A medium of exchange is...

...a class of general purpose technologies that facilitate liquidity, or the easy exchange of goods, services, and information (sometimes called real factors) in the presence of various transactional frictions -- including physical distance and inter-temporal gaps between producers and consumers, the presence of uncertainty and mistrust, and the infrequency of coincidences of wants between diverse, independent economic agents.


To date, the function of “medium of exchange” (MoE) has been almost exclusively identified with the specific set of technologies known as money. However, the two terms are not synonymous. Money is generally understood to embody several other functions in addition to that of MoE -- e.g., means of payment, store of value, unit of account, et al -- each of which is distinct, and may conceivably exist in isolation from the others. 


While each of these functions is generally regarded as independent by monetary theorists, it may be more accurate to say that money’s capacity to serve as a MoE for the scarce physical goods and services that characterize the conventional economy is contingent on the bundling together of these services within a single instrument. For example, when individual goods are rival -- that is, they are wholly transferred or consumed (and hence unavailable to the selling party to conduct further transactions) in the process of exchange -- then it makes sense for the MoE to undergo a similarly absolute (functional if not physical) transference in the opposite direction. In this sense, money is well suited to serve as a MoE in the conventional economy because it circulates, or changes hands, typically in the opposite direction of the goods and services being exchanged. The common term for this function is means of payment.


Money’s operation as a MoE is contingent on its widespread acceptance as a means of payment in exactly this sense. In turn, money’s broad utility as a means of payment depends on its capacity to function as both a store of value and as a unit of account. The first of these features implies that money earned in one place today can be held with confidence that it will continue to be valuable -- i.e., to the useful for executing purchases -- in the same way (if not necessarily to the same degree) at other places and times in the future. The second of these features implies that the instrument is decomposable into discrete units that are widely used as a benchmark for calculating the relative value of real factors, or goods and services, which can be bought and sold using the MoE. Together, these collateral features enable consumers to hold money acquired here and now with confidence that it could be used to execute transactions at different places and times in the future. They also provide current producers with some basis for setting prices for the goods that they produce today but hope to sell at other places and times in the future.


While each of the secondary functions associated with money has a role to play in the Internet, none is required in the same way for the Internet’s MoE to work. Unlike the conventional economy’s value-bearing features, the packet-borne goods and services that are characteristic of the Internet are not consumed in the process of exchange. A website viewed by one Internet user does not cause that website to become unavailable or less valuable in any way to subsequent users, nor does the act of receiving a file transfer cause the file in question to become unavailable to its original creator, or to anyone else that might access it thereafter. This is because, for transactions that are executed by means of packet switching, the act of exchange is at root identical to the act of duplication. In economic terms, this implies that Internet exchanges are fundamentally nonrival.


Since Internet-based transactions do not absolutely consume the value-bearing resources that are being exchanged, a circulating MoE is unnecessary to provide liquidity for such exchanges. Arguably, such circulation would be absolutely inefficient for this narrow purpose, since such MoE movements would likely result in a reduction in overall liquidity, as some real factors would become periodically or perhaps permanently detached from the market and hence unavailable for exchange. One should note that this this does not imply that a separate, possibly circulating means of payment can have no role to play in the Internet, but rather that a mechanism of this kind is not -- and likely should not be -- bundled with the Internet’s MoE


Typically, individual public IP addresses are uniquely and persistently associated with the specific points of attachment (or interfaces) between an individual device that plays host to one or more of the Internet’s real factors, and the universe of similarly addressed devices and real factors. In effect, they are the static tokens that mediate the dynamic, ad hoc, demand-driven interaction of nonrival factors across the vast distributed exchange mechanism that we call the Internet. Viewed in this light, this particular combination of nonrival real factors and non-circulating MoE provide the foundation for the Internet’s distinctive attachment-oriented economic model, in which association with the virtual economy and levels of participation in that economy are mutually independent.


While the above description might prompt some to conclude that the liquidity services provided by IP addresses and money exhibit few commonalities that might merit comparison or the use of a common functional terminology for explanatory purposes, such a conclusion would be premature. Subsequent research will illuminate how both MoEs are vulnerable to the same kinds of systemic risks -- “same” in the sense of arising from the same root causes, giving rise to the same differential and progressive effects, and responding to the same basic mitigation strategies. It is also a telling historical fact, given the complete absence of intellectual cross-fertilization between the disparate MoE expert communities, that the same basic systemic risk management mechanisms emerged organically in both MoE-defined domains (note: a preliminary exploration of this historical symmetry is now online here).


Looming Challenges to the Internet’s Medium of Exchange


Although the Internet’s attachment-oriented economic model might be well suited to any system of pervasive, nondirectional exchange of nonrival factors, and arguably fits well with the kind of radical factor-level experimentation that is characteristic of the Internet, it is clear that the attachment model presents some intrinsic challenges of its own. The most basic of these is the incongruence of this model with the finite nature of some of the Internet’s basic building blocks, as well as with the broader, scarcity-based habits and constraints of the encompassing “conventional” economy. These incongruences have given rise to two distinct kinds of concerns, both of which have become increasingly salient in the recent past.


The first kind of challenge arises as a result of the inherent mismatch between an effectively infinite quantity of exchangeable goods and services, and the ultimately finite carrying capacity of the diverse infrastructure elements required to execute those exchange transactions. Long recognized but never clearly diagnosed, critics often describe this particular challenge as an manifestation of the Internet’s “lack of a business model,” prompting them to periodically call for the development of mechanisms to “route money,” or embed a means of payment into the very structure of TCP/IP-based exchanges. Conversely, it has often been identified by advocates as one of the key features that makes the Internet so amenable to “peer production,” and other voluntaristic, non-contractual, and/or gift-like exchange conventions. For better or worse, the inherently binary nature of this attachment-based system has also led at times to MoE-centered commercial strategies for regulating demand for Internet-based services by end users, and also to anticompetitive strategies for discouraging market entry by aspiring new Internet service providers.   


If the first kind of challenge stems from levels of participation in Internet-based exchange that meet or exceed the carrying capacity of required physical infrastructure inputs, the second kind of challenge results when levels of participation fall below the level that would justify the commitment of such inputs, or of associated IP number resources required to make such gross inputs interoperable. Although this problem is not intrinsic to the Internet protocol itself, it is increasingly relevant today because of the relatively small quantity of unique IP addresses that are supported by the current version of the Internet protocol -- and because that pool of globally interoperable IP addresses is now almost fully distributed to current or former Internet service providers.


The proximate concern arising from this looming development is that some IP addresses that were distributed in the past are no longer used, or perhaps were never used. Technology changes over time have reduced or eliminated some older requirements for unique IP addresses, at the same time that administrative improvements have permitted the distribution of IP addresses to be ever more closely aligned with the concurrent requirements of address resource seekers. Unfortunately, the terms under which IP number resources have been distributed generally provide no effective mechanism for recovering number resources that are no longer used by their original recipients. As a result, some IP addresses that may have been needed in the past are now idle, but cannot be easily recovered to support additional Internet growth. As the numbers of aspiring Internet participants and the supply of new exchangeable factors continue to grow unabated, the persistence of these stranded and idled IP addresses has become increasingly contentious.


However, even if this proximate concern did not exist -- e.g., because all IP addresses were 100% utilized, or all remaining idled IP addresses were 100% unrecoverable -- the fundamental problem would still remain: the foundational MoE that is supported by current Internet technology is too limited in quantity to continue providing basic liquidity services for an ever-expanding global base of demand drivers -- including user-generated interactions as well as commercial content and services -- that are actively exchanged over the Internet.


This problem has three possible long-term solutions:


  1. A logical partitioning or segmentation of the Internet into separate domains, each with its own MoE and its own conversion mechanisms for interacting and avoiding unwanted collisions with other MoE-based domains

  2. The introduction of a new public MoE pool that is large enough to support all current and future global demand

  3. A fundamental redesign that replaces the Internet’s protocol-based, attachment-centric economic model with something else


Several solutions of the first kind are already in various stages of implementation or development. Some (e.g., conventional and “carrier grade” NAT, LISP, et al.) employ strategies involving durable, possibly nested MoE diversification at the enterprise and/or industry levels. Others, like the various regional proposals to effectively privatize IP address resources, would likely induce patterns of MoE segmentation that follow the contours of national property laws. While their descriptions vary considerably, the actual effects of introducing such MoE conversion requirements into the Internet are identical to those transactional frictions that are encountered by every financial authority, bank, and money user that is required to engage in frequent currency conversion transactions. In both cases, the additional cost, increased uncertainty, and variability of results tends to marginally reduce incentives both to exchange, and to invest in producing goods and services that have to traverse such conversion frontiers.


Given the long and widespread understanding of these adverse effects within the Internet community, efforts to develop a new and vastly larger IP address resource pool were initiated nearly two decades ago. However, the structure of the IPv6 addressing format that resulted clearly reveals that those adverse effects were not recognized as telltale indicators of the existence of a MoE-mediated system. Consequently, issues like the need for backwards compatibility (i.e., “full convertibility”) and the challenges of facilitating the integration, conversion, and ultimately migration to the new MoE resource pool by diverse autonomous actors did not achieve the requisite level of salience until now.


Unfortunately, history suggests that MoE-mediated systems tend to be highly resistant to many kinds of changes, including the kinds that many originally assumed would drive the spontaneous adoption of IPv6. Historically, new MoEs that are perceived to be inferior by their intended users tend to be rejected in the presence of alternatives, or accepted only with great reluctances and/or at a deep discount relative to the preferred MoE, which severely undercuts the new MoE’s potential utility as a liquidity enhancer. Conversely, new MoEs that are regarded as superior are often hoarded by their recipients in anticipation of future uncertainties, once again with the effect of undermining much of the rationale for introducing a new MoE in the first place.


Potential implementations for the first two kinds of long-term solutions represent the primary subject matter of this investigation. Any implementation of either kind would necessarily involve steps and likely produce outcomes that are highly reminiscent of policies and outcomes that have been recorded in the much longer and more varied history of that other globally ubiquitous MoE, i.e., money. Accordingly, much of the content that follows investigates possible points of mutual illumination between the history of international monetary arrangements and financial flows and the history and future of Internet addressing and routing -- with likely subjects including:


  1. The quantity theory and other mechanisms for matching the MoE supply to the supply of real economic factors

  2. Inflation, deflation, and stagnation: causes and consequences, preventives and correctives

  3. Currency additions, withdrawals, and migrations: historical successes and failures

  4. Causes and consequences of Gresham’s Law

  5. Defining optimal currency areas

  6. Monetary integration and dis-integration, i.e., establishment and dissolution of shared MoE domains


Although beyond the scope of the current undertaking, investigations of the third alternative are also underway. Interested readers are encouraged to see John Day’s recent book, Patterns of Network Architecture, for more information.