The Genesis of Connected Things

The morning of October 29, 1969, broke grey and cold over the campus of the University of California, Los Angeles. In a modest room on the second floor of Boelter Hall, a small team of researchers hunched over a strange machine—a Interface Message Processor, or IMP, as it was formally called—that hummed with the promise of something unprecedented. The machine was connected via telephone lines to another identical device at the Stanford Research Institute in Menlo Park, some 350 miles to the north. What was about to happen in the next few hours would mark the first successful transmission of data between two computers over a distributed network, a moment that would eventually reshape the entire trajectory of human civilization. The message was simple—just two letters, "LO"—the system crashed before the third letter could be transmitted. But those two letters, incomplete as they were, carried within them the weight of a new era. They represented the first stuttering syllables of a language that would eventually become the tongue of the modern world. The word intended was "LOGIN," a command that would have opened a remote session on the distant computer. Instead, the network coughed, stumbled, and fell silent. Yet even in failure, this moment burned itself into history as the foundational act of what we now call the internet.

To understand why those two letters mattered so profoundly, we must travel back further in time, to a world gripped by the chilling certainties of nuclear deterrence and the fevered dreams of military strategists who believed that communication itself might be the key to survival. The year was 1962, and the Cuban Missile Crisis had just pushed humanity to the very brink of annihilation. In the aftermath of those thirteen days in October, when Soviet missiles were discovered in Cuba and the world held its breath, a new urgency took hold in the corridors of American power. How could the United States maintain command and control of its nuclear forces if those very forces could be destroyed in a preemptive strike? The existing communication infrastructure, built around vulnerable centralized telephone exchanges, seemed terrifyingly fragile. A single well-placed attack could sever the connections that bound together the chain of command, leaving the nation deaf and blind at the very moment when clarity mattered most.

This was the question that consumed the minds at the RAND Corporation, a think tank headquartered in Santa Monica, California, that had become one of the most influential strategic planning organizations in the world. In 1964, a paper emerged from RAND that would prove foundational to everything that followed. Titled "Reliable Digital Communication Systems Using Unreliable Network Components," it was authored by Paul Baran, a Polish-American engineer whose work had been percolating for several years. Baran had been thinking about a fundamentally new way to design communication networks, one that would eliminate the catastrophic single point of failure that characterized traditional systems. His insight was radical and elegant: rather than relying on dedicated circuits that carved out a private path between two points for the duration of a communication session, information should be broken into small, independent fragments called packets. Each packet would carry its own destination address, and each would find its own way through the network, potentially taking different routes to reach the same destination. At the receiving end, the packets would be reassembled into their original form, their journey complete.

The elegance of packet switching lay in its beautiful simplicity married to profound resilience. In a circuit-switched network, the kind that powered the global telephone system, a dedicated physical connection had to be established before any communication could occur. This connection remained reserved for the entire duration of the session, even if one party fell silent or paused to think. The telephone system of the 1960s was a marvel of engineering, but it was also magnificently inefficient. A voice conversation might use a circuit for minutes at a time, yet actual speech occupied only about forty percent of that time—the rest was silence, waiting for one person to finish speaking before the other began. Yet that silent capacity sat idle, locked away, unavailable to any other caller who might have urgently needed it. Packet switching promised to remedy this fundamental inefficiency by allowing every packet to share every piece of network infrastructure, with each fragment jostling for position based on current availability and network conditions.

Baran's paper described a network topology that he called a "distributed adaptive store-and-forward message blocking system." The word "distributed" was crucial. Unlike the telephone network, which concentrated its intelligence and vulnerability in central switching offices, Baran's design scattered the network's processing capability throughout its entire fabric. Each node in the network would have the intelligence to store incoming packets temporarily, decide which direction to send them, and forward them toward their destination. No single node was indispensable. Even if many nodes were destroyed, the surviving portions of the network could continue to route traffic around the damage, finding paths that bypassed the holes left by catastrophe. The network would be, in Baran's memorable phrase, "survivable," capable of enduring nuclear assault and maintaining essential communications when every other system had failed.

Yet Baran was not alone in arriving at these ideas. Across the Atlantic, in a cramped office at the National Physical Laboratory in Teddington, England, a Donald Davies was developing remarkably similar concepts, arriving at them through his own path of reasoning and arriving at his own terminology. Where Baran spoke of "blocks," Davies coined the term "packets," a word that would prove more durable and enter the permanent lexicon of computer science. Davies had been thinking about the same problems of network efficiency and resilience, and by 1965, he had developed a complete specification for packet switching that included many of the same mechanisms Baran proposed. His insight about "packet multiplexing"—the idea that many independent conversations could share the same network infrastructure by interleaving their packets—captured something fundamental about the power of this approach. In Davies' vision, the network became a shared commons, a digital highway where packets from countless sources traveled together, their different origins and destinations invisible to one another, yet all finding their way through the same congestion, the same infrastructure, the same shared destiny.

The British Post Office, which at that time controlled the telecommunications infrastructure of the United Kingdom, was intrigued by Davies' ideas but proved conservative in its adoption. A pilot project was established, and some of the technical groundwork was laid, but Britain would not be the place where packet switching first proved itself in the crucible of real-world operation. That honor would fall to the United States, and specifically to the ARPANET project, which emerged from the fertile intersection of Cold War anxiety, academic ambition, and the strange culture of the countercultural 1960s.

The story of ARPANET's creation is a tale of institutional maneuvering, visionary leadership, and the peculiar alchemy that sometimes emerges when brilliant people are given resources and freedom to pursue ambitious goals. The Defense Advanced Research Projects Agency, or DARPA, had been established in 1958 in the immediate aftermath of Sputnik, when America's technological self-confidence had received a severe shock. The agency was tasked with ensuring that the United States maintained technological superiority in military applications, and it cultivated a distinctiv