In late 1991, roughly 15 researchers in the Systems Group at the University of Cambridge Computer Laboratory shared one coffee machine. It stood in the corridor outside a workspace known as the Trojan Room.
That was convenient for the people inside the room. It was less convenient for colleagues working two or three flights of stairs away, who could make the journey only to discover that someone closer had emptied the pot.
According to Quentin Stafford-Fraser’s retrospective account of the project, the solution began with a few pieces of spare laboratory equipment: a video camera, a frame-grabber and an old 680×0 VME-based computer.
They pointed the camera at the coffee pot.
A small problem worth solving
The coffee machine was not important in any grand technological sense. That was precisely what made the project so recognisable. Nobody wanted to spend several minutes crossing the building, navigating the stairs and returning empty-handed because the latest pot had already disappeared.
It was a recurring irritation with a simple yes-or-no question at its centre: was there coffee available right now?
Stafford-Fraser and his colleague Paul Jardetzky realised that the researchers already had everything needed to answer it remotely. The camera was fixed to a laboratory retort stand. Its cables ran beneath the floor to the computer and frame-grabber in the Trojan Room.
What XCoffee actually was
The software came in two parts. As Stafford-Fraser explained in the project’s original non-technical biography, Jardetzky wrote the server program that captured images of the pot. Stafford-Fraser wrote the client application, called XCoffee, that researchers could run on their workstations.
The client displayed a tiny grayscale picture in a corner of the screen. The server could capture images every few seconds at different resolutions, although the image seen by users was initially updated only about three times a minute.
That was enough. Coffee does not vanish frame by frame. Researchers only needed to distinguish an empty pot from one that was full, filling or worth walking towards.
The internal system used custom networking software rather than the World Wide Web. It ran across the laboratory’s own network stack and required users to have the specialised XCoffee client.
Calling the 1991 installation a webcam therefore compresses two related but distinct moments into one. Stafford-Fraser and Jardetzky created the camera-monitoring system. The system became a true web camera two years later.
The moment it reached the web
By 1993, graphical web browsers had made images a practical part of web pages. Daniel Gordon and Martyn Johnson revived the coffee-camera project with newer equipment, and Gordon modified its server so that it could respond to HTTP requests.
Instead of running XCoffee, a person could now open a web page and request the latest image through an ordinary browser. That change made the coffee pot visible far beyond the Cambridge laboratory.
The later public service used a video-capture board fitted to an Acorn Archimedes. Cambridge’s surviving technical description of the web service says the board grabbed one frame every second, encoded it as a JPEG and supplied it to the web server when requested.
The picture generally associated with the coffee cam was only 128 pixels wide and 128 pixels tall. It was grayscale, grainy and barely animated. What mattered was not its visual quality but the fact that somebody elsewhere could look at an object in Cambridge and see its current state.
Why such a modest tool made sense
The project followed a pattern that still shapes workplace technology: a minor inconvenience happens repeatedly, somebody notices how much attention it consumes, and a tool is built to make the state of things visible.
A company-sponsored 2025 Dayforce survey conducted by Hanover Research questioned 6,178 workers, managers and executives across six countries. It reported that 88 percent encountered friction in their working environment, while 69 percent said their organisations used too many technology platforms.
The Cambridge researchers faced a much smaller version of the same design problem. Information existed — the pot was either empty or it was not — but obtaining it required an unnecessary journey.
The coffee cam moved that information to the researchers instead.
From internal joke to early-web landmark
Once the image became publicly accessible, the Trojan Room coffee pot developed an audience that nobody involved had planned for.
People around the world began checking the level of coffee inside an ordinary university corridor. The camera attracted press coverage, appeared in broadcasts and became one of the early web’s best-known novelties.
By 2001, contemporary reporting put its audience at 2.4 million visitors. American tourists reportedly contacted Cambridge’s tourist-information office to ask where they could see the pot in person.
Its appeal was partly the absurdity. The growing World Wide Web promised access to information across continents, and one of its first famous live images showed whether a group of computer scientists had remembered to make more coffee.
What the coffee pot started
Web cameras soon appeared in offices, bedrooms, aquariums, city centres and remote landscapes. Jennifer Ringley launched JenniCam in 1996, continuously publishing images from her apartment for seven years. Traffic cameras, wildlife cameras and weather cameras followed the same basic pattern.
The Trojan Room project did not directly invent modern video calls, security cameras or livestreaming platforms. Those systems developed from broader histories of digital video, networking and telecommunications.
What it demonstrated was a simple and durable web format: point a camera at something, publish updated images at a stable address and allow distant viewers to check what is happening.
The habit of glancing
The camera could not guarantee that coffee would still be available by the time somebody reached the corridor. Another researcher could arrive first and empty the pot between the latest image and the journey downstairs.
It did, however, reduce uncertainty. Checking the image became a small part of the laboratory’s working rhythm: glance at the corner of the screen, judge the level in the pot and decide whether the trip was worthwhile.
That rhythm is familiar in modern offices. Status indicators, delivery trackers, shared calendars and monitoring dashboards all allow people to check the condition of something before acting.
The coffee cam’s achievement was not that it recovered dramatic amounts of working time. It turned an invisible state into a visible one.
The camera’s final image
In 2001, the Computer Laboratory prepared to leave its old central Cambridge premises for the William Gates Building in West Cambridge. The Trojan Room would not survive the move, and the ageing camera system had become increasingly difficult to maintain.
The feed was switched off at 10:54 UK local time — 09:54 UTC — on 22 August 2001. Cambridge’s record of the shutdown says the final frame showed the fingers of Daniel Gordon, Martyn Johnson and Quentin Stafford-Fraser pressing the power switch on the Acorn Archimedes.
That machine had captured the public images for seven years and nine months.
The surviving Krups coffee machine did not remain in the old laboratory. It is now displayed at the Heinz Nixdorf MuseumsForum in Paderborn, Germany, where it appears in an exhibition on the history of the internet.
128×128 pixels, one ordinary object
The original story survives because the purpose of the system is immediately understandable. The researchers were not trying to build a global broadcasting platform. They were trying to avoid walking two or three flights of stairs for an empty pot.
Stafford-Fraser and Jardetzky built the internal answer. Gordon and Johnson made it accessible through the web. Together, those two stages turned a minor laboratory convenience into the first webcam.
The last image was not an empty table or a disconnected camera. It was a few fingers reaching towards a switch: one final grainy frame before the coffee pot disappeared from the web.
