frustration with a grain of salt. Still, the difference between waiting for a machine and waiting on a machine is a sizable one. It wasn’t the first time he’d been forced to stand over the printer, watching pages print out one by one. As a person who spent the bulk of his days and nights improving the efficiency of machines and the software programs that controlled them, Stallman felt a natural urge to open up the machine, look at the guts, and seek out the root of the problem.
Unfortunately, Stallman’s skills as a computer programmer did not extend to the mechanical-engineering realm. As freshly printed documents poured out of the machine, Stallman had a chance to reflect on other ways to circumvent the printing jam problem.
How long ago had it been that the staff members at the AI Lab had welcomed the new printer with open arms? Stallman wondered. The machine had been a donation from the Xerox Corporation. A cutting edge prototype, it was a modified version of the popular Xerox photocopier. Only instead of making copies, it relied on software data piped in over a computer network to turn that data into professional-looking documents. Created by engineers at the world-famous Xerox Palo Alto Research Facility, it was, quite simply, an early taste of the desktop-printing revolution that would seize the rest of the computing industry by the end of the decade.
Driven by an instinctual urge to play with the best new equipment, programmers at the AI Lab promptly integrated the new machine into the lab’s sophisticated computing infrastructure. The results had been immediately pleasing. Unlike the lab’s old laser printer, the new Xerox machine was fast. Pages came flying out at a rate of one per second, turning a 20-minute print job into a 2-minute print job. The new machine was also more precise. Circles came out looking like circles, not ovals. Straight lines came out looking like straight lines, not low-amplitude sine waves.
It was, for all intents and purposes, a gift too good to refuse.
It wasn’t until a few weeks after its arrival that the machine’s flaws began to surface. Chief among the drawbacks was the machine’s inherent susceptibility to paper jams. Engineering-minded programmers quickly understood the reason behind the flaw. As a photocopier, the machine generally required the direct oversight of a human operator. Figuring that these human operators would always be on hand to fix a paper jam, if it occurred, Xerox engineers had devoted their time and energies to eliminating other pesky problems. In engineering terms, user diligence was built into the system.
In modifying the machine for printer use, Xerox engineers had changed the user-machine relationship in a subtle but profound way. Instead of making the machine subservient to an individual human operator, they made it subservient to an entire networked population of human operators. Instead of standing directly over the machine, a human user on one end of the network sent his print command through an extended bucket-brigade of machines, expecting the desired content to arrive at the targeted destination and in proper form. It wasn’t until he finally went to check up on the final output that he realized how little of the desired content had made it through.
Stallman himself had been of the first to identify the problem and the first to suggest a remedy. Years before, when the lab was still using its old printer, Stallman had solved a similar problem by opening up the software program that regulated the printer on the lab’s PDP-11 machine. Stallman couldn’t eliminate paper jams, but he could insert a software command that ordered the PDP-11 to check the printer periodically and report back to the PDP-10, the lab’s central computer. To ensure that one user’s negligence didn’t bog down an entire line of print jobs, Stallman also inserted a software command that instructed the PDP-10 to notify every user with a waiting print job that the printer was jammed. The notice was simple, something along the lines of “The printer is jammed, please fix it”, and because it went out to the people with the most pressing need to fix the problem, chances were higher that the problem got fixed in due time.
As fixes go, Stallman’s was oblique but elegant. It didn’t fix the mechanical side of the problem, but it did the next best thing by closing the information loop between user and machine. Thanks to a few additional lines of software code, AI Lab employees could eliminate the 10 or 15 minutes wasted each week in running back and forth to check on the printer. In programming terms, Stallman’s fix took advantage of the amplified intelligence of the overall network.
“If you got that message, you couldn’t assume somebody else would fix it”, says Stallman, recalling the logic. “You had to go to the printer. A minute or two after the printer got in trouble, the two or three people who got messages arrive to fix the machine. Of those two or three people, one of them, at least, would usually know how to fix the problem”.
Such clever fixes were a trademark of the AI Lab and its indigenous population of programmers. Indeed, the best programmers at the AI Lab disdained the term programmer, preferring the more slangy occupational title of hacker instead. The job title covered a host of activities-everything from creative mirth making to the improvement of existing software and computer systems. Implicit within the title, however, was the old-fashioned notion of Yankee ingenuity. To be a hacker, one had to accept the philosophy that writing a software program was only the beginning. Improving a program was the true test of a hacker’s skills.[1]
Such a philosophy was a major reason why companies like Xerox made it a policy to donate their machines and software programs to places where hackers typically congregated. If hackers improved the software, companies could borrow back the improvements, incorporating them into update versions for the commercial marketplace. In corporate terms, hackers were a leveragable community asset, an auxiliary research-and- development division available at minimal cost.
It was because of this give-and-take philosophy that when Stallman spotted the print-jam defect in the Xerox laser printer, he didn’t panic. He simply looked for a way to update the old fix or “hack” for the new system. In the course of looking up the Xerox laser-printer software, however, Stallman made a troubling discovery. The printer didn’t have any software, at least nothing Stallman or a fellow programmer could read. Until then, most companies had made it a form of courtesy to publish source-code files-readable text files that documented the individual software commands that told a machine what to do. Xerox, in this instance, had provided software files in precompiled, or binary, form. Programmers were free to open the files up if they wanted to, but unless they were an expert in deciphering an endless stream of ones and zeroes, the resulting text was pure gibberish.
Although Stallman knew plenty about computers, he was not an expert in translating binary files. As a hacker, however, he had other resources at his disposal. The notion of information sharing was so central to the hacker culture that Stallman knew it was only a matter of time before some hacker in some university lab or corporate computer room proffered a version of the laser-printer source code with the desired source-code files.
After the first few printer jams, Stallman comforted himself with the memory of a similar situation years before. The lab had needed a cross-network program to help the PDP-11 work more efficiently with the PDP-10. The lab’s hackers were more than up to the task, but Stallman, a Harvard alumnus, recalled a similar program written by programmers at the Harvard computer-science department. The Harvard computer lab used the same model computer, the PDP-10, albeit with a different operating system. The Harvard computer lab also had a policy requiring that all programs installed on the PDP-10 had to come with published source-code files.
Taking advantage of his access to the Harvard computer lab, Stallman dropped in, made a copy of the cross-network source code, and brought it back to the AI Lab. He then rewrote the source code to make it more
