Séance 3 - Chartes, codes, contrats : modes d’adhésion et d’application de la norme (1)
Première publication : 7 juillet 2004 -
Dernière mise à jour : 28 juillet 2005
Pourquoi et comment construire un système de co-régulation de l’Internet ?
Résumé : L’exposé proposera d’abord un rapide tour d’horizon des différents modèles de régulation possibles de l’Internet (I). Il se concentra ensuite sur le modèle de co-régulation, dont on tentera de montrer la logique et la prévalence dans le contexte de l’Internet et au-delà (II). On s’attardera quelque peu aux bases juridiques qui peuvent servir de base à un modèle d’auto- ou de co-régulation aux Etats-Unis, en Europe et en France (III). Enfin, les différents éléments qui concourent à la constitution d’un système de co-régulation seront discutés et mis en cohérence (IV).
Texte proposé à la discussion
Sandvig, C. (forthcoming). Shaping Infrastructure and Innovation on the Internet : The End-to-End Network that isn’t . In D. Guston & D. Sarewitz (eds.), Shaping Science and Technology Policy : The Next Generation of Research. Madison : University of Wisconsin Press. Plus de détails
This chapter approaches the question of how we should best reason about the design of
communication infrastructures by examining a particular debate about the Internet : specifically at issue are the benefits of the Internet for innovation. Some argue that the Internet’s gift is found in an obscure design feature called the "end-to-end argument." This is a network engineering strategy that promotes "stupid" networks : designs where the center lacks intelligence and performs only a few functions, while nodes at the edge of the network - the ends - build complex applications by employing the simple building blocks of the core. This is the Internet (smarter PC, dumber router) as the opposite of the telephone (dumber telephone, smarter switch). Proponents say that with the Internet ?s end-to-end design, experiments can be deployed from the edges (ends) by anyone at all. Success of the Internet can be explained because experiments like the World Wide Web did just that.
On the other side are commercial interests currently deploying intelligence inside the
network’s core ; this logic speeds some traffic over others (caching), blocks traffic (firewalling, filtering), eavesdrops (snooping), and disguises some nodes as others (masquerading). This is so worrying for innovation - and for the freedom of users - that end-to-end proponents have asked the US government to take action to preserve the Internet’s "natural" form. They argue that we may even need a new Internet that retains an end-to-end design if the present one continues to erode.
This chapter takes a third position. Forward from the earliest organized communication
systems such as the horse couriers of the early Chou dynasty, circa 1000 B.C.E., history teaches that networks tend to complexity at the core as more is asked of them. Contrary to the end-to-end argument, there is no reason to think that the Internet will evolve to be faster and more reliable than earlier electronic systems, and in a reversal of three millennia this will require fewer intermediaries and less intelligence at the center.4 While a more complex "middle" is already here, the question at issue has never been the preservation of a simpler network structure but how and where new complexity is implemented. The key to innovation rests somewhere else entirely : the key is not the degree of logic within intermediary nodes, but which nodes we trust. While Moors (2002) elaborates some of the technical implications of this critique, this chapter
addresses the implications for innovation and public policy.
I will address the future Internet by first recalling the oldest data networks - chiefly to
remind us that though the structure of a communication network may have a technical veneer, it is a political bargain. Then, considering the Internet, I will unpack the end-to-end argument and suggest that : (a) it is not an organizing principle, (b) if it is a principle it is probably not true, and (c) if it is true it is probably not useful. The best outcome that normative claims premised on the end-to-end argument can offer us is to produce the right result for the wrong reasons, but we might be even better at promoting innovation if we act for the right reasons. Even worse, a dogmatic belief in end-to-end will simply retard the development of the Internet’s infrastructure by limiting needed improvements in the "middle" or core. I will suggest that the right values to support are transparency and participation. I then conclude with the suggestion that underlyingprinciples that support innovation need to be addressed explicitly, not silently embedded in technical arguments.