The network stacks conceived in the 1970s and 1980s by teams led by Vint Cerf (TCP/IP) and Charlie Bachmann (OSI) (among many others) were designed to solve complex but very specific transmission and communication problems. The big idea was not to disrupt modern geopolitics. However, for The Stack, we recognize them to represent a more universal topography and geographic machine, one that we may come to see as having real effects of a similar order of magnitude as the loop topologies ratified at Westphalia. It must be said that for a book of design theory, The Stack platform ’ s logical structure is far more important than the technical details of existing networks, but one provides for the other. As indicated, however, a crucial consideration for these models was massively distributed infrastructural modularity as a way to accommodate unplannable new demands and new machines within discrete layers as the key to future growth of the system. For this they exemplify platforms far more than master plans, and they are direct technologies for governance at least as much as tools of governments. At the same time, the armature for that modularity guarantees the predictability of these inclusions and the scalability of the whole, and so as for any platform, the governing regularity of stack protocols is necessarily inflexible and regular. What is simplest and most fixed provides for Hermes ’ s ephemeral work of transit and translation.
All of this is was and is highly contingent. It was quite possible that neither OSI nor TCP/IP would become anything like a central systems mechanism for global communications. The standards wars of this era divided phone companies, which preferred a system that would support discrete circuits between one sender and receiver, like older telephony networks, versus many computing companies, such as IBM, which lobbied hard for packet switching technologies that could treat all messages (e.g., voice, data, image) as recombinant bits flowing over whatever future hardware that could connect with the network. The models of communication (equally technical and social) posed by both options contain profound downstream implications for the geopolitics of an information society. A polity of circuits and a polity of packets are in epistemological and functional opposition. For the circuit model, its stack is a bounded utility for which use is metered by monopolistic caretakers who, by guaranteeing the circuit between sender and receiver, retain de facto sovereignty over the channel. For the packet switching model, at least in the minds of Cerf ’ s group, the platform would prioritize the edges of the network, asking them to do more of the work to reassemble transmitted packets and calculate the content of messages. Cerf ’ s group presumed increasing computerization of networks from node to edge, but even more than its authors intended, their model would depend on the exponential growth of the computational capacity of all network equipment, soup to nuts (e.g., terminals, routers, servers), in order to bring the network to life at a global scale. The greater the computational carrying capacity of the entire network apparatus, the more information it can convey with greater speed, and the more information it conveys, the more demand for capacity by organizations increasingly reliant on its competitive efficiencies. In this way, Moore ’ s law is not just the cause of the software society; it is also a contingent effect of a platform architecture design decision.
62
Platform and Stack, Model and Machine