The Birth of UNIX — Bell Labs

The UNIX story is often told as a recovery from a failure. The framing is accurate.

After Multics

In the mid-1960s, MIT, General Electric, and AT&T Bell Labs had collaborated on an ambitious operating system called Multics—the Multiplexed Information and Computing Service. Its goals were advanced for their time: timesharing, a hierarchical file system, dynamic linking, ring-based protection, a command language. All of them at once.

Multics did, eventually, ship. But Bell Labs withdrew from the project in 1969—too large, too slow, with no clear commercial outcome.

Ken Thompson, Dennis Ritchie, and Doug McIlroy, the Bell Labs researchers who had worked on Multics, were redirected back to their own work. Thompson wanted to keep what had been good about Multics—the hierarchical file system, the shell, the idea of treating devices as files—and reimplement it in something smaller, simpler.

Then he found a DEC PDP-7 that nobody was using.

On the PDP-7

The PDP-7 was already a slightly aged machine by 1969—an 18-bit minicomputer with 8 KB of memory, released in 1965. Bell Labs owned one for another purpose; it sat idle.

Thompson first wrote a small timesharing system for himself: a hierarchical file system, a handful of utilities, a simple shell. The skeleton was reportedly complete in three weeks while his wife was away on maternity leave with their newborn.

In 1970, Brian Kernighan (later the K of K&R) half-jokingly suggested the name 'UNICS'—a kind of anti-Multics. It would eventually be spelled UNIX.

C and Portability

The first UNIX was written in assembly. That meant it ran only on the one machine it had been written for.

In 1972, Dennis Ritchie extended Thompson's B language into a typed successor he called C. In 1973, UNIX was rewritten in it.

This was the property that would distinguish UNIX from every operating system before it. No operating system before UNIX had been written in a portable language. The OS had always been hand-fitted to the machine it ran on.

If you wanted to move UNIX to a different machine, you needed a C compiler. That was all. In a world where OS and hardware had always been inseparable, this was a basic reframing.

Bell Labs' Constraint and Gift

Under the 1956 consent decree, AT&T was forbidden from entering the computer business. Bell Labs could not sell UNIX as a commercial product.

So UNIX went out at nominal cost—essentially the price of the magnetic tape and the postage—to universities and research institutions. MIT, UC Berkeley, the University of Tokyo. Graduate students around the world got UNIX with full source code.

This determined the system's later fate. UNIX was, commercially, an unsellable OS for more than a decade. But during that decade it sank deep roots in academic computing. The PhDs and masters' students who emerged from those campuses carried the UNIX shell, the file system, and the philosophy of small composable tools into industry as a matter of habit.

When Linus Torvalds at the University of Helsinki began writing Linux in 1991, his model was UNIX. When Apple built iPhone OS in 2007, the OS X kernel beneath it was a UNIX descendant (a hybrid of BSD and Mach).

Most people who have never heard of UNIX now use a UNIX-descended operating system every day.