Another idea on it

new media & society Copyright © 2000 SAGE Publications London, Thousand Oaks, CA and New Delhi Vol1(3):7–28 [1461–4448(199912)1:3;7–28;010895]

National borders on the world wide web ...................................................................................... ALEXANDER HALAVAIS University of Washington ............................................................................................................ Abstract The internet is often seen as a significant contributor to the globalization of culture and the economy. It is also seen as an inherently international medium, unimpeded by national borders and removed from the jurisdiction of the nation-state. This paper argues that although geographic borders may be removed from cyberspace, the social structures found in the ‘real’ world are inscribed in online networks. By surveying 4000 web sites, it is determined that the organization of the world wide web conforms to some degree to traditional national borders. Web sites are, in most cases more likely to link to another site hosted in the same country than to cross national borders. When they do cross national borders, they are more likely to lead to pages hosted in the United States than to pages anywhere else in the world.

Key words global communication • internet • transborder data flow • world wide web

Any recent discussion of globalization – economic or otherwise – is likely to make mention of the internet. The exponential growth of the internet in the United States, now being overtaken by growth in the rest of the world, has led many to question the relationship between a new global network and the future of sovereignty for the nation-state. McLuhan’s ‘global village’ is frequently evoked and politicians strive to be seen as deliverers of information highways and wired schools. At the same time, many worry that instantaneous and ubiquitous transmission of cultural material may lead to a homogeneous world culture at best, and an American world culture at worst. That the internet leads inexorably toward globalization is often a foregone conclusion.

When evidence of internet-based globalization is presented it is all too often anecdotal. In the pages that follow, I will argue that national borders have a measurable effect on the topography of the world wide web. Surveying a sample of web pages allows us to determine the geographic distribution of hyperlinks. I will argue that these data do not support claims that ‘cyberspace’ exists as an anarchic unvariegated universe, unimpeded by national borders. Rather, while national borders seem to be less intrusive on the web than they are in earlier networked media, the resilience of cultural structures is demonstrated in the organization of this new medium.

NATIONAL BORDERS AND COMMUNICATION The question of identifying national borders on the internet is complicated by the fact that there is no clear agreement as to what ‘national borders’ are. By introducing a provisional definition of what national borders entail, we may be better able to detect homologies between these borders and the topography of the internet. A traditional view of the national border suggests itself as a starting point. In this conception, a national border is an imaginary boundary tied strictly to geographical territory in which a state’s sovereignty may be exercised (Goodwin, 1974: 100). While such a definition is well suited to discussions between national governments, determining the sovereignty of nations by their territorial borders does not account for two vital ingredients of a nation: its people and their culture. People have always made connections across national borders, but improvements in communication and transportation technologies have made such connections far easier as the 20th century comes to a close. A political or legal sense of national borders ignores, by and large, these vital connections.

A second view of national borders is that they represent a gap in the totality of relationships between individuals.1 Such an approach focuses more heavily on networks of association rather than more static or institutional notions of the state. John Burton (1972), for example, suggests that international political theory must change to recognize increases in transnational practices, especially among non-governmental groups. He argues that the traditional ‘billiard-ball’ model of international relations, in which countries of differing sizes act and react only in terms of a cohesive foreign policy, is lacking. He proposes, instead, the ‘cobweb’ model of international relations, in which layers of interaction define a world of highly complex interdependency. Such a view holds that while physical terrain may play a (decreasing) role in determining nationality, it is the imagined community, determined more by the propinquity of ideas than by the exigencies of physical distance, that defines nations and their borders. Such a network or systems approach is very much in vogue among global communications scholars, due in no small part to the recent increased potential for distanced communication brought about by networking technology in general and the internet in particular. The idea that communications media have a substantial effect on the emergence of largescale

political formations is most often associated with Harold Innis (1972).

However, it is Karl Deutsch’s work in drawing relationships between measured communication flows and national boundaries that lays the foundation for the study presented here. Deutsch argues that although a number of factors contribute to establishing a nation, many of these are clearly measurable in the form of patterns of communication. In his words, we are able to draw from ‘the observable ability of certain groups of men and women to share with each other a wide range of whatever might be in their minds, and their observable inability to share these things nearly as widely with outsiders’ (Deutsch, 1953: 65). By examining the degree to which countries communicate internally and externally, and the character of that communication, we should gain some understanding of who becomes the ‘outsider’; in other words, which ‘peoples are marked off from each other by communicative barriers, by “marked gaps” in the efficiency of communication’ (1953: 74). Measuring communicative flows on a large scale should provide some clues as to where these borders are being drawn in the collective imagination (Deutsch, 1956; also Janelle, 1991).

As one Singaporean minister has suggested, the modern nation is like a cell in a larger organism: porous in some respects, walled off in others, part of larger structures, containing sub-structures, ultimately in control of its own actions (Yeo, 1995: 23). By studying communication flows, we may gain some understanding of the structure of imagined nations and some idea of where their borders lie. These borders may not be as arbitrarily exacting as those found on a world map, but by measuring the relationships between individuals and groups, we may arrive at a more dynamic and realistic measure of nations and their borders.

[ . . . ]

[ . . . ]

ANALYSIS An analysis of the data provided by this survey leads to findings in two areas. First, while the world wide web is a very international medium, the number of hyperlinks that cross international borders are significantly less than those that link to sites within the home country. Second, links are far more likely to be directed toward the United States than toward any other country, though this appears to be due in large part to the imbalance in the number of sites hosted in each country.

If the destination of hyperlinks are aggregated for the 12 countries best represented by the sample, it becomes clear that domestic links are far more common than international links (Figure 1). With the exception of those hosted in Canada, sites were more likely to link to another site within the same country than to cross national borders (Table 2). This is not at all what we would expect if the world wide web were, indeed, an undifferentiated network. Rather, we would expect a fairly even distribution of linkages across the web. The tendency to link to domestic sites is particularly significant, given that outside of domain names in the URL there is very little to indicate or to restrict the user from crossing national boundaries while surfing the web.

Of course, web sites are not evenly distributed among countries, so we cannot assume that links will be. The initial diffusion of the web was certainly heaviest in the United States, and the relative maturity of the web in the US means that a majority of pages are hosted by American servers. Using the distribution of hosts as a guide, we would expect, for example, about 70 percent of all links on the web to lead to the US and about 1 percent of all links to lead to Japan. In fact, as shown in Table 3, the United States receives a lower percentage of links than we would expect from sites around the world (except from those sites located within the US). Table 3 (and Figure 2) show how the percentage of linkages between countries differs from the distribution of host machines. While the percentage of linkages to the United States is quite high in absolute terms, it is unremarkable when the distribution of the web is taken into account.

There remains a bias toward domestically produced content in the US, but this bias is fairly small when compared to the relatively inwardly linking web in France and Japan. We are left with an ambiguous picture. A majority of web content is created in the United States and this content is linked to nearly as frequently as material produced indigenously around the world. However, a very large part of the bias toward the United States seems to be as a result of the distribution of content at this stage of the internet. If this is, as this survey shows, an accurate depiction of today’s web, we might expect to see this disparity diminish as more of the world begins to use the internet. On the other hand, the population, economic power, and technological position of the United States makes it unlikely that countries like Canada or individual nations in Europe will be able to challenge the centrality of America on the web in the very near future, especially in terms of networking infrastructure (Evagora, 1997; OECD, 1998).

The structure of content on the web may prove to be far more dynamic, especially if the rapid changes in demographics of web users are any indication of the volatility of this new medium (Bloomberg, 1998). As has already been noted by a number of technically as well as socially minded internet researchers, the measurement and mapping of the internet is of vital

importance. In this vein, Tim Berners-Lee has called for ‘parameters of measurement of restlessness and stability analogous to hormone levels or body temperature of the human organism’ as an indication of how this structure is changing (Berners-Lee, 1997). Such measures should provide not only information about ‘cyberspace’ but about how networking affects real space.

Strikingly clear in both Tables 2 and 3 is the tendency of sites to link domestically rather than internationally. When compared to two traditional networked media – the telephone and postal systems – the web appears to be much more internationalized. Yet the degree to which there are gaps at national borders is remarkable; even more so when it is noted that unlike the postal and telephone systems, the web provides no differential pricing for domestic and international linkages. As cost is reduced, we would expect the network to become increasingly interconnected. In the case of the web, however, it is clear that there are other, non-economic barriers to distanced networking. While a hyperlink from Paris to Nice may cost the same as one from Paris to Tokyo, the former is far more likely. The exact location of the most significant gaps in communication cannot be determined by the approach taken here, but it is clear that these gaps are to a degree correlated to national borders. When other borders are removed, social homophily guides the selection of necessarily scarce hyperlinks (Van Alstyne and Brynjolfsson, 1997).

As seen in Table 4, the degree to which these gaps are present differs depending on the subject matter found on the site. The most internationalized pages tend to be those related to the international scholarly community, while the least tend to be pages related to governmental bodies. News, sports, and (strangely enough) travel tend to be less oriented toward international hyperlinks than web sites centered on other topics. We might approach this information in two ways. We might begin by considering the web as an indicator of the global social environment. For example, scientific and scholarly communities have long been international in nature, as have certain political movements. As the topical index suggests, these groups have quickly migrated to the new medium. Other groups have only recently seen an increase in the need for transnational communication. For example, the elimination of many economic impediments has driven even small businesses into the international market.

However, the web has also provided excess capacity for transnational practices. While some Americans may be ‘bowling alone’, many others are taking up hobbies and interests – from anime to macram´e – for which they find support from outside of their physical communities. Many businesses approach the web as a cheap source of advertising or another venue for sales and ‘stumble into’ the international aspect of the medium. As users come to depend on the web, they enter into negotiation with its conventions, adopting those they like and adapting to those they do not. Because of the reciprocal relationship between public conceptions of the web and its actual structure, the future of that structure remains difficult to predict.

The preponderance of the sites in this sample contain pages in English, as shown in Table 5. The overwhelming presence of English on the web is a cause for concern, given the potential of language as perhaps the most powerful of ways to establish borders in this new medium (Castells, 1997: 52). As noted above, the sample may be slightly skewed toward sites in countries in which the English language is dominant, so some caution should be taken before inferring too much from the large number of English-language sites. However, the percentage of English-language sites that link domestically is instructive. We might expect English-language sites to be more likely to link internationally if English is indeed the new global lingua franca. One possible explanation for this is that commercial sites – which are far more likely to be ‘dead-end’ sites, without hyperlinks outside the site (Terveen and Hill, 1998) – appear more often in English. Less surprising, perhaps, is the fairly large amount of international linkage associated with multilingual sites.

CONCLUSIONS The findings presented here lead to an immediate set of conclusions and a more forward looking set of suggestions. The first of these is related to a reading of the internet at a particular point in time. The novelty of the internet forces discourse about it to the extremes, and hyperbole abounds when questions of national borders, sovereignty and the internet are addressed. The survey undertaken here demonstrates clearly that, as with earlier ‘new media’, this technology is both ‘so new that people can’t even imagine it’ and ‘never so new as people imagine’ (Nord, 1986). On one hand, while the internet incorporates little in the way of technological, regulatory or economic impediments to transnational interconnections, the web demonstrates that national cultures continue to exert a substantial influence on how these connections are made. While national borders may be eroded, they certainly remain significant.

On the other hand, when compared to other media, the internet is considerably more internationalized. If you examine postal flows, none of the 12 countries considered here receives more than 5 percent of its total letters from abroad (UPU, 1997). The United States, the most insular (in absolute terms) of the countries considered here, has over 9 percent of its links fetching information from abroad, while other countries have much higher rates of international content. This presents a novel opportunity for people to be exposed to information and ideas from outside their own national cultures (to the extent that such can be said to exist). I would suspect that email and other more ‘personal’ uses of the internet would be more geographically localized and this presents an interesting area of inquiry. The existence of national borders, though perhaps not in the more traditional sense, has important policy impacts. The chief argument against national regulation of the internet is that it is inherently global. This survey indicates that for the web, national borders are neither absent nor absolute. Certainly there are significant challenges to designing regulation for a social environment that is less reliant on geography (Lenk, 1997). However, ridiculing policy-makers who claim that the internet can be segmented or controlled ignores the crucial impact of social structure on the structure of the medium.

In her dissent to the Supreme Court’s striking down of the

Communications Decency Act, Justice Sandra Day O’Connor argued that cyberspace could indeed be segmented if the desire to do so was made clear in social policy (Reno v. ACLU, 1996). In making such an argument, she explicitly drew on ideas presented by Lawrence Lessig, a law professor presently at Harvard University. Lessig (1996) argues that there are a number of elements that lead to structures in the networked environment. While some of these are technological, most of them emerge as social (and often commercial) constructs. Legal borders – national and otherwise – emerge as social conventions. As such, they need not rely expressly upon geography. As the internet becomes more socialized, law will develop that takes into account the new borders of cyberspace (Johnson and Post, 1997). The present position of the United States as central to the web may also be a cause for alarm. The future of this distribution is in no way certain – it may go the way of earlier mass media and the US may maintain a central position on the internet, as Herbert Schiller and others have argued (Gillespie and Robins, 1989; Schiller, 1992, 1995). However, the widely noted speed of diffusion of the internet outside of the United States, and the relatively open linking of US sites to sites abroad is likely to present a significant challenge to the centrality of the United States on the web (Maherzi, 1997: 46–7).

More important than these conclusions, which given the ephemeral nature of the internet remain necessarily of the moment, are the conceptual underpinnings of this study. An attempt to describe the social impacts of the internet must include some indication of the structure of this medium. A description that suggests the internet is an undifferentiated space outside of ‘real’ space – as many popular and academic accounts do – must be approached with some skepticism. We must recognize that ‘social borders have their own cartographies’ and go about mapping these structures (Harvey, 1996: 282). This is in no way a new concern – Georg Simmel noted that modernity has provided any number of examples of ‘a group whose cohesion depended upon geographic and physiological factors, terminus a quo, [being] entirely replaced by a group whose cohesion was based on purpose, on factual considerations, or, if one will, on individual interests’ (Simmel, 1955: 128). The internet provides a very promising way to observe how these borders and groups evolve. While the future of internetworking will most certainly surprise us, the need to investigate the social and informational structure of the medium will continue to remain among the most important tasks of the researcher.