The Internet: New and Larger Network Effects
Digitalization put an end to the monopoly of the Bell System. For the Bell System, disruption did not come from the use of computers to better run the telecommunications networks, nor from the new data processing services to be open for competition. Instead, disruption came from the new network structures that made the historic centralized network of the Bell System obsolete and from the larger network effects of a network that could connect both humans and computers.
In 1960, a psychologist turned computer scientist named J.C.R. Licklider envisioned the creation of a network of computers connected by telecommunications infrastructure in his seminal paper entitled “Man-Computer Symbiosis.” In it, Licklider stated: “The picture readily enlarges itself into a network of such centers, connected to one another by wide-band communication lines and to individual users by leased-wire services. In such a system, the speed of the computers would be balanced, and the cost of the gigantic memories and the sophisticated programs would be divided by the number of users.”18
Licklider’s contribution was not only theoretical. In 1962 he was appointed head of the Information Processing Techniques Office at the United States Department of Defense Advanced Research Projects Agency, better known as ARPA.19 ARPA wanted to save resources by pooling together all the computing capacity of the different projects financed by the office across universities and research centers, meaning that researchers at one university could remotely use the computer in another university.
One of the options for ARPA was to create a centralized network connecting computers, following the traditional structure of the telephone network. It was proposed to set up the switching point in Omaha, Nebraska, near the geographic center of the US,20 where the Air Force was already centralizing communications.
Licklider’s ARPA had a better idea: an innovative network model developed at the RAND Corporation by Paul Baran for the US Air Force and described in the paper entitled On Distributed Communications.1' Instead of centralizing switching in a single node (centralized network), or even in a reduced number of nodes (decentralized network), Baran proposed establishing redundant links between the different points to be connected, in such a way that communications would be handled from point to point without being switched in a centralized node. In other words, every point in the network would have switching capabilities; every point would be a small node. The US Air Force had identified that the centralized network of the Bell System would collapse in case of a nuclear attack, and Baran was commissioned to identify an alternative to solve this problem. The distributed network was his solution.
The new ARPA network would still have another fundamental feature that Baran had envisaged: packet switching. Inspired by road transportation (another distributed network), a message would be divided into small blocks (packets), which would be independently transported across the distributed network, to be reconstructed at the destination using computing techniques.
Baran spent five years unsuccessfully trying to convince the Bell System to implement a packet-switched distributed network. His proposal was not only dismissed, but he was put though a training course in which 94 speakers would explain how a telephone network works, so he could understand why his idea was not feasible. He was told how a centralized network worked; how messages could not be broken down into packets, but could only be transported by opening a dedicated circuit making full use of the capacity of the circuit for the specific communication. Nevertheless, Baran was not convinced.22
The network financed by ARPA, called ARPANET, would prove engineers in the Bell System wrong. On October 29, 1969, the first connection of ARPANET, between UCLA and Stanford, started to operate.23 This was the beginning of the internet, which would grow from a small network of computers connecting US universities to a network connecting basically everyone, everything, and everywhere.
The engineers of the Bell System were blinded by the nature of their network. A fully standardized and centralized network developed for voice transmission could not be easily transformed into a distributed network connecting computers owned by different institutions. There was no common operating system and no common standard for the computers to communicate with each other.
Had the engineers been more sensitive to network effects, as Vail had been decades earlier, they would have understood that a network connecting more points (not only humans to humans but also humans to computers and computers to each other) would be more valuable than a mere telephone network. Vail understood the power of converging the telephone and telegraph networks. His successors did not understand the power of converging voice and data networks.
Still, the challenges were enormous. Each university had developed its own communications system to connect their computers, so the universities were not interested in changing their systems to adopt a new common standard. The solution was to keep each independent network working under its own rules and to create a protocol on top of it so that different networks could communicate with each other. The solution was to create a network of networks.
A specific, simple protocol was developed so that the different networks could interact. The internet protocol (TCP/IP) created a new layer on top of the different networks that allows the interaction of previously isolated computer networks. It would not be necessary to fully standardize the pre-existing networks in order for them to interoperate. Each network would use its own standards, but they would all be able to interoperate thanks to a new standard on top of the legacy local standards. The prerequisite would be for the networks to use the TCP/IP protocol.
The internet would develop as a distributed network of networks. No central authority would run a non-existent central node. Eventually, everyone - universities, private companies, and even telephone carriers - was invited to join the network of networks. The larger the number of participating networks, the larger the value for them and for other networks. The internet would become the ultimate example of the power of network effects.
But before the internet outgrew the telephone network, digitalization would kill the telephone monopoly. Transistors empowered alternative communications systems such as microwave networks. Computers increased demand for point-to-point private services. Such services were not served by the regular centralized network that was the backbone of the Bell System, but by specific infrastructure, dedicated lines, provided by the Bell System at a high price. Alternatives started to flourish.