Telephone switchboard operators were very slowly automated

I’ve had a look at a job that was absolutely obliterated by automation: telephone switchboard operators.

This is how a normal day would look like. you’re at a switchboard in the early 20th century, working a panel of jacks, cables, and blinking lights. Each jack links to a phone line — a home, a shop, an office. Someone picks up their phone, a light flashes. You put a cable into their jack, hear them say, “Connect me to the tailor on district xyz,” and find the tailor’s jack, checking a directory if needed. You plug in another cable, send a signal to ring the line, and their voice, turned into an electrical signal, travels through copper wires to the tailor. For long-distance calls, you patch to another operator’s switchboard, chaining circuits across cities. You juggle dozens of calls, plugging and unplugging cables fast to avoid mistakes. One wrong plug sends a call to the wrong person.

In the last century, hundreds of thousands had this job, most of which were women

The tech was simple. Any engineer would see these repetitive, rule-based, tasks as ripe for automation. And indeed, the tech was already there decades earlier, yet it rolled out slowly. In the 1890s, the Strowger switch let callers dial directly. Each digit sent pulses (six for “6”) driving a rotary selector to connect the call without an operator. By the 1920s, these switches spread. In the 1950s, crossbar switches used relay grids to handle multiple calls. By the 1980s, digital switches with microprocessors turned voice into binary, routed over fiber-optic lines. Machines were faster, cheaper. Operator numbers dropped from over 350k in the U.S. in the 1940s to under 100k by 1980, per government data.

Why did automation take decades despite early technology? The Strowger switch, while innovative, was clunky and expensive to scale. Retrofitting existing networks required massive infrastructure investment — new exchanges, wiring, and phones with dials. Rural areas lagged, relying on operators longer due to low call volume and high upgrade costs. Public habits also resisted change; many preferred human operators for complex calls or lacked trust in machines, sounds familiar?

The transition above took nearly a century (1890s - 1980).

So when switchboard operating was no longer needed, many women operators shifted to clerical jobs (typing, filing, customer service etc) using their organization and calm. Some retrained for technical roles like data entry or early programming, where precision fit. But not all landed well. Older workers or those without training faced unemployment or low-wage jobs.

I note a few reflections from this.

• UBI (Universal Basic Income, a social security for everyone without exception) seems like a destiny, as AI can now mass replace white and blue collar jobs like manufacturing and coding.

• Society needs strong principles and values rooted in family and community to prevent collapse.

• People need purpose and jobs. (As seen in several UBI experiments)

Given the above: create more community roles and never automate teaching or caregiving. Protect these industries. Reskill and upskill in emotional and social intelligence, the root of social and economic value. As these jobs reinforce principles and values. Without focus on these, little can prepare governments and societies for the mass unemployment looming across industries. That’s the crux of how I see the shape of things to come. Combine this with collapsing populations, it will be very hard to glue your society to anything but good old fashioned 1984 big brother and a very hot stick to keep things in order.

The good news is there is time to soothe the transition. Even if AI can do knowledge and robotic work now, automation tech took decades to phase out operators, slowed by information asymmetry, adoption, and infrastructure. I think a similar but much shorter curve is happening, no one will use AI when they have a human to hold accountable and even have a morning coffee with. This gives years for governments to prepare through reskilling and policy.

Last reminder to the reader: given this expected time lag, now is a good time to take a bet on something big. So many markets haven’t yet been captured. It seems likely many unicorns will be born these couple of years. Build that vertical agent. Chat with me if you are.