Source: 2005 World Bank, ITU

Measuring the level of internet use within a given society presents several methodological difficulties. The numbers that are most easily collected and verified (e.g. the number of internet hosts in a given country) provide only an approximation of the number of people actually using the internet. Constructing estimates of the number of users from sources like the number of hosts is a hotly debated topic, and many of the participants in the debate have commercial interests concerning a particular side of the issue. Even if the number of currently possible internet addresses is known, they can be dynamically assigned (and because of other technical reasons), there is no correspondence between the number of addresses and the potential maximum number of users. Furthermore, several of the most popular top level domains (.com, .net, and .org for instance) are not geographically fixed. By convention, these addresses are counted with the United States, but there is no systematic and definitive way to determine their location, and as the international development of the internet continues, assigning these addresses to the US will increase the inherent inaccuracies. Estimating internet users is thus likely to remain a matter of approximation, more akin to measuring television audiences than to counting votes.

Despite these difficulties, statistics like numbers of hosts allow for comparative evaluation. The correlation between wealth and internet penetration is not particularly strong. At the present stage of development, English as a primary language, government policy, and even individual effort have strong secondary impacts.