It is sometimes said that:

* IP addresses are hoarded by “developed nations” - if only “underdeveloped” nations were given more IP addresses, the Internet would grow more/better…

Assertions like this mistakenly conflate the administrative process of requesting and receiving public IP addresses with the economic or commercial act of routing IP addresses - of engaging in what is sometimes called “Internet production.” The former, administrative process involves relatively little in the way of overhead, and confers nothing more than the potential to develop public Internet resources—i.e., to create new Internet users (provide access) and/or Internet uses (provide content and other online services).

Actually achieving that potential takes much more than IP addresses. At minimum, it also takes equipment, engineering staff, a physical medium through which the new Internet resource can be delivered to end users, and a strategy for putting all of these pieces together so that they are worth more then the sum of their parts—in other words, a strategy for creating value for users. This rule holds—or should hold—whether or not those users happen to be called customers, coworkers, community members, or citizens.

True, an institution that possesses all of these features but lacks public IP addresses cannot easily create new, globally visible Internet resources, so addresses do indeed represent a kind of bottleneck to “Internet production.” However, the only requirement for getting (or more precisely, for securing conditional exclusive beneficial control of) public IP addresses is, in fact, to possess those exact features, or means to secure them through external suppliers. The process through which independent network operators request new IP addresses from the Regional Internet Registries involves nothing more than demonstrating that one has the minimum prerequisites needed to engage in actual Internet production, and a plan for producing specific new Internet resources. Once this potential is credibly demonstrated, a new “allocation” is made, conferring upon the applicant beneficial control over a quantity of IP addresses that matches the demonstrated need. Viewed in this light, each public IP address request is like a promise undertaken by a network operator to produce a specific quantity of Internet resources within a fixed period of time. Each actual IP address allocation represents a promise of Internet production that has been validated. By the best measure currently available, an estimated 90% of networks that enter into such arrangements start delivering on those promises within 75 days of receiving their public IP allocations.*

So, no public Internet production without public IP addresses—but more importantly no public IP addresses without Internet production, or at least the promise of near-term Internet production. What happens if this production doesn’t materialize? In that case, no further IP allocations will be made to the non-producing institution. This closed loop arrangement represents the best possible system for balancing the interests of current and future Internet participants, both “Internet producers” and end users. Current Internet beneficiaries know that they won’t face any local, artificial scarcity; one day everyone may have to transition to a new addressing scheme, but this requirement won’t be imposed arbitrarily or asymmetrically because of “hoarding.” Meanwhile, those on the less fortunate side of the digital divide can be assured that public IP addresses will always remain available to them—and/or to any other institution that aspires to facilitate their incorporation into the global information economy—for as long, and to the same degree, as they are available to everyone else.

Although this description captures the essence of the public IP allocation process, it is true that the process and requirements for securing IP addresses have not always been the same for all Internet participants at every point in history. The above description is based primarily on the current process used by multihomed “Internet producers” (i.e., those with external links to at least two other networks) to secure IP addresses directly from the Regional Internet Registries (RIRs). This admission points to two distinct dimensions of variance, one historical the other related to network scale and topology. In the past, each of these variables has been used in different contexts by different parties to support claims that the system, or perhaps the results of the system, are unfair. Occasionally each reason is also used to explain or obscure the other.

Recently, for example, representatives from some information economy latecomers—mostly PSTN-dominated developing countries—have sharply criticized older networks—mostly based in developed economies with competitive telecom/Internet markets—because the latter received IP address allocations under earlier, more relaxed terms, when Internet production technology was much less efficient than it is today. Such critics would prefer to lay the blame for the current digital divide on this “historical unfairness,” rather than to consider whether and how their own current domestic circumstances—which often include monopoly telco control over basic Internet inputs—adversely affects their own Internet production possibilities. In some cases, this control extends beyond essential telecommunications infrastructure (especially “last mile” access), to encompass public IP addresses themselves, sometimes through the vehicle of a National Internet Registry (NIR), or more commonly through exploitive but unregulated commercial practices of a local service provider, or LIR.

On the other side of the argument, representatives of the old, well-established networks—many having grown far beyond the confines of their original host economies—are often quick to point out the relationship between provider diversity and public IP addresses. They note, quite correctly, that countries that play host to more independent network operators tend to make more public IP allocation requests, which ultimately result in more public IP address allocations. This empirically verifiable fact would seem to suggest a clear path forward for aggrieved developing countries—use domestic policy mechanisms to encourage the formation of more local providers. Arguably, such a suggestion glosses over substantial variations in the real cost and availability of critical Internet inputs, and thus obscures any role that established, globe-spanning networks have in influencing those costs at the international level. And yet a variety of domestic levers for empowering local Internet production are available to every country, ranging from regulatory changes that would widen access to telecom infrastructure and other critical inputs, to the establishment of Internet Exchange Points (IXPs), where local Internet producers can trade access to new Internet resources with each other directly, without the mediation of national PSTN or foreign network operators.

It is possible that a strategy for fostering national network provider diversity could be pursued in bad faith by a powerful incumbent, simply in order to secure administrative control over more public IP addresses for their own sake. However, so long as each public IP request is reviewed objectively, on the same terms developed through international consensus and executed at the regional/global level, the expected result would extend beyond the merely administrative, beyond the simple transfer of inert public IP addresses; they would represent the same kind of credible and verifiable promise to produce real Internet resources—new users, new usage, new content and other uses—that has to date paid off so handsomely throughout the global information economy.

For these reasons, national-level experiments with address allocation policy should be strongly discouraged, just as nonconforming IP address assignment policies at the enterprise level are strongly discouraged today. Better for all if the rules linking public IP addresses with real Internet production continue to be defined and administered outside of the equally competitive commercial and national/political spheres, and that these rules continue to be treated as transitive features, applicable and enforceable over every actor and every transaction involved in the distribution of public IP addresses. That is the best way to make the Internet grow more/better, for now…

—-
Further Reading:
* Support for this statistic, and many other valuable insights on the relationship between measures of address allocation and Internet production can be found in Meng, Xu, Zhang, Huston, and Zhang, IPv4 Address Allocation and the BGP Routing Table Evolution, also published in the January 2005 SIGCOMM Computer Communications Review.

Additional research on Internet economics, IXP construction, and the relationship between telecom policies and Internet development is available at Packet Clearing House.