The Internet Protocol Journal, Volume 12, No.4

Book Review

Protocol Politics

Protocol Politics: The Globalization of Internet Governance, by Laura DeNardis, MIT Press, 2009, ISBN 978-0-26204257-4.

In Protocol Politics, Dr. Laura DeNardis assembles a variety of stories gleaned from official and unofficial Internet Engineering Task Force (IETF) records and firsthand accounts, and supplements them with primer-level descriptions of successive generations of Internet addressing and routing protocols to create a broadly accessible overview of the factors that have shaped the present and evolving state of these most central features of Internet technology.

The author, a former enterprise networking consultant and technology analyst, joined the Yale Law School Information Society Project as a Post-Doctoral Fellow in 2006, and became the Executive Director of the program in late 2008. DeNardis approaches the challenge of organizing these disparate materials by adopting an interpretive framework that highlights the role of power—interpersonal as opposed to electrical—as both the primary input and most important output or consequence of the definition, selection, and implementation of Internet protocols.

The book knits together a wealth of important historical information that has to-date remained largely neglected outside of the technical community. Although DeNardis' choice of framing is perfectly legitimate—and in fact quite common within the academic disciplines that delve into the influence of institutions on industries, economies, and society—in this case it leads her to overreach a bit, and arguably to draw a few prominent conclusions that are not well-supported by the balance of available historical evidence.

Organization of the Book

DeNardis employs this interpretive framework across six densely written chapters, the first four of which directly address the significance of power in a different functional context of relevance to the evolution of Internet addressing and routing. The introductory chapter investigates the significance of scarcity and its effect on protocol resource management and Internet Governance. Here she devotes considerable space to detailing the critical importance of IP addresses as the single element among Internet protocols that is both indispensable and nonsubstitutable. DeNardis' insightful overview of the general characteristics of IP addresses is somewhat marred by her mixing together of some basic, intrinsic functional properties of addressing (for example, identifier and locator functions) with various necessary but extrinsic correlates or consequences of those functional properties (for example, universality, external observability), or with contingent features of current IP address usage conventions (for example, indifference to underlying technologies).

In addition, despite the ostensible focus on scarcity in the chapter, no reference is made to that other, equally essential and quantity-constrained feature of the Internet service landscape—that is, the inherently limited, occasionally overtaxed carrying capacity of Internet routing subsystems, particularly the collectively provisioned interdomain routing system. Overall, Protocol Politics provides almost no exposure to the technical, operational, and economic constraints that define the routing environment, much less to the constraints that those factors impose on number resource distribution arrangements. Chapter One closes with an overview of the priorities that justify and define the sphere of Internet Governance which anticipates many of the concluding observations in the book's final chapter on "Opening Internet Governance." Both chapters acknowledge "technical expertise" only as a source of institutional or political legitimacy, without according any special significance to the content of such expertise, or why it matters at all. Readers of Protocol Politics may thus come away with insufficient appreciation of the fact that before Code can become Law (or anything else), it first must be running code—and that not every wish is translatable into running code. [1]

Piercing the Fog of Protocol War

n the three chapters that follow, DeNardis presents her observations about how power shapes and flows from the definition and selection of Internet protocols. Chapter Two covers the first half of this proposition, focusing on the events that followed the December 1990 IETF meeting where, DeNardis suggests, the twin challenges that would shape the development of Internet addressing intersected with the chief institutional impediment that would ultimately reveal the true political nature of Internet standards development.

The first challenge that she identifies is the foreseeable inadequacy of IPv4 as the exclusive addressing resource pool for a rapidly growing and globalizing Internet. In keeping with the overall theme of the book, the second challenge that DeNardis chooses to highlight is the implicitly political challenge of accommodating greater international participation in the U.S.-centric Internet technical coordination and decision-making bodies. Against this backdrop, DeNardis introduces the other chief protagonist in her story, the International Organization for Standardization (ISO), which backed the rival Open Systems Interconnection (OSI) family of protocols as an alternative, non-TCP/IP-based foundation for the ongoing, global proliferation of data networking. DeNardis details the convoluted, multidimensional deliberations that followed that 1990 IETF meeting, which eventually culminated in 1994 in the formal recognition of IPv6 as "The Next-Generation Internet Protocol."

Chapter Three goes on to explore the implications of both IPv4 and IPv6 for important civil liberties—especially privacy—and how such considerations did and did not, but hypothetically might have, influenced the choice and form of the most important features of TCP/IP.

Chapter Four rounds out the central thesis of the book by illustrating how various national-level considerations—especially government-directed foreign and domestic economic policies—have resulted in an increasingly diverse global pattern of IPv6 adoption. DeNardis' detailed account of the complexities surrounding the IP Next-Generation (IPng) debate and its aftermath incorporates a diverse mix of sources, from pointed remarks made on various mailing lists, to conference presentations and official Internet Architecture Board (IAB) meeting minutes, and represents a major feat of historical scholarship. That said, her presentation of "relevant historical facts" from the 1990–1994 period is by no means complete, nor is her interpretation of the facts that she does cover or the conclusions that she draws from them immune to criticism. For example, in puzzling over possible hidden forces behind the selection of IPv6, DeNardis states that:

"If anything, there was market pressure to adopt an OSI rather than TCP/IP-based protocol. The ISO alternative had the political backing of most Western European governments (sic) influential technology companies, and users invested in OSI protocols, and was even congruent with OSI directives of the United States. The selection of IPv6…" (p. 61)

Although these facts may be beyond dispute, they do not represent the full picture. To give one illustration, in 1989, almost 2 years before the date that DeNardis marks as the start of the IETF's lone struggle against the combined forces of Europe, influential carriers and hardware manufacturers, and the U.S. government, an indigenous movement of European network operators emerged and began self-organizing to facilitate the exchange of TCP/IP-based traffic, contact information, and operational tips, and to discuss best practices in areas of networking where individual network-level decisions could have far-reaching effects on internetwork performance.

That organization would go on to become Réseaux IP Européens Network Coordination Centre (RIPE NCC), the first independent, transnational registry for Internet Protocol number resources, and the institution that would provide the organizational template for the Regional Internet Registries (RIRs) that subsequently sprang up in Asia (APNIC, 1993), North America (ARIN, 1997), Latin America (LACNIC, 2002), and Africa (AFRINIC, 2004). These facts point to a level of active indigenous European support for TCP/IP-based networking that would seem to be at odds with any suggestion of a continent united in support of OSI against a less-attractive standard being pushed by an insular foreign organization.

Thus, regardless of whether DeNardis' concerns about institutions and power relations are well-founded, her intuitions about the division of contestants in the great protocol power struggle clearly are not. 2]

Market Contrast

Another question that DeNardis raises, obliquely but repeatedly, relates to the possibility of "free markets" as an alternative mechanism for defining, selecting, and distributing Internet protocols and the virtual resources that they create.

In no less than a dozen separate passages scattered across each of the chapters in the book, DeNardis sharply contrasts a range of IETF and RIR institutional processes to the workings of the "free market." For example, she observes that the value of IP addresses is unknown because they have never been exchanged in free markets (p. 16); that Internet addresses have never been exchanged in free markets (pp. 23, 190); that the privacy potential of Internet technologies is enhanced by selection pressures from free markets (p. 74); that the IETF refused to countenance an IPng protocol selection made by free markets (p. 51); that the selection of IPv6 happened outside the realm of free markets (p. 69); that widespread adoption of IPv6 is impeded by the absence of a free market for protocols (p. 137); that IETF philosophy holds that it would be inappropriate to exchange protocol resources in free markets (pp. 163, 183–184); that the Internet Assigned Numbers Authority (IANA) refused to relinquish IP addresses to free markets (pp. 163, 164); that traditional opposition to the exchange of protocol resources in free markets fortified and centralized the IETF's institutional control (p. 184); and that exchanging IPv4 in free markets has pragmatic appeal, if only as a temporary stopgap (p. 228), although such exchanges might have unintended consequences (p. 229).

Given this frequency of repetition, it is impossible to avoid forming a strong impression of DeNardis' underlying opinion about the intrinsic merits of "free markets" as compared to the seemingly market-antithetical goals and practices of the IETF and the other TCP/IP-centric standards-setting and technical coordination bodies. However, even if one stipulates that "free markets" would by definition represent a superior alternative to the enumerated protocol design and distribution mechanisms, DeNardis never provides any clear indication of where a model for such "free markets" might be found—whether in Europe, the United States, or anywhere else, now or anytime in the past.

Even her own description of that fateful moment in networking history when IPv6 was selected clearly suggests that the alternative to the IETF process that ultimately prevailed was itself neither "free" nor especially market-like:

"…congruent with OSI directives of the United States. The selection of IPv6, an expansion of the prevailing IPv4 protocol over such a politically sanctioned OSI alternative solidified and extended the position of the Internet's traditional standards-setting establishment as the entity responsible for the Internet's architectural direction." (p. 61, emphasis added).

Arguably, the non-inclusion of a pure "free market" example is not merely a coincidence, but rather reflects a more fundamental problem inherent in the concept itself. Further, if one grants that the market mechanism that is most free is the one that fosters the broadest participation in those activities that make markets attractive—including openness to participation, exercise of individual choice, competition, accelerated innovation, and wealth creation—then one might interpret the two-plus orders-of-magnitude growth in the number of independent network services providers operating on both sides of the Atlantic since that time as a solid indicator that markets have not suffered too badly from the 1994 decision to extend the lifetime of TCP/IP through IPv6.

Clearly the looming inflection point in IP addressing will provide many irresistible opportunities to revisit that choice in the days ahead. Meanwhile, the question of whether the embrace of an OSI-friendlier IPng by the IETF would have been sufficient to offset the varied negative externalities that might have accompanied such a choice must forever remain unanswered. Would an IETF endorsement have trumped the as-yet incomplete state of OSI standards, as well as OSI's tighter associations with non-standards-based operating systems, proprietary hardware platforms, and the connection-oriented networking technologies favored by then Internet-averse incumbent Public Switched Telephone Network (PSTN) operators? Would that choice alone have created or been likely to foster a freer market, or to have led to a more enthusiastic, widespread embrace of a different post-IPv4 addressing format—or alternately would it have led to the appearance of books like Protocol Politics, albeit written from the opposite perspective, and possibly a decade sooner? Contrary to the popular adage, hindsight is not 20/20, any more than is our vision of where to go from here. [3]

Beyond the Clash of Idealizations

Writing a book review is an inherently risky undertaking, one that is vulnerable to many of the same human biases and errors that have unquestionably informed both the selection and development of various technical standards, just as they have influenced the embrace, rejection, or modification of various market arrangements throughout history.

Even when people (book reviewers, for example) recognize that real-world decisions and their consequences tend to be irreducibly complex—or perhaps precisely because they recognize that complexity—they nevertheless tend to gravitate toward explanatory frameworks and cognitive models that promise to invest their perceptions and choices with the kind of absolute certitude that is very rarely found outside of the physical world (and only infrequently found there).

The problem, of course, is that many such explanatory frameworks can be found to fit quite nicely with the same set of human experiences, even though some of those models may be mutually orthogonal, and some may be quite mutually and actively antagonistic. In this sense, the juxtaposition of pure, frictionless "free markets" alongside the idea of absolutely pure scientific or technical decision making divorced from all other human considerations, while well-calibrated to inflame passions, represents less a contrast of opposites than a rather less illuminating pairing of two deeply unrealistic ideal types. Distilling a book as rich and informative as Protocol Politics down to one possible review-sized essence is much easier to accomplish from just such a privileged vantage point, and no doubt this particular review suffers from the all-too-predictable effects described herein. However, with that caveat firmly established, a few more things about Protocol Politics deserve to be mentioned here.

First, Protocol Politics is an important book. It is the well-written and informative, and is the first to be written for a general audience that draws on the right historical sources (or at least most of the right ones that remain accessible) to cover this critical period in the development of the Internet's core addressing and routing protocols. Even those who are least likely to be sympathetic to its findings are likely to find Protocol Politics to be a thoughtful and engaging read.

Second, IPJ readers and other technologists should not dismiss the inherently political, power-oriented framework that DeNardis employs in Protocol Politics. In general, the most honest and effective response to an assertion of systemic political or institutional bias is not to claim an equally absolute, otherworldy detachment from the affairs of man, but rather to remind the critic that in a world where all institutions are regarded as manifestations of somebody's will to power, specific targeted criticisms based solely on that fact lose all coherence. Would-be institutional critics who espouse such views thus have no choice but to make a positive argument as to which arrangement, among all of the equally power-tainted institutional arrangements that are possible, should be regarded as the preferable outcome, for whom, and why. Judged in this light, this reviewer feels that "the IETF way" still stands up pretty well, foibles and all. There is always room for improvement, but just as in matters of code, a concrete proposal for improvement is worth a thousand critiques of the past.

Finally, the careful reader may notice a pattern within this review, one composed of points highlighted here even though they may not be equally central to the story presented in Protocol Politics (for example, about the role of technical expertise in Internet governance, the dynamic limitations of routing system carrying capacity, the possibility of free market alternatives to current Internet address distribution arrangements, and so on).

Each of these points merits special attention because taken together they help to illuminate the existence of an identical set of critiques that have reappeared periodically in the course of another, much older (actually, centuries-old) debate that parallels the as-yet unresolved debates outlined by DeNardis in Protocol Politics.

In both instances, the question at issue involves the relative merits of nonmarket, technical expert-based systems as a means of managing resources that are uniquely central to economic growth, and for mitigating the systemic risks that can threaten that growth. In that other debate, arguments in favor of pure free market solutions have generally been dismissed as extreme and unrealistic for more than a century, ever since the last real-world implementation of such a system finally succumbed to its own chronic instabilities and was replaced by a nonmarket coordination arrangement. More recently, however, a resurgence of extreme turmoil in that parallel industry has undermined belief in expert management, if not in the underlying "hard realities" that were supposed to constitute the managers' technical domain of expertise. In turn this turmoil has sparked renewed interest in the long-marginalized pure free market proposals, as well as in alternative remedies involving much tighter industry control by nonmarket authorities.

How the current chapter in either of these parallel stories will play out remains to be written. However, those who are eager to anticipate the kind of language that is likely to play a central role in both outcomes will find that a close reading of Protocol Politics provides a wealth of possibilities to consider, and more than a few to keep one up at night.

—Tom Vest, Consultant
tvest@eyeconomics.com

References

[1] DeNardis makes several references to the idea that Code is Law, which was first articulated by Larry Lessig in Code and Other Laws of Cyberspace (1999) [Editor's note: Code was reviewed in IPJ Volume 11, No. 3]. Here the phrase is juxtaposed with David Clark's famous paean to "rough consensus and running code," which DeNardis describes as an "articulation of the IETF's core philosophy" (p. 47), and amended with a paraphrasing of an early (c. 1992) observation made by Marshall Rose about a common problem encountered when attempting to implement code to satisfy a non-operationally developed standard. The original staying was, "The problems of the real world are remarkably resilient to administrative fiat."

[2] Several formerly obscure insights on the events of this period were recently illuminated by RIPE co-founders Rob Blokzijl and Daniel Karrenberg, during RIPE's 20th Anniversary Commemoration at the RIPE 58 meeting in Amsterdam (May 2009). Some of these are available at:
http://www.ripe.net/ripe/meetings/ripe-58/content/presentations/Blokzijl-RIPE-20-years.pdf
and
http://www.ripe.net/ripe/meetings/ripe-58/content/presentations/the-origins-of-ripe.pdf

[3] Those wishing to investigate these questions further may benefit substantially from yet another unique historical resource that has recently been made available online. Thanks to the Charles Babbage Institute and the Institute of Technology at the University of Minnesota, the entire ten-year archive of ConneXions—The Interoperability Report (1987–1996) is now available online at: http://www.cbi.umn.edu/hostedpublications/Connexions/index.html

In keeping with its mandate to track the interoperability of emerging network technologies, ConneXions published more than sixty substantive articles on OSI and GOSIP during the period leading up to and following the IPng debates recounted in Protocol Politics.

Read Any Good Books Lately?

Then why not share your thoughts with the readers of IPJ? We accept reviews of new titles, as well as some of the "networking classics." In some cases, we may be able to get a publisher to send you a book for review if you don't have access to it. Contact us at ipj@cisco.com for more information.