Internet Deployment Worldwide: The New Superhighway Follows the Old Wires, Rails, and Roads
Eric ARNUM <firstname.lastname@example.org>
Sergio CONTI <email@example.com>
This paper details the extent to which some of the most popular inventions of the 19th and 20th centuries have been diffused into the economies and populations of 100 countries. It includes a statistical analysis of per capita diffusion rates for Internet hosts, Web pages, telephones, televisions, electrical production, and paved roads and railways, comparing the world's new superhighway with the old infrastructures it replaces and/or complements.
Keywords: Internet host, Network Wizards, Web pages, AltaVista, Internet domain, roads, railways, electrical power, telephones, televisions, diffusion rates, GDP, national statistics, transportation, energy, telecommunications, population density, per capita wealth, scatter plot, generic top-level domain, gTLD.
In a study of 100 countries that represent 99.65 percent of the hosts on the Internet in January 1998, it was revealed that in general, the nations with the highest diffusion rates for transportation, energy, and telecommunications also are among the leaders in Internet usage.
Countries in this category include Finland, the United States, Iceland, Norway, Sweden, Australia, Switzerland, Canada, Denmark, and the Netherlands. They are leaders in the diffusion of the old superhighways (which switch cars, trains, phone calls, electrical current, etc.) and they also are the leaders in the diffusion of the new superhighways (which connect Internet hosts and Web sites).
This sort of comparison proves what many Internet researchers believed intuitively, that countries which deployed high per capita amounts of roads, railways, electrical power, phones, and TV broadcasting in earlier generations are also leaders in the current generation's deployment of Internet hosts and Web pages. This paper proves that whether the pundits call it the Infobahn or the Infostrada, the Internet really is the next superhighway, the next network grid. One network's diffusion facilitates the next: extensive transportation and communications infrastructures enabled commerce and created wealth, and brought with them mass markets for TVs, appliances, and phones. These indicators in turn produced a demand for the Internet, both within the household and among businesses.
What TV was to the second half of this century, what the telephone and the paved road were to the early 20th century, and what the railroad was to the 19th century, so too is the Internet to the current generation. It is another wire grid to lay across the continents that enables communications. And sure enough, the early adopter countries for those older networks also are leading the way onto the Internet today.
Still, there are huge areas of the world where there is not much Internet activity . . . yet. The nations of South America, the Middle East, Africa, and the bulk of the Asia/Pacific Region (with notable exceptions such as Japan, Australia, Singapore, Hong Kong, Taiwan, and New Zealand) are behind the curve in terms of deployment of paved roads, wired infrastructure, and -- no surprise -- Internet usage.
In this paper, countries are compared on universal statistical measures such as size, population, and Gross Domestic Product (GDP), as well as on the reported number of kilometers of roads and railways. Statistics also are presented for each country's annual electrical output and the number of phones and televisions in use. In all cases, the latest available data was used, primarily from 1995 to 1997, although some of the Internet host data is from 1998. Sources included the Grolier's Encyclopedia, available online from CompuServe, and the CIA World Factbook for 1997, available on the Web.
Statistics for the number of Internet domains and Internet hosts in existence in January 1997, July 1997, and January 1998 were compiled by researchers at Network Wizards. The researchers changed their methodology for the January 1998 count in order to allow for a more complete counting of the world's Internet hosts. Because of that change, the three most recent sets of Internet domain and host statistics were averaged in the ratio computations described below.
Statistics also were compiled for the number of Web pages created month by month between December 1993 and January 1997. The number of Web pages created worldwide (which does not account for those that have subsequently gone "out of print") stood at 84 million in January 1997. The page count rose to at least 124 million by December 1997 and at least 175 million pages by April 1998. However, because up-to-the-present page counts were completed for only 40 of the 100 countries by the time of this publication, only the worldwide count as of January 1997 was used in the ratio computations. The source for the Web page counts was the AltaVista  search engine.
These statistics were gathered, compiled, and compared to create five ratios--for Density, Wealth, Paved, Wired, and Internet--for each of the 100 countries. Worldwide medians for each of these five ratios were then created, considering data from public sources for worldwide totals.
The 100 countries were selected from the possible 244 using a rough methodology that examined the number of hosts detected in January 1998, the population of the country in 1997, and/or the country's estimated GDP in 1995. All countries with more than 500 Internet hosts in January 1998 were automatically included. All countries with at least 25 million inhabitants were automatically included. And all countries with at least $100 billion GDP were automatically included.
The worldwide totals for communications networks, population, landmass, and other parameters used in this study are as follows:
Sources: ITU Telecommunication Indicators , 1996 CIA World Factbook , DEC AltaVista search engine , Network Wizards Internet Domain Survey, January 1998  & 1997 Grolier Multimedia Encyclopedia .
As more current data becomes available, these ratios can continue to be updated. In fact, it is hoped that particularly for the Internet ratio, annual updates will allow for further examinations of the changing rate of diffusion of Internet and Web infrastructure over time within each country. The Internet is growing at such an incredible rate (Web pages over 100 percent annually, Internet hosts around 40 percent annually) that it significantly impacts the ratios computed in this study.
The resulting list of 100 countries selected for this study represents more than 99 percent of the world's Internet hosts, 92 percent of its population, and 96 percent of its GDP. It represents 93 percent of its roads and 97 percent of its rails, 89 percent of its phones and TVs, and 98 percent of its electrical production. So although it is a list of barely four-tenths of the world's assigned ISO country codes, it is nevertheless one that represents a high percentage of its communications and transportation infrastructure.
Five populous countries with little or no recorded Internet activity were included in the group of 100 because of their size and their use of other paved and wired networks: Bangladesh, Sudan, Myanmar, Zaire, and Afghanistan. The only countries of appreciable size (but fewer than 25 million inhabitants) that were omitted from the study were Ghana, Iraq, Ivory Coast, Nepal, North Korea, Syria, Uganda, and Yemen. These eight countries had but 684 Internet hosts among them in January 1998.
The five ratios rank the 100 countries from highest to lowest diffusion rates, using the same formula for each. For example, the Paved Ratio is the sum of kilometers of paved road and railways, divided by the square kilometers of area within the country's borders. In that measure, Singapore ranked highest and Sudan ranked lowest.
Here are the formulas for the five ratios:
Hosts + Domains + Web Pages Internet Ratio = ------------------------------------- PopulationElectric (kWh) + Phone Lines + TVs Wired Ratio = -------------------------------------- PopulationRailway (km) + Road (km) Paved Ratio = -------------------------------------- Area (sq km)Gross Domestic Product Wealth Ratio = --------------------------------- PopulationPopulation Density Ratio = --------------------------------- Area (sq km)
In all, there were 14 data points per country, for a total of 1,400 different statistics. An additional 430 statistics were compiled for the uncounted countries, primarily to ascertain if there was significant economic or Internet activity within their borders. As mentioned, none of the excluded countries had more than 500 hosts, 25 million people, or $100 billion in GDP.
Though there is no longer a Soviet Union, the SU domain is still frequently used on the Internet. Likewise, though Hong Kong has been reunified with China, they continue to have separate identities within the Internet domain hierarchy. Just to see what would happen, combined ratios were computed for HK and CN, and for RU and SU. Those ratios were found to be essentially the same as for China and Russia alone. In the final list, HK, CN, and RU were included, but SU was not.
For the sake of simplicity, the six major "generic Top Level Domains" (gTLDs) were counted entirely in the USA total, alongside the US domain. Around 61 percent of the world's Web pages and 70 percent of the world's Internet hosts are in one of the gTLDs. However, researchers have found that somewhere around 10 to 15 percent of the hosts, domains, and Web pages within the com and net domains are in fact not physically located within the USA. But if such a deflator were applied to the Internet Ratio for the USA, the USA would move only two slots down the list, from its current second place into fourth place, between Norway and Sweden.
Even with all the gTLDs in its column, the USA is not the country with the most per capita Internet activity. That honor goes to Finland. This is by far the most surprising calculation of the study: Finland has the highest diffusion of Internet technology in the world. Once an allowance for the international aspect of the gTLDs is considered, then Finland, Iceland, and Norway each would have a higher rate of Internet activity than the USA.
Given the formulas listed above, and the data to which they were applied, the ratios for the whole world were as follows:
These are the worldwide medians. Each worldwide median was then set to equal 1.0 by multiplying the results of the calculations described above it by its mathematical inverse. So, for instance, 5,389, the worldwide median for per capita GDP, was multiplied by 0.00017127. Each of the 100 country ratios was then multiplied by this same constant, resulting in five charts of ratios clustered around 1.0. In all cases, those with ratios above 1.0 were above the worldwide median. Those below 1.0 were below the worldwide median. The magnitude they were above or below 1.0 inferred the magnitude of their deviation from the worldwide median.
The computation of the Internet Ratio, because it is based on such highly volatile statistics, is highly dependent upon the moment in time at which the statistics are gathered. Unlike the installed base of other network infrastructures, the worldwide Internet is growing at a very rapid pace. Within some countries, the size of the Internet has doubled within a year's time.
As mentioned previously, the World Wide Web page count stood at 84 million in January 1997, but had increased to 124 million pages by the end of 1997. Because Web growth has accelerated to an even faster pace in the early months of 1998, the growth rate of the Web is back over 100 percent a year worldwide.
According to Network Wizards, the Internet host count in January 1997 was just over 21.8 million, but by January 1998 it was nearly 29.7 million hosts. So although growth worldwide averages 40 percent, within countries such as Taiwan, New Zealand, and Singapore, it continues to exceed 100 percent a year.
Keeping in mind how fast the data is changing, here are the 100 computations of the Internet Ratio, along with the corresponding Web page counts for January 1997 and the Internet host counts for January 1998:
There was a decided east/west split noticeable within Europe, with the east tending to be below the median and the west tending to be above the median. Only three exceptions were noted to this trend. Malta was the only western European country below the median for Internet activity. Estonia and Slovenia were the only eastern European countries above the median for Internet activity.
There also was a noticeable north/south split within Europe. As a region, the Scandinavian countries were towards the top of the list, claiming four of the top five slots. The Mediterranean countries, meanwhile, while they generally were above the median, were usually clustered farther down the list, below their neighbors to the north.
A closer examination of growth trends (data available upon request) shows that while the northern countries have the largest installed bases, the Mediterranean countries have the highest annual growth rates. If northwestern Europe was part of the first wave of Internet adoption, then the Mediterranean countries are part of its second wave.
Worldwide, there was a decided continental pattern. In general, the nations of Europe and North America were towards the top of the list while South America, Africa, and Asia were towards the bottom. However, there were some notable exceptions to this trend. In North America, the Spanish-speaking countries were below the median, while in Asia, the English-speaking countries were above the median.
This in no way implies that English is the language of the Internet. English-speaking countries had an early advantage in the first wave of deployment, because of the use of the ASCII character set (ASCII is the acronym for the American Standard Code for Information Interchange). However, because most Web browsers and e-mail clients now support multiple character sets, there is now a significant amount of Web activity in other languages besides English.
In fact, in a study conducted last year by Databank Consulting on behalf of the European Commission Directorate General XIII, researchers found that 40 percent of all European Web sites are written in the native language only, with no English text visible on the home page. Around 38 percent are bilingual or better (with English as the second language), and just 22 percent are written in English alone. Even in highly English-literate countries such as Norway and the Netherlands, English is the second language of the Web, not the first.
Similar procedures were used to calculate the Wired Ratio and the other ratios, using the formulas outlined above. The Wired Ratio was found to be above average in 55 countries, of which these are the top 25:
The Wired Ratio was lowest in Ethiopia, Afghanistan, Tanzania, Myanmar, Bangladesh, Mozambique, Sudan, Zaire, Kenya, and Nigeria. Countries very near the worldwide median included Saudi Arabia, Romania, Uzbekistan, and Chile.
All countries above the median for Internet usage also were above the median for per capita income (the Wealth Ratio). This suggests that all countries with above-average Internet activity also have above-average per capita incomes. Of course the Internet didn't produce the wealth, but wealth enabled the purchase of a computer, phone line, modem, and subscription, which led to demand for Internet services.
However, there are many countries with high incomes where the Internet has not quite caught on yet. There were 21 countries above the worldwide median for Wealth that were significantly below the median for the Internet Ratio. This list includes Saudi Arabia, Oman, Venezuela, Thailand, Trinidad, U.A.E., Argentina, Mexico, Brazil, Bahrain, Uruguay, Chile, Kuwait, Poland, Malaysia, Slovakia, South Korea, Cyprus, Latvia, Malta, and Taiwan. In other words, these countries have high levels of economic activity but currently low levels of Internet activity.
Could these countries form the next wave of growth on the Internet, after North America and Europe? They are all countries that have demonstrated their interest in road, rail, phone, and electrical infrastructures, as shown by the other ratios. Perhaps they are the next wave, if one accepts that the Internet arrived early in North America and Western Europe and is arriving currently in the Mediterranean region and in Eastern Europe. Perhaps the Middle East and South America will be the next regions to experience explosive Internet growth.
The next step was to compare the Internet Ratio and the other four ratios, looking for correlations. This was done by means of an X-Y scatter plot, with each of the 100 countries represented on a chart by a dot. Inevitably, most of the 100 dots will be clustered near the vertex, but the exceptions will stand out from the pack. The farther to the right and above the (1,1) value a dot is located, the farther that country is above the worldwide median. A dotted line at 45 degrees (allowing for the different scales used for the X and Y axes) was inserted as a visual aid, to show which of the ratios was dominant.
The scatter plots reveal that the amount of Internet activity within a country, as measured by the detected number of hosts, domains, and Web pages, does seem to have a strong correlation to the wealth as well as to the energy, communications, and transportation infrastructures within a country. Correlations with road, rail, and population density are much weaker. In general, though, countries with more per capita communications infrastructure also had more per capita Internet activity.
The highest correlation of all was found between the Wired and Wealth Ratios (Figure 1). Other strong correlations were found between the Internet and Wealth Ratios (Figure 2), and between the Internet and Wired Ratios (Figure 3).
The countries depicted the farthest to the right in Figure 1 are those with the highest per capita GDP. The countries farthest towards the top are those with the highest per capita distributions of telephones, televisions, and electrical production. The dotted line, though it appears to be much closer to the X axis, is actually positioned along the 45 degree line, because of the unequal scale of the X and Y axes.
Countries above the dotted line are farther above the worldwide median for TV, phone, and electrical usage than they are above the median for per capita GDP. In other words, they use these networks at a higher rate than their wealth suggests they would. Norway, Canada, Iceland, and Sweden are notable in this respect.
Countries below the dotted line, such as Hong Kong, Luxembourg, and the United Arab Emirates, for some reason have fewer phones, TVs, and electrical usage than their wealth suggests. Most countries, however, are very close to the diagonal, suggesting a correlation between the Wired and Wealth Ratios. As per capita income rises, so does usage of telephones, televisions, and electricity.
The correlation in Figure 2 is less strong than in Figure 1. One reason may be the relative youth of the Internet as compared to the telephone, television, or electrical grid. However, even after only a few years of growth, the Internet Ratio is taking on a pattern similar to that found with the Wired Ratio. In future years, it will be interesting to see if the scatter plot for the Internet Ratio begins to mirror the scatter plot for the Wired Ratio.
At present, there are some noticeable differences between Figures 1 and 2. All the dots plotted for the 100 countries have the same relative location along the X axis, but different locations along the Y axis. For instance, Bermuda still has the wealthiest inhabitants, but its per capita Internet usage is a bit below its per capita usage of TVs, phones, and electricity.
The countries towards the top of the chart are those with the most per capita Internet usage, as measured by their registration of Internet hosts, their creation of Web pages, and their formation of Internet domains. Finland tops this list, followed by the USA, Iceland, Norway, Sweden, Australia, Switzerland, and Canada. These are countries that have a disproportionate share of the Internet -- not only given their populations, but also given their wealth. The more typical countries are those just above or just below the diagonal, such as the United Kingdom, Ireland, Belgium, Germany, and Austria.
Japan, France, and Italy are noticeably below the diagonal. What this means is that these are the countries with the most room to grow, coupled with the means to do so. They have the money required to deploy an Internet infrastructure and to support an Internet market. Indeed, Japan and Italy are among the growth leaders on the Web, as computed by the number of Web pages published each month. Perhaps if this study is repeated in a few years' time, they will have reached their potential.
Two special cases deserve mention. Kuwait (KW) and the United Arab Emirates (AE) are virtually along the X axis. This means that for some reason, these countries have the wealth to support an Internet infrastructure but they have little of it currently. Both are proportionally closer to the norm in their usage of phones, TVs, and electricity, as shown in Figure 1. Why they are so far below the norm for Internet usage remains something of a paradox.
Finally, a direct comparison of the Internet Ratio and the Wired Ratio shows which countries are leaders in both respects. Norway, Finland, Canada, the USA, Iceland, and Sweden seem to form one group. They are the leaders in their usage of both the Internet and other wired networks. On the one hand they have the highest diffusion of phones, TVs, and electricity, and on the other hand they have the highest diffusion of Internet technology.
Another group is defined by the 16 or so dots that cluster around the diagonal. This group includes Australia, New Zealand, Switzerland, Denmark, the Netherlands, and others. Again, the diagonal cuts this group in half. Those above the diagonal have more Internet than phone/TV, while those slightly below, such as Japan, France, Bermuda, and Estonia, have more TVs and phones than Internet hosts and Web pages.
It is beyond the scope of this paper to examine the socioeconomic reasons why the development of these infrastructures occurs at different rates in different countries. However, the study verifies what many people intuitively believe: that the deployment of the world's new "superhighway" of the Internet follows familiar patterns begun in the late 1800s for rail and phone, the early 1900s for road and electric, and the mid to late 1900s for television. The correlation is very strong with highly related technologies such as the telephone, but it is less strong with unrelated networks such as the railroads.
There were some notable exceptions. Three countries were above the worldwide median for Internet activity but below the median for road and rail: Iceland, Australia, and Canada. This can be explained by those countries' unique geographies, which in the case of Canada and Australia feature a few population centers coupled with vast amounts of scarcely traveled territory. In Iceland, there are no railroads and a relatively small amount of roads.
There was only one country that was above the median for Internet activity but below it for the Wired Ratio. This was San Marino, which also was the only country among the 100 that does not produce any electricity (it relies on the capacity of surrounding Italy). However, assuming the people of San Marino consume the same amount of electricity as their Italian neighbors, and therefore recalculating their Wired Ratio, then there are no exceptions to the rule: all countries with above-average usage of TVs, phones, and electricity also have above-average usage of the Internet.
There were no countries below the median for the Wealth Ratio which were above the median for the Internet Ratio. In other words, all countries with per capita GDPs above $5,839 also had above-average rates of Internet usage.
The highest Paved Ratios were observed in geographically small countries. Singapore was the most densely paved, followed by Belgium. In fact, all the Benelux countries were high on the list. Interestingly, four countries near the top of the list apparently have no railroads: San Marino, Bahrain, Bermuda, and Malta.
Iceland and Canada, which are among the leading per capita users of the Internet and other networks, are among the lowest on the Paved scale. As mentioned, Iceland has no railroads, so of course it scored low on the Paved Ratio. Norway and Sweden, leaders on the Internet, are only slightly above average on the Paved scale. Conversely, heavily paved countries such as Sri Lanka, Trinidad, Bahrain, Cyprus, and Malta are very low on the Internet scale. This means the correlation between road & rail and the Internet is weak at best.
Those with the lowest Paved Ratios included Sudan, Botswana, Ethiopia, Chile, Afghanistan, Mozambique, Panama, U.A.E., Algeria, and Myanmar. Countries very near the worldwide median included Kuwait, Sweden, Pakistan, Uzbekistan, Brazil, South Africa, Nicaragua, Ecuador, Indonesia, and Mexico.
A Density Ratio also was computed, to help discern whether comparisons of infrastructure to population or area correlated to population density. They did not. There were vast and empty countries such as the USA, Norway, Sweden, Canada, and Australia which had high amounts of Internet activity. There were vast and empty countries such as Sudan, Afghanistan, and Algeria which had little or no Internet activity. Therefore, being a big and empty country, in and of itself, is not related to Internet usage.
At the other end of the spectrum, there were densely populated countries such as Hong Kong and Singapore which had high levels of Internet activity. But there also were densely populated countries such as Bangladesh, India, Vietnam, and Pakistan where there was little or no Internet activity. Once again, there was no correlation between population density and Internet usage.
Therefore, although the Internet could be useful in sparsely populated and lightly paved countries, where telecommunications networks could be used as the means to span the distance between people, this does not seem to be the case. The popularity of the Internet does not seem to correlate well with population density.
Among the 20 largest countries, as measured by population, the USA, Britain, Germany, and Japan had the highest level of Internet activity, and Bangladesh, Nigeria, Ethiopia, and Iran had the lowest.
Among the 20 largest economies, as measured by total GDP, the USA, Canada, Australia, and the Netherlands had the highest per capita Internet Ratios. Iran, India, China, and Indonesia had the lowest.
Among the 20 richest populations, as measured by the Wealth Ratio, the countries with the most Internet activity were the USA, Iceland, Norway, and Sweden. The wealthy countries with the lowest Internet activity levels were the U.A.E., Hong Kong, France, and Italy. Of those four, however, only the U.A.E. was also below the worldwide median for Internet activity.
There is one inescapable conclusion that permeates the data: the Internet is more popular in countries that long ago established large infrastructures for communications and transportation. Conversely, the Internet is less popular in countries without extensive networks for electricity, television, or telephone. This is independent of geography or region, although most of North America and Western Europe are consistently above the median.
If the data has any predictive value, it suggests that Internet activity in the rest of the world has a lot of room to grow before it approaches the high rates currently seen in Scandinavia or North America. Assuming that those regions were part of the first wave of Internet growth implies that a second and third wave are about to sweep the world. The second wave seems to be washing across the Mediterranean and Eastern Europe, while the third wave seems headed for the Middle East and South America.
Nine countries in Scandinavia and North America together have only 6 percent of the world's population, but they contain 73 percent of its Web pages and 79 percent of its Internet hosts. For the world to reach their somewhat saturated levels for communications infrastructure, it would need
That is a lot of room for growth in the next century!