SAGE and a glimpse of group computing from before the PC

22.18, Tuesday 21 Dec 2021

There was a fork in the road away from group computing, way back when.

Right now we’re in the era of personal computers. Collaboration, social use of tools, togetherness: all of these are hacks on top of something that, at its core, was designed for the individual first.

But there’s a particular photograph of group computing from the 1950s, from before the PC was invented…

Ok I need to rattle out a story here about SAGE, and that will let me get to the photograph.

I’m going to do this from memory so apologies in advance for any factual errors, but I think I’ve got the bones of it in order.

(Earlier this year I did a three part talk about the pre-history of computing and so a lot of this is in my head from then. It was a super fun talk with a novel format – three x 1 hour talks on successive evenings across a single conference, each picking out and storytelling around particular moment in the evolution of today’s technology. And it got great ratings in the feedback. I was pleased about that.)

To get to SAGE and to put it in context, I need to rewind a whole way.


One way (not the only way) of telling the story of computers goes like this.

Tabulating machines were invented in 1890 for the US Census and went on for the next 50 years without any fundamental changes but with great popularity in business. They were electromechanical sideboards that basically ran a small set of Excel commands on stacks of what became standard issue IBM punchcards – which we retain today in the shape of air flight boarding cards.

The Second World War was a catalytic event. The need to quickly calculate ballistics tables led to the first fully electronic computer, ENIAC (which also became one of the world’s first programmable computers in the modern sense), although the initial task in 1945 for this room-sized machine was numerical modelling for the hydrogen bomb at the tail end of the Manhattan Project, the vast secret project to create nuclear weapons.

Let’s take a moment to name the first of the first, the original programmers of ENIAC, all women: Kay McNulty, Betty Jennings, Betty Snyder, Marlyn Meltzer, Fran Bilas, and Ruth Lichterman.

Skip ahead to 1968 and the invention of the personal computer. Douglas Engelbart’s team attaches a screen to a military computer and, in 90 minutes in The Mother of All Demos, demonstrates a whole new user interface: interactive text, video collaboration, modern office furniture, and the mouse. The demo consisted of Engelbart managing his shopping list.

The PC was an audacious conceptual leap: the idea of an individual computer being used by a single person for their own specific work tasks was akin to the idea of a baseball stadium being used by one player (Alex Handy, The New Stack).

Between the milestones of ENIAC and Engelbart there was SAGE.

For comparison: The Manhattan Project in the 1940s cost about $20bn in today’s money. SAGE cost $60bn. The Semi-Automatic Ground Environment was built out over the 1950s and was a direct defence against the atomic weapons developed in the mega-project of the previous decade, a continent-wide sensing, synthesis, and rapid response platform blending human intelligence and technology. It ran for over 20 years.

The SAGE system, by the time of its full deployment, consisted of 100s of radars, 24 direction centers, and 3 combat centers spread throughout the U.S. The direction centers were connected to 100s of airfields and surface-to-air missile sites, providing a multilayered engagement capability. Each direction center housed a dual-redundant AN/FSQ-7 computer, evolved from MIT’s experimental Whirlwind computer of the 1950s. These computers hosted programs that consisted of over 500,000 lines of code and executed over 25,000 instructions – by far the largest computer programs ever written at that time. The direction centers automatically processed data from multiple remote radars, provided control information to intercepting aircraft and surface-to-air missile sites, and provided command and control and situational awareness displays to over 100 operator stations at each center.

Imagine a network of radar stations covering the whole of North America, constantly looking out for nuclear bombers. SAGE never spotted one; bombers were quickly replaced as delivery mechanisms by the ICBM. It was the prototype for today’s air traffic control system.

Radar signals came into the 24 direction centers, were analysed by people and computers, and instructions sent out again to bombers and missile silos: the original WarGames.

Each direction center was built around an AN/FSQ-7 – at 250 tonnes, the largest computer ever built. (With the transistor replacing vacuum tubes a few years later, the largest that ever would be built.) Then what you’ve got with SAGE is 100+ operator stations plugged into the same computer.

You can see the weapon’s director console here. There’s a circular screen, like a radar display, and a light gun to select radar traces. There are toggle switches and a numerical rotary dial to tag the traces with numbers.

This is interactive computing in the real world for the first time!

So, for me, this is the pre-Engelbart breakthrough moment.

And in a way, it’s more authentic than any research project. What we see in SAGE is a new interface forged under perceived existential threat, something actual, not philosophically derived or imagined by lone genius, but contoured along the grain of known human behaviour.

On the left hand side of the console there is an ashtray.

(Also it is not lost on me that the light gun was the particular interface device that was popularised by its use in this military system. Imagine being so accustomed to weapons that it feels entirely natural to select an icon on your computer screen by pointing a gun at it.)

The successor to the AN/FSQ-7 was intended to be the much smaller, transistorised AN/FSQ-32.

Engelbart’s team in Stanford got its start in 1963 when it was given its first research contact by J. C. R. Licklider, director of ARPA: the US Defense Department’s Advanced Research Projects Agency. Licklider made it a condition that, instead of using a standalone computer in Stanford, Engelbart had to start by connecting a display to the new AN/FSQ-32.

So a connection from SAGE to Stanford and that brings us back to the PC.

But it wasn’t inevitable that the Semi-Automatic Ground Environment would be followed by the personal computer.

I’m kinda obsessed with a particular road not taken…

LOOK AT THIS. Here’s an archive photo on Wikipedia: Subsector Command Post of SAGE Combat Center at Syracuse Air Force Station with consoles and large Photographic Display Unit display, which was projected from above.

Group computing.

What you can see is ten men sitting around a Kelvin Hughes Photography Display Unit – a large screen for displaying graphics and characters. The third floor of a SAGE direction center: The Pit.

Each of the men (yes all men) has their own computer console at their desk. But they’re working together around the PDU. One of the men is holding what is either a light gun or a laser pointer/equivalent. They’re assessing potential threats and ordering missiles and bombers. Together.

(Here’s a great article about the physical architecture of SAGE: The Futuristic Cold War Era SAGE Air Defense Bunkers Looked Right Out Of A Kubrick Film.)

So we’ve got a system here in the 1950s which is on some axis more sophisticated than the rooms with similarly large yet typically less intelligent screens that I regularly sit and have meetings in, seven decades later.

This isn’t a setup for presentations and discussions. It’s for collaboration and action. The whole room is an environment for the team to work.

And I often, often wonder this:

WHAT IF, instead of the Personal Computer, the dividend of SAGE had been the Team Computer?

A computer that wasn’t used individually but as a group, together in a room or perhaps remotely. Not desktops but environments. An alternate history of computing that doesn’t involve user IDs or ownership as primary concepts but is instead oriented around collaborative, co-created artefacts, spaces that are jointly inhabited. It’s hard to mentally unfold such a world, from such different initial conditions, imagining its progress lensed through Microsoft Office analogues, video games analogues, World Wide Web analogues, over such a stretch of time.

It would look something like DynamicLand but with decades of evolution. We got just a glimpse of the path.

More posts tagged:
Follow-up posts:

If you enjoyed this post, please consider sharing it by email or on social media. Here’s the link. Thanks, —Matt.