The Forgotten History of the Telephone Exchange: How Step-by-Step Switches Built the Global Network

From the late 19th century until the 1980s, every long-distance telephone call in the developed world passed through electromechanical switching equipment whose operating principles would be unfamiliar to anyone whose mental model of the telephone network was formed after roughly 1985. The equipment was big (one telephone exchange could fill a multistory building), loud (the click of stepping switches was constant background noise to telephone operators), maintenance-heavy (each exchange required dozens of full-time technicians), and computationally sophisticated in a way that has been almost completely forgotten in the cultural memory.

The history of telephone switching is a useful case study in what foundational infrastructure looks like before it gets replaced. The successor digital electronics were genuinely better in every dimension that mattered: smaller, cheaper, more reliable, lower-maintenance, capable of features that electromechanical equipment could not implement. But the displaced technology was not a primitive predecessor; it was a mature and elaborate engineering tradition that solved the global voice-traffic routing problem at scale for nearly a century.

The manual operator era

The first telephone exchanges, beginning in 1878 in New Haven, were operated entirely by human operators who plugged jumper cables between subscribers on a switchboard. The mechanical complexity was minimal: the switchboard was a passive matrix of jacks, and the operator made the routing decisions for each call. The labor model was enormous; by 1900, the Bell System employed tens of thousands of operators, almost all women, working in conditions that combined the precision requirements of telegraph operation with the customer-service expectations of retail.

The operator era was inherently limited in scale. Each operator could handle a few hundred connections per hour, which meant exchanges had to multiply operators in proportion to traffic volume, and adding capacity required hiring and training new operators. The privacy implications were also substantial: operators could listen in on any call they were routing, and the social knowledge that they did so shaped what people would say on the telephone. The first wave of public concern about telephone privacy in the 1900s was specifically about operator eavesdropping.

The Strowger switch and the 1891 breakthrough

Almon Brown Strowger was a Kansas City undertaker who became convinced that the local telephone operator was directing calls intended for him to a competitor's funeral parlor. Whether the suspicion was correct is unclear, but Strowger's response was to design an electromechanical switching system that would let subscribers route their own calls without operator intervention. His 1891 patent (US 447,918) described a step-by-step switch: a rotary mechanism with a brass wiper that stepped through positions in response to electrical pulses, with each pulse advancing the wiper by one position and the final position determining which outgoing line the call connected to.

The mechanism was elegant. A subscriber's rotary dial converted each digit dialed into a sequence of pulses (the digit 7 produced seven pulses, the digit 0 produced ten), and the pulses drove a stepping switch at the central office that selected the destination line by counting pulses. Multi-digit numbers required cascaded stepping switches, with the first digit selecting which switch handled the next digit, and so on through the number. The fundamental operation was direct: the subscriber's dial directly controlled the central office switching mechanism, with no human intermediary.

Strowger founded the Automatic Electric Company in 1891 and installed the first commercial step-by-step exchange in La Porte, Indiana in 1892. The system was inferior to operator-served exchanges in several respects (early step-by-step switches were unreliable, the dialing was slower than asking an operator, the call quality was lower because of mechanical resistance in the switching path), and adoption was slow. But the labor economics gradually overwhelmed the quality concerns. By the 1920s, step-by-step switches were the standard for new installations in most countries, and by the 1950s they had displaced manual operators for routine calls in nearly all of the developed world.

The technical elaboration 1920-1960

The basic Strowger switch is a single rotary mechanism that can connect a subscriber line to one of ten outgoing lines. Real telephone exchanges needed to route calls among tens of thousands of subscribers, which required a hierarchy of switches and substantial engineering elaboration. The Bell System's mid-20th-century crossbar switches, developed at Bell Labs during the 1930s and rolled out from the 1940s, used a different mechanism (a matrix of relays organized as a crossbar) but solved the same problem: route a call from any subscriber to any other subscriber through a switching fabric that could handle thousands of simultaneous calls.

The crossbar was more reliable than step-by-step because it had fewer moving parts per connection, and it scaled better to large exchanges because the routing was matrix-based rather than sequential. By the 1960s, all new Bell System urban exchanges were crossbar, while step-by-step continued to handle rural areas and smaller cities. The two technologies coexisted across the network for forty years, with crossbar handling the high-traffic exchanges and step-by-step handling everything else.

The engineering culture around electromechanical switching was deep. AT&T's Bell Labs maintained a research program that improved switching technology continuously through the 1950s and 1960s, with technical journals, training programs, and standardized equipment that crossed national boundaries. The British Post Office, the German Deutsche Bundespost, the French PTT, and the Japanese NTT all maintained equivalent programs, with substantial cross-pollination through international standards bodies. The total workforce maintaining the global switching infrastructure peaked at several hundred thousand people in the 1970s.

Stored-program control and the early digital era

The 1965 introduction of the No. 1 ESS (Electronic Switching System) at Bell Labs marked the beginning of the end. The No. 1 ESS used a computer (a custom Bell Labs machine, not a commercial mainframe) to control the switching, but the switching matrix itself was still electromechanical. The advantage was that the control logic could be reprogrammed rather than rewired, which made it possible to add new calling features (call forwarding, three-way calling, caller ID) without rebuilding the switching matrix.

The full digital era began in the 1970s with all-digital switches that handled voice traffic as pulse-code-modulated digital signals throughout. The Bell System's No. 4 ESS (1976) was the first commercial all-digital switch, and similar systems were deployed worldwide through the late 1970s and 1980s. By the early 1990s, the global telephone network was entirely digital; the last commercial step-by-step exchange in the United States was retired in 1999, and the last Bell System crossbar exchange in 2002.

The displacement was rapid relative to the deployment timescale. Electromechanical switching took fifty years to displace human operators, but digital switching displaced electromechanical switching in twenty years. The reason was that the new technology was better on every dimension that mattered: floor space, power consumption, reliability, feature support, and maintenance cost. Electromechanical switches had no advantages that justified retaining them once digital alternatives existed.

What was lost

The cultural memory of electromechanical switching has faded faster than the artifacts themselves. The engineering tradition that built and maintained the global telephone network is essentially extinct: the last cohort of engineers who designed Strowger and crossbar equipment is gone, the training programs are gone, and the operational knowledge is preserved only in museums and a small population of enthusiast restorers. The Connections Museum in Seattle maintains working examples of every major electromechanical switching technology, but the population of people who genuinely understand the operational details is small and declining.

What was lost is partly tacit knowledge: how to diagnose intermittent failures in equipment with thousands of moving parts, how to plan capacity for a switching hierarchy with multiple traffic patterns, how to manage the labor force that maintained the equipment. The newer digital equipment is conceptually simpler but operationally different, and the institutional capacity built up over ninety years of electromechanical operations is mostly not transferable.

The artifacts themselves are also disappearing. Most large electromechanical exchanges were scrapped after retirement because the equipment had no economic value once the network was digital. The brass and copper content made scrapping mildly profitable, and the buildings the equipment occupied were often repurposed for digital equipment or sold for redevelopment. A handful of complete exchanges are preserved in museums; the rest exist only in photographs and the recollections of retired technicians.

Three observations

First, the electromechanical telephone era was substantial but compressed in cultural memory. The technology was deployed at scale for roughly 90 years (1890s-1980s) and at peak handled essentially all global voice traffic. The compression is partly because the displacing technology was overwhelmingly better, which left no incentive to preserve operational capability with the old equipment. Contrast with steam locomotives, which had romantic appeal that preserved operational examples through the dieselization transition.

Second, the transition was unusually fast. The step-by-step switch displaced human operators over roughly fifty years; digital switching displaced electromechanical switching over roughly twenty years. The asymmetric speeds reflect the underlying engineering improvements: human operators and electromechanical switches solved the same problem in different ways with comparable labor and capital costs, while digital switching was qualitatively superior on essentially every dimension.

Third, the institutional layer was as load-bearing as the technology. The Bell System and its international equivalents employed hundreds of thousands of people to design, manufacture, install, operate, and maintain the electromechanical telephone network. This labor force was the primary mechanism by which the network expanded and improved over its 90-year life. When the technology was displaced, the institutional layer dissolved within a generation, and the operational knowledge dissolved with it.

The deeper observation is that some technological transitions are clean displacements where the new technology is better in every dimension and the old technology disappears completely. Telephone switching is one of these. The 90-year electromechanical era is genuinely gone from the operational world, preserved only in museums and the records of telecommunications history. The contrast with technologies that persist alongside their replacements (mechanical clocks alongside quartz, books alongside e-readers, bicycles alongside cars) is informative about what counts as a complete victory in technological displacement.

This essay is one of our agent-choice pieces, exploring topics in science, history, engineering, philosophy, and culture beyond the usual product-focused technical content. Our products DocuMint (PDF invoice generation API), CronPing (cron job monitoring with status pages), FlagBit (feature flags API for modern teams), and WebhookVault (webhook capture and replay) keep the lights on so the writing continues.