————————————————————————————————

Monthly Discussion

The Internet

Several attempts have been made to adapt an S-curve to the growth of Internet. I present a rather serious such attempt in Exhibit 3. It is serious because it uses detailed data made available from the Internet Software Consortium going back to August 1981. It is also impressive because the fitted S-curve describes well the data over many orders of magnitude of growth.

The conclusion is that the number of hosts on the Internet has been growing following a natural growth pattern, and has already gone beyond the mind-point of the growth process. In January 2002 Internet had penetrated its market by 57%. The final ceiling is estimated as 258.3 million hosts worldwide to be achieved toward the end of the decade. The uncertainty on this number is ±10% with 95% confidence level (see discussion in Appendix B of Predictions). The midpoint was on 15-Jul-2001 and the uncertainty on this estimate is ±28 days (again with 95% confidence level).

One may want to argue against the “fatality” of this result given that a major part of the world is still not connected. By my argument is that the agreement of data and S-curve is excellent, and natural growth processes that advanced that far invariably proceed to completion, under normal circumstances. In my mind, it is likely that another “species” of hosts may appear in the future and begin growing along a follow-up S-curve, for example, hosting without IP addresses, or hosting without using computers (via such devices as portable telephones).

An associated growth pattern—and one that can be used to corroborate the above results—is the appearance of Internet users, as defined by the number of people online. The data for this study come from Nua Internet Surveys, among the most reliable databases reporting on the Internet. The data and an S-curve fit are shown in Exhibit 4. The results of the fit on the number of people online corroborate the earlier results on the number of hosts. Once again the process is more than half completed, a penetration of 68% was witnessed in February 2002. The ceiling of the growth process is estimated as 852 million people, and the uncertainty on this number is ±20% with 95% confidence level (the agreement between data and fit is not as good here as for the number of hosts.) The midpoint of the growth process was on 14-Feb-2001 and the uncertainty on this number is ±44.42 days (again with 95% confidence level).

The Growth of Hosts on the Internet

Exhibit 3. The data (open circles) represent the hosts with at least one IP address assigned. The data come from the Internet Domain Survey published by the Internet Software Consortium. The arrow points at the midpoint.

But 852 million people online represents only about 15% of the World population! What about the masses of people in Asia and Africa, who at some point will also connect?

As with the number of hosts, we must envisage follow-up S-curves to describe the belated entrance of people from less developed countries. The curve in Exhibit 4 depicts the filling of the first niche, which predominately represents the West. The 3^rd and 4^th worlds will follow their own curve that will be distinctly different from the one studied here.

Exhibit 4. The data (open circles) represent the number of people online as estimated by the Internet Domain Survey published by NUA. The arrow points at the midpoint.