartographic form. The maps posted last week, for example, divided GDP and HDI figures at regular intervals to yield subsets of equal size. Seven such quantiles were used, but the data can just as easily be broken down into more or fewer categories. The information set can also be divided in completely different ways; using “natural breaks,” for example, allows one to keep areas that deviate slightly from each other within the same categories. The remaining posts in this series will therefore focus on alternative methods of mapping the information in the Demic Atlas database. This endeavor will reveal otherwise invisible spatial patterns while also showing how the same set of information can yield strikingly different maps depending on how it is arrayed.  Such maps will be constructed in both the demic and the state-based frameworks.

Demic Atlas Preface, Part III

As the past several GeoCurrents posts have explained, sovereign states make poor units of socio-economic comparison due to their vast size disparities. But issues of scale are not the only reasons for considering an alternative scheme of division. In the standard model of global affairs, countries are the all-purpose and essential units of human organization. According to this view, independent states are neatly demarcated geo-bodies administered over their entire expanses by globally recognized, fully autonomous governments, their inhabitants bound together by common sentiments of national solidarity. Yet the nation-state ideal is seldom fully realized. The world is replete with nationless states, stateless nations, contested nationalities, vacuums of sovereignty, and so on. As Jim Wilson perceptively pointed out in a GeoCurrents comment, such geopolitical “anomalies” are too common to be considered anomalous. The political structure of the world, in short, is far too complex for the reigning model. As a result, the global representation found on country-based political maps is simplistic at best and misleading at worst. On a world map, Somalia has the full appearance of a sovereign state; in actuality it has virtually none of the substance.

Over-reliance on a flawed world model results in more than intellectual mischief. In the early years of this century, U.S. military and political planners had little doubt that Afghanistan and Iraq could be quickly and cheaply stabilized and democratized, as it was assumed that they were coherent nation-states, their people tied together by bonds of common affiliation, and hence willing to work together to achieve national aspirations. A decade on, trillions of dollars and many thousands of lives have not proven adequate in either case. As the United States lurches through its second economic crisis in three years, its financial resources stretched near the breaking point, those billions upon billions of dollars that have flowed into nation-building endeavors in Iraq and especially Afghanistan increasingly seem like poor investments. Had we been less beholden to a normative geopolitical model, perhaps such inordinately optimistic policies would not have been pursued.

The goal of the Demic Atlas is to denaturalize the state-based picture by viewing the world through an alternative lens. It is not to argue that states are passé or unimportant in any way, much less that the standard political map should be abandoned. The contention is rather that countries should not be the only units used for depicting the socio-economic differentiation of the human community. Different modes of division can show different patterns and, we believe, yield new insights.

The classification scheme employed in the Demic Atlas will seem odd to most viewers, and shocking to some. The standard world map and model are so ubiquitous, so taken-for-granted, that any alternative is bound to appear perverse. Many if not most educated people have also been so schooled in nationalism that their own countries at least seem inviolable, forming natural units of not just political power but also of social and economic organization. Quick previews of the demic world map—which will be posted here next Tuesday or Wednesday—provoked quick objections, as viewers found it wrong to see their own countries sundered and then re-aggregated with pieces of other states. But this is precisely the purpose of the exercise: to unsettle conventional notions of global geographical organization by challenging the essentialism of the sovereign state.

Modified Map of Europe  from 1751As instinctive as it has become for us to divide the world along geopolitical lines, it is not a particularly long-standing maneuver. Old maps reveal that before the 1800s, European cartographers typically deployed a hybrid system of terrestrial division. In the early modern period (1500-1800), they typically began by splitting the world into continents, and then carving each of the resulting landmasses into a handful of major divisions. Some of these sub-continental entities were geopolitically delimited kingdoms, empires and the like, such as France, Russia, Persia, and China. Others were former sovereign powers that had long lost that standing, such as Hungary and the German (or Holy Roman) Empire, while others were (at the time) mere regions with no political coherence (Italy, Arabia, and India, for example).* Such units were by no means of equal size, as those of Asia dwarfed those of Europe, reflecting the Eurocentrism of European cartographers. But within each continental frame, all the constituent “countries” were roughly comparable. It would have struck an 18th century cartographer as absurd to elevate the pocket states of northern Italy, let alone those of western Germany, to the same level as France or Spain on their basic maps. Early modern cartographers did occasionally depict the small polities of Italy and the Holy Roman Empire, but only when they were explicitly illustrating the geopolitical order. Maps devised for general purposes relied instead on a hybrid divisional scheme.

A somewhat similar situation obtained in early modern Japan. Before the Meiji Restoration of 1868, Japan was a semi-unified state; the Tokugawa Shogunate exercised hegemony over the main islands, but over 200 feudal lords maintained autonomy within their own domains, many of which were spatially dispersed. Tokugawa cartographers by and large ignored the complexly fractured political order that resulted, and instead mapped the archipelago in accordance with the provinces of classical Japan—units that had no political significance at the time. The continued use of a system of defunct subdivisions in basic maps has struck many observers as a quirky anachronism, but as Kären Wigen has shown in her recent book, A Malleable Map, the strategy had its own compelling logic. The provinces of Tokugawa Japan were observational rather than administrative units, providing spatial containers for place-specific information. Provinces served this purpose well because they were deeply rooted in historical memory, were no longer politically charged, and were spatially stable. Also, unlike the domainal territories of Tokugawa Japan, they were relatively compact and of roughly similar size.

Like the provinces of Tokugawa Japan, the spatial divisions in the Demic Atlas are designed for observational purposes: to serve as politically neutral containers for marshalling socio-economic data. Unlike the old Japanese provinces, of course, they are nakedly artificial constructs without historical precedent. Whether or not such manufactured units prove useful is for readers to decide.

*Intriguingly, mapmakers of the time sometimes claimed to base their divisions on geopolitical criteria, yet in practice they did not exactly do so. Consider Thomas Kitchin’s 1787 map, “Europe Divided into its Empires, Kingdoms, States, Republics, Etc.,” published in London by Robert Sayer. As actually mapped, the “principal parts” of Europe included such non-sovereign entities (at the time) as Italy, Germany, Ireland, and Hungary. The accompanying text notes that “Germany is full of sovereign princes and counts …, every one of which is more free and absolute than several crowned heads.” A 1742 version of Guillaume Delisle’s Atlas Nouveau (published in Amsterdam) likewise contains a map purporting to show Asia “accurately divided into Empires, Kingdoms, States, and Peoples,” yet it depicts all of mainland Southeast Asia as one entity, and all of central and northern Asia as another (“Tartary”). These kinds of maneuvers were typical of the time.

Enlightenment-era views on the geopolitical division of space are perhaps best represented in the Encyclopédie Méthodique par Ordre des Matières (“Methodical Encyclopedia by Order of Subject Matter”), a 200-plus-volume reference work published by Charles-Joseph Panckoucke, designed to follow the more famous encyclopedia of Denis Diderot. Here the Europe entry (vol. 2, page 574) details the division of the landmass into three empires (Russia, Germany, and Turkey-in-Europe), twelve kingdoms, one great ecclesiastical realm (the papal state), one archduchy, one grand duchy, four great republics, and four less powerful republics. Not all of Europe, however, was classifiable in this scheme, with northern Italy presenting particular challenges. In another entry, the encyclopedia describes Italy as “a great country of Europe,” noting that it has too many political divisions to report (vol. 1, page 94). The work also notes that the kingdoms and duchies of Europe are not all ruled by their own sovereigns. Although Hungary is listed as a kingdom (joined with that of Bohemia), the author allows that it is currently “under the House of Austria” (vol. 94, page 2).