Blog
Why Cambridge Put a Coffee Pot on the Web
How a shared Cambridge coffee pot led to XCoffee in 1991 and a Web camera in 1993, with the original desktop viewer and the people behind it.
Quick Answer
Cambridge researchers pointed a camera at their shared coffee pot so they could check whether coffee was left before walking to it. The local XCoffee system began in late 1991; its images reached the World Wide Web in November 1993.
On This Page7 Sections
A Shared Pot and an Uneven Walk
In his original coffee-pot biography, Quentin Stafford-Fraser describes roughly fifteen researchers sharing a filter machine near the Trojan Room. Some worked several flights of stairs away. By the time they reached the machine, colleagues nearer the pot might have emptied it.
They mounted a camera on a laboratory retort stand and ran wires under the floor to a video frame-grabber attached to an unused test computer.
Paul Jardetzky wrote the server program and Stafford-Fraser the client that displayed the camera view on colleagues' screens. XCoffee showed a small grayscale image, updated roughly three times a minute, or one image every 20 seconds. That let someone check the pot from their desk before setting off.
From XCoffee to the World Wide Web
Stafford-Fraser's timeline records the local system and Web connection separately. Dan Gordon modified the server for Web requests, and Martyn Johnson connected the system to the Web.
| When | What changed |
|---|---|
| Late 1991 | A purpose-built local application let colleagues see the pot |
| November 1993 | The image became accessible through the World Wide Web |
| August 22, 2001 | The Cambridge camera was switched off |
The audience changed with the Web connection. A colleague wanted to know whether to fetch coffee; a stranger elsewhere could look into an ordinary corner of Cambridge simply out of curiosity.
The Clever Step Was Changing What a Request Returned
A browser could request an updated camera image without the group's special client. In his 2001 Communications of the ACM account, Stafford-Fraser explains that an image request did not have to return the same static picture every time. The server could supply a new view from the camera.
What the Cambridge Service Showed
The later Cambridge service page describes a capture board in an Acorn Archimedes and JPEG-encoded images. It also mentions lighting the pot at night. Those details describe the documented later service, rather than every stage of the system that began in 1991.
Someone still had to brew the next pot and share it. For the practical work of serving coffee today, coffee for six guests covers orders, quantities and the sequence of brewing and pouring.
The Camera Is Off, but the Record Survives
The live Cambridge feed ended on August 22, 2001. The surviving Cambridge service page and participant accounts are historical records. They preserve the story of the shared pot and the route from a local desktop application to a public Web destination.
A personal note: I like the smallness of the question behind the Trojan Room camera: is it worth leaving the desk for coffee? The picture only needed to remove enough uncertainty to make that decision easier. That is a satisfying piece of design to me. I also find the two network milestones worth keeping apart, because the local tool already had a purpose before the Web gave it a wider audience. The same pot could be useful to a colleague and intriguing to a stranger. For me, the story connects an ordinary shared resource with a change in how people could see distant places.
Sources
- Quentin Stafford-Fraser: The Trojan Room Coffee Pot, a non-technical biography
- Quentin Stafford-Fraser: The Trojan Room Coffee Pot Timeline
- Quentin Stafford-Fraser (2001): The Life and Times of the First Web Cam, Communications of the ACM 44(7)
- University of Cambridge Computer Laboratory: The Trojan Room Coffee Machine service page
- Quentin Stafford-Fraser, Trojan Room coffee pot xcoffee.png, published in 1995. CC BY-SA 3.0. Image unchanged.