Trends in Internet Map Use A Second Look
Dr. Michael P. Peterson
Department of Geography / Geology
University of Nebraska at Omaha
Omaha, NE 68182 USA
Beginning in 1993, with the release of the Mosaic, the
first widely available World Wide Web browser, usage of the Web has increased
dramatically. The new medium is used to distribute a wide variety of documents,
from scientific data to government reports. Maps represent an important
and growing component of Internet traffic. The World Wide Web has led to
a major change in how maps are distributed and used. It is important that
we understand this new medium and especially the extent to which it is used
for delivering maps to a potential map user. This multi-year study presents
the results of monitoring a representative sample of Web sites that distribute
maps. At the ICA meetings in Stockholm in 1997, it was reported that a selected
group of sites on the World Wide Web respond to nearly 1 million requests
for maps on a daily basis. Based on this and other statistics, it was estimated
that at least 10 million maps a day were sent via the Internet. It was also
noted that some map sites were not increasing above 100,000 hits per day
likely due to slow response times and competition from other map sites on
the Web. The growth in Internet map use, exponential since 1994, showed
signs of slowing and increasing in a more linear or straight-line fashion.
In retrospect, this seems only to have been the case for individual map
sites. The results now indicate that web map usage continues to grow in
a dramatic fashion. Yearly averages of "map hit data" show a renewed
exponential growth in map distribution through the Web. A four-fold increase
in web map usage is estimated over this two year period.
The World Wide Web (WWW) has become a major communications medium and maps have emerged as one of its major components [Peterson 1997a]. A study performed by the author in 1997 [Peterson 1997b] examined five map sites on the web and found a combined usage of about 1 million maps a day. It was estimated that total map usage was about 10 million a day. The purpose of this study is to examine general trends in the growth of the web, web user demographics, and provide an updated estimate of daily Internet map use.
Examining how maps are distributed and used is important in understanding trends affecting cartography. Through the web, a new generation of map users are interacting with maps in an entirely different way. The traditional cartographic products may have little meaning to these "new" map users once they have been exposed to interactive mapping through the web. Or, perhaps, the web will generate greater interest in maps of all kinds. As cartographers, it is important for us to understand changing use and attitudes about maps.
More than any other technological development in the past
century, the web forces us to examine the purpose of cartography and our
means of map distribution. For many applications, the web may serve as the
best medium to communicate spatial information in an economical and efficient
manner. If so, cartographers should find ways to adapt their maps to this
2.0 The Internet and Web Usage
Having its beginnings as ARPANET in 1969, the Internet now consists of several data communications protocols including e-mail and the file-transfer protocol (FTP). The dramatic increase in the use of the Internet during this decade can be attributed the WWW. Conceived at the European Particle Physics Laboratory (CERN) in Switzerland, the WWW introduced the principle of "universal readership," a concept that networked information should be accessible from any type of computer in any country with a single program. It was originally intended to assist researchers in high energy physics research by linking related documents. The developers wanted to create a seamless network in which information from any source could be accessed in a simple and consistent manner. A prototype of the new protocol was finished in 1991. The first widely available browser, Mosaic, was introduced by the National Center for Supercomputer Applications (NCSA) in 1993. Netscape, a commercial successor to Mosaic, was introduced at the end of 1994 and Microsoft followed with Explorer in 1995.
The Web grew quickly. In June of 1993 there were only 130 web servers. By mid-1995 there were 23,500 web servers and this had grown to 230,000 by 1996. In June 1998 there were 2.4 million web servers. Measured in terms of unique domain names, there were 29.7 million in January 1998. This had grown 45% to 43.2 million by January of 1999 [Thompson 1999a]. The top graph in Figure 1 illustrates the growth of Internet hosts from 1993 to 1999. The bottom graph shows growth by domain from 1998 to 1999.
Figure 1. Growth of World Wide Web Domains. Most new domains are either ".com" or ".net." (www.thestandard.net/metrics/display/0,1283,848,00.html)
The web now dominates the Internet. By 1999, the WWW generated 68% of all Internet traffic while e-mail and FTP each had about 11% (www.cyberatlas.com). The remaining traffic consisted of a variety of other protocols. It is expected that the percent of web-related Internet traffic will begin to decline as audio and video protocols begin to dominate.
As the web has become increasingly commercialized, a considerable amount of data has been collected on web usage. Simple "hit" data tracks the number of times a page is accessed by visitors. Table 1 lists the top 25 web sites for March of 1998. The sites are ranked by the number of unique visitors age 12+ in the United States. The reporting period is from March 2, 1998 through March 29, 1998. The map provider, MapQuest, is listed as number 21. Intense competition among commercial sites caused MapQuest to fall out of the top 25 in 1999.
Table 1. RelevantKnowledge  "Top 25 Web Properties" list for the month of March.
Number of "Hits"
|The Weather Channel||3,662,000|
* Site data on the "Top 25 Web Properties" list is based upon aggregation of multiple domain names. A complete list of domain names/URLs, for each site can be found at www.RelevantKnowledge.com.
The number of Internet users is growing rapidly. Estimates of Internet use in the United States are fairly consistent. Intelliquest reports that 79.4 million Americans were are online in the fourth-quarter of 1998 (www.intelliquest.com). The Fall 1998 Cyber Stats by Mediamark Research found that 72 million American adults have access to the Internet (www.mediamark.com). The Computer Industry Almanac estimates there were 76.5 million Americans online at the end of 1998 (www.mediamatrix.com). How the web is used is also measured. NetRatings reports that the average web user spends nearly six hours looking at 355 pages per week over 10 sessions (see Table 2). They also found that each page is examined for less than a minute.
Table 2. Web Usage Statistics for the Week of February 28, 1999. Statistics represent average activity for a Web user, as reported by NetRatings, Inc. (www.mediamatrix.com).
|Page Views Per Week||355|
|Time Spent Per Week||5: 52: 00 m|
|Number of sessions per week||10|
|Pages visited per surfing session||36|
|Time spent during session||35: 12 m|
|Duration of a page viewed||00: 59 m|
|Banners viewed week||124.28|
|Banners clicked on during week||1.18|
Once primarily used the upper-middle class and the well educated, the Internet has become more mainstream. The report found that 51 percent of those planning to get Internet access are over the age of 35. Almost half (49 percent) of the group have only a high school education or less. More than half of those planning to go online (58 percent) make less than $50,000 a year.
Web users in the United States are also divided relatively equally by sex. 52% of web users are male and 48% are female (the actual percentage of male and female is 52% female and 48% male). Web surfers are only 9% more likely to be male. In terms of age, web usage remains high until about the age of 55. Only 6% of 55-64 year old people had accessed the web in the past 30 days. This is compared with 26% in the 25-34 year old category and 28% in the 35-44 age group (Thompson 1999b)
The number of Internet users around the world is constantly growing. The Computer Industry Almanac has reported that by the year 2000, 327 million people around the world will have Internet access. This is up from 61 million in 1996 and 148 million in 1998. Estimates for 2005 are 720 million. The top 15 countries will account for nearly 82% of the these worldwide Internet users (including business, educational, and home Internet users). By the year 2000 there will be 25 countries where over 10% of the population will be Internet users (www.cyberatlas.com/big_picture/geographics/stats.html).
Access to the Internet varies considerable by country. Table 3 shows the approximate populations that have Internet access. A common definition "Internet Access" is use of the Internet in the past week. The percentage of people that have access to the Internet varies from 61% in Iceland to .02% in Vietnam.
Table 3. Internet Access by Country. Number of users are given in millions. The percent of users is specific to the population of each country.
# of Users
% of Users
This general distribution is reflect in the map in Figure
2 that depicts the world distribution of Internet hosts in July of 1998.
North America, Europe, and the Pacific rim dominate in the location of Internet
hosts, but a large number of Internet hosts can also be found in South America,
the South Pacific and South Africa.
Figure 2. World Distribution
of Internet Hosts in July 1998.
3.0 Internet Map Use
As was discovered by the author in 1997, assessing map use on the Internet is not an easy task. There are large number of web sites that distribute maps but no coordinated method of measuring the number of maps that are downloaded. Beyond this, of course, there is no way to determine if the maps are effectively utilized. A centralized "counter" is still needed to at least determine the number of maps that are distributed via this new medium.
Determining "hit rates" at commercial sites has become even more difficult than in 1997. Although these sites maintain such data to attract advertisers, they have become less willing to share such data because of competition with other sites. When these sites were first started, press releases issued by the companies detailed how quickly usage was growing. MapQuest.com was especially open about the number of maps it was distributing as it was attempting to dominate Internet mapping. This data has now become "a property" and is handled by independent companies that then assess how much each web site can charge for advertising banners. The independent "auditing" companies demand that their clients not reveal "hit" data because this represents the basis of their business.
Data remains publicly available at some non-commercial mapping sites. The best of these remains the XEROX Parc site (pubweb.parc.xerox.com). Although operated from a commercial domain, the site is essentially run by a single employee, Steve Putz, who wanted to demonstrate how the web could be utilized for interactive mapping. Begun in June of 1993, his site is among the longest running interactive mapping sites on the web. However, as was noticed in 1997, usage of the site has leveled off to around 100,000 maps a day (see Figure 3). This is probably because of computer and network constraints. Generating 100,000 user-defined maps a day would still mean that the computer would need to make 70 maps a minute if the load were equally distributed. Of course, this is not the case and slow-downs occur at peak times. Web users tend not to wait long for a site to respond and quickly move to another page.
Figure 3. Graph of XEROX Parc Map Server Daily and Monthly Requests. (Source: http://pubweb.parc.xerox.com/mapdocs/usage.html)
Another non-commercial web site that is mostly operated by a single individual is EarthView (www.fourmilab.ch/earthview/). John Walker, an American who works at the research lab in Switzerland, is responsible for programming the site that produces maps of the earth and of the moon (see examples in Figure 4). He reports the following:
In recent months, actual requests for Earth Viewer images are running about 35% of the total daily requests to my site. This count includes only those requests which actually delivered a custom image to the requestor--accesses to the Earth Viewer welcome page, background documents, etc. are not included in this figure. With total daily valid accesses (not counting erroneous ones) averaging about 180,000, this gives an average of about 63,000 custom map requests per day. Average accesses per day have been rising since the site's inception, often jumping discontinuously when the site is mentioned in the press or a link is placed on a frequently visited page of a high-profile site. Since last November accesses have been on a plateau of about 180,000 per day due to the simple fact that the 512 kb leased line is now in saturation about 10 hours per day. Shortly I will bring up a mirror site on a T1 line in North America with very good connectivity to the Internet backbone. By diverting requests from the western hemisphere to that site I expect to be able to simultaneously provide faster response, support more traffic, and reduce the load on the leased line here. [Walker 1999]
In commenting on a question concerning daily and weekly variations, Walker says:
I see a strong and consistent diurnal variation with the peak right where you'd expect it, when all North American time zones are hitting the site, but no strong weekly or monthly cycle (though the bandwidth saturation may be masking such a cycle). I do notice a slight decrease during long week-end holidays in the U.S., which indicates that at least some people spend their holidays somewhere else than on the Web. [Walker 1999]
Figure 4. Maps of the earth and moon from the www.fourmilab.ch/earthview/ site. Both depict areas illuminated by the sun at the time the images were downloaded.
It seems that non-commercial sites have difficulties exceeding 100,000 maps a day (70 maps a minute daily average). This limitation is a result of lack of funding to support both faster equipment and faster connections. In 1997, computers at the commercial MapQuest site were able to generate 1000 maps a minute. MapQuest now responds to an average of about 2,500 user-defined maps a minute. Peak map production would need to be much higher than this.
Table 3 shows the comparison between the 1997 and 1999 map use data at the five sampled map sites. Data for one of the sites was not available in 1999. Usage over the two year period at the five sites (data for one site was not available in 1999) has increased from 1 million to a little over 4 million. Most of the increase is map web map usage is due to the increase at MapQuest. But, with the exception of XEROX Parc, non-commercial map sites with minimal funding register about a two fold increase in use. It was estimated in 1997 that the 1 million confirmed maps transmitted via the web represented only 10% of all maps distributed through the web on a daily basis. Following this estimate, this would mean that approximately 40 million maps are distributed on a daily basis in 1999.
Table 3. Hits on Selected Map Sites 1979 and 1999.
|Map Site||1997||1999||Source of Information|
|Tiger Mapping||35,000||75,000||Information from Chris Stuber, Census Bureau.|
|EarthView||35,000||63,000||Information from John Walker, CERN, Switzerland.|
|MSU Weather||159,000||NA*||Data not available for 1999.|
|Total Estimate||10 M||40 M|
Usage of the commercial site, MapQuest, has increased the most ñ from 700,000 to about 3.8 million maps a day (see map Figure 5). Crampton  labels MapQuest the biggest map publishing house of all time. MapQuest's goal, as described on their "About Us" web page, is to "pioneer new ways for businesses and consumers to use interactive mapping on the World Wide Web." It is consistently rated as the top travel site on the Web. MapQuest provides a variety of services including:
Figure 5. Map of Ottawa from MapQuest.com. The interactive mapping site is the largest on the web.
Chris Stuber of the Census Bureau reports that the Tiger Mapping Service (see Figure 6) now has two computers and has doubled its daily distribution from 35,000 to 75,000 map a day [Stuber, 1999].
Figure 6. Map from U.S. Census
Bureau's Tiger Mapping Site. The site presents a variety of options for
modifying the map.
Estimating map usage on the web remains a difficult task. Commercial sites that have the resources to invest in computers and programmers and distribute the largest proportion of maps through the web but provide only occasional data on usage. Non-commercial sites continue to grow in map usage but at a much slower pace, probably because of computer and network restrictions that lead to slowdowns.
Better means need to be established to monitor the extent of map usage through the web. Certainly, a large number of maps are being distributed and used through the web. It is difficult to draw any conclusions on Internet map use without having more data. As was suggested, a centralized "counter" needs to be established that that counts maps distributed by the major mapping sites. It will likely remain impossible to count every map that makes it through the Internet, either through the web, e-mail or FTP.
The Internet, particularly the web, has established itself
as a major form of map distribution. In addition to determining the number
of maps distributed in this way, it is also vital to understand how these
maps are being used and whether map usage will become more interactive.
Certainly, the possibility exists that users will not accept the use of
static maps. It is also possible that the use of maps on paper will increase
after people have become accustomed to using maps on the Internet.
Crampton, Jeremy (1998) The Convergence of Spatial Technologies, Cartographic Perspectives, no. 30, pp. 3-6.
Peterson, Michael P (1997a) Cartography and the Internet: Introduction and Research Agenda. Cartographic Perspectives, 1997, No. 26, pp. 3-12.
Peterson, Michael P (1997b) Trends in Internet Map Use. Proceedings of the 18th International Cartographic Conference, Stockholm, Sweden, Vol. 3, pp. 1635-1642.
RelevantKnowledge (1998) www.RelevantKnowledge.com, April 8, 1998.
Stuber, Chris (1999) Personal communication, March 10.
Thompson, Maryann J (1999a) Spinning an Ever-Widening Web. The Industry Standard, March 1, 1999. (www.thestandard.net/metrics/display/0,1283,848,00.html)
Thompson, Maryann J (1999b) Spotlight: The Demographics of Who's Online. The Industry Standard, March 8, 1999. (http://www.thestandard.net/metrics/display/0,1283,846,00.html)
Walker, John. (1999) Personal
communication, March 1.