"I get called in by my boss, one day," says Gordon, a fast-talking New Yorker who grins as he starts into a story he has obviously told many times before. "He was the head of the math section, and he says, 'We have a new IBM 604 electronic calculator, and we'd like one of the actuarial students to learn how to use it. What do you think?' I said, 'Sounds terrific!' I didn't know what the hell he was talking about -- I'm 23 years old, on my first job -- but I said it sounded great."

Just by studying the manual, Gordon mastered the essentials of the 604 well enough that he was able to fix it when it malfunctioned a couple of weeks later, even without ever having previously laid eyes on the machine. Within days, he was transferred into the "tabulating division." He felt blessed.

"We used to sit in the math section," says Gordon, "churning out rates and forfeiture values and dividends and whatnot, and now I'd come across this fabulous electronic machine that does stuff miraculously at lightning speeds, and I'm thinking, I'm in on a revolution! I'm going to free mankind from the drudgery of sitting at calculators!"

There was just one little problem. After setting up the 604 to do whatever particular operation was required, the machine would spit out "reams and reams" of punch cards that in turn would be fed into the "tabulator" which would then promptly produce printouts of neat columns of numbers. But only numbers -- there were no headings, no indications of what the numbers referred to. The tabulator could only handle numbers. To make the information "camera-ready" for later copying and distribution, those headings had to be added.

"So one day I'm wondering, where do those headings come from?" says Gordon. "And one evening, on a break during overtime, I go down to the math section where all the people had been freed from the drudgery of hitting the calculators. And they are sitting there, cutting and pasting headings on the pages to make camera-ready copy! And I thought, this is what I have freed them for? There is something terribly wrong with this whole system! At that point I almost quit computing to do something else."

Mort Bernstein's living room explodes into laughter. Barry Gordon is sharing his story with a roomful of aged programmers -- veterans of the industry who all started out in the early '50s. The breadth of experience shared by the nine men and one woman gathered together in this Santa Monica suburb is impressive. They have literally seen it all.

Today, they trade tips on avoiding Microsoft Outlook viruses, or dual-booting their personal computers with Linux. But their experiences are grounded in the biggest of "big iron" -- cumbersome IBM mainframes like the 701 "Defense Calculator" and its successor, the 704. Most were employed by Southern California aerospace companies or think tanks: RAND, Lockheed, Douglas, North American Aviation. Notwithstanding Gordon's insurance tales, these programmers were accustomed to solving problems of slightly more significance than, say, handling Web site traffic or calculating life expectancies. Irwin Greenwald simulated the H-bomb explosion on Eniwetok at RAND, for example, while Westinghouse's Frank Engel modeled the possible deadly malfunction of a Nautilus submarine nuclear reactor.

Some, like Phil Cramer, are still working -- he runs a company that specializes in software for auto dealerships. Others, like Frank Wagner, the patriarch at this meeting, have been retired for several decades. Still others, like the host, Mort Bernstein, amuse themselves with hobbies: Bernstein is trying to program an "emulator" for RAND's Johnniac computer on his PC, but is finding it to be quite a task. Not only does he have to emulate the Johnniac's processor, but he also has to emulate the punch cards that fed it, and the punch card reader, and so on.

The youngest doesn't look a day over 65, but that isn't slowing anyone down. Jokes, anecdotes about long-forgotten computers and disputes about ancient programming lore zing back and forth. They delight in insulting each other, their long-dead colleagues, and most of all, IBM. Programmers of any age appreciate the art of the deftly delivered sarcastic jab, or, in the parlance of these coders, the "cut-down." When Gordon recalls a going-away party at IBM for a manager who appears to have been spectacularly nondescript, and quotes the parting toast "John, your departure will fill the void that was created when you first got here," the laughter is hardly polite -- it's an uproar.

The occasion is a reunion for former members of SHARE, an IBM user's club. Founded in 1955, SHARE may have been the first official computer user's group ever. Certainly, for a time, it was one of the most powerful.

SHARE was an outgrowth of an earlier collaboration between Southern California aerospace companies called PACT -- the Project for the Advancement of Coding Technologies. PACT's goal was to write a compiler for the 701, IBM's first commercial digital computer. SHARE started out as a proactive measure to prepare for the arrival of the 704. Although the name predated a later reverse-engineered acronym -- "The Society to Help Alleviate Redundant Engineering" -- from the very beginning SHARE aimed to save individual programmers from the sorry fate of writing basic code for essential tools that had already been written by someone else.

"We wanted to do something about this silly business of everybody programming their own square root routine," recalls Wagner, who managed programmers at the Mustang fighter plane manufacturer North American Aviation. The 704, which filled a large, specially built room with its card punch, card reader, printer, CRT, magnetic tape reels, magnetic drum, magnetic core storage, central processing unit and operator console (each of which was a separate machine), was not only hard to use, but also didn't come equipped with much in the way of pre-installed "software." Instead, the 704 arrived with a 103-page "Principles of Operation" manual, an assembler and some very basic utility programs, such as a single punch card "bootstrap loader" to get the machine started.

These huge mainframes were more than welcome to the aerospace companies -- they were essential. In the tense days of the Korean War, the pressure was on the defense industry to keep up with every move made by the Soviets and Communist Chinese. But, as Wagner recalls, it was getting harder and harder to tote up the necessary numbers involved in high-tech aeronautic design. "We were drowning in arithmetic," says Wagner, a genial, albeit occasionally sharp-tongued man treated by the other programmers with a mixture of respect and friendly deference. "Whenever an aircraft design changed, you had to go way back and start all over again."

Cooperation, for the purpose of the elimination of "redundant" effort, was the order of the day. The companies didn't share all their software -- they kept their structural analysis programs to themselves. But they did share the tools that they used to build such programs. Such cooperation only made sense to programmers who hated wasting their time, and who, according to Gordon, "tended not to be company loyal or commercially oriented."

"They were loyal to their profession," says Wagner.

And their drinking buddies. Certainly, as an example of programmer pragmatism prefiguring such open-source standbys as the Apache Web server, or Linux itself, by decades, SHARE is historically noteworthy. But SHARE wasn't just about saving money and time -- it was also about having fun with your community. Going to SHARE meetings was a blast, and not least because every night there was an open bar "SCIDS" meeting: The "SHARE Committee for Imbibers, Drinkers and Sots." SCIDS was where the action was, where the "technical" people gathered to lubricate themselves on alcohol and algorithms. As John Backus, the principal author of IBM's Fortran programming language, noted during a 1980 commemoration marking 25 years of SHARE, "There were two principal pleasures of SHARE: Blasting IBM -- giving them hell for fouling up and not giving them what they wanted -- and the second activity, known as SCIDS."

Programmers labor under a stereotype that maligns them as anti-social shut-ins. But while it is true that the act of programming is solitary toil, programmers are also intensely social. It's hardly an exaggeration to suggest that the Internet was built mainly so that programmers would have a place where they could get together and chat about their favorite science fiction novels; but before the Internet, you had to gather at conventions, or conferences. That's where you shared notes with your colleagues, that's where you learned your craft and honed your programming chops. That's where you figured out how to fix your Giant Brains.

The Free Software Project
Read Andrew Leonard's book-in-progress on Linux and open source -- and post your comments.