Software Wizards: Joel Berez

Robert Levering, Michael Katz, and Milton Moskowitz

---

This is an excerpt from:

Levering, Robert, et al. The Computer Entrepreneurs: Who's Making it Big and How in America's Upstart Industry. New American Library, 1984, pp 120-26.

---

Joel Berez

"It's clearly the most complicated, convoluted, disgusing piece of code that's ever been written"

---

---

The story of Infocom could be told as a piece of Infocom fiction: Once upon a time there was a strange galaxy called MIT. In the galaxy was a nebula called the Computer Science Lab, and in the nebula was a world-system known as the Programming and Technology Division. It was a long time ago, around the year 6 PCE (personal computer era). The inhabitants of this world-system had learned to travel through neuronic space on a mighty ship named the Mainframe. One day the Mainframe carried them to a place where they had never been before. It was called Adventure. It was a game.

If this were really an Infocom game, it would get interactive at this point. You would enter the nebula, find out all you could about it, have hair-raising skirmishes with creatures and forces, and through your choices determine the course of the story, which would have dozens of possible endings.

The Infocom story, since it actually happened, is less flexible (though not unwhimsical). When the MIT computer whizzes discovered Adventure in the circuitous guts of their mainframe in 1976, they were enchanted. That trailblazing game, created by a pair of earlier MIT graduates, offered a fantasy world with which the player could interact through simple verb-noun commands like "Take sword." But soon the young researchers felt a competitive itch. They knew they could create a game with many more complex options and a far more sophisticated text.

Three of them — Marc Blank, Dave Lebling, and Tim Anderson — started working nights on what was to become a cult: Zork I, which they initially set loose on the mainframe network in 1977, and which is now the first part of a Zork trilogy. Blank made up the nonsense name. With the help of MDL (pronounced Muddle), a powerful programming language previously developed at the lab by Blank and Anderson, Zork was several giant steps ahead of Adventure.

With Zork, you can type in a command like this: "Pick up the troll-wand and the rubber gerbil and give them to the Bog Monster. Then follow the magic chicken." When you find yourself facing the evil, axe-wielding troll in a gory grotto and decide to defend yourself by throwing a sack at him, you might get this shocking news: "The troll, who is remarkably coordinated, catches the brown sack and, not having the most discriminating taste, gleefully eats it. The flat of the troll's axe hits you on the head. I'm afraid you are dead."

Other members of the MIT programming research group were Joel Berez, Stuart Galley, Scott Cutler, and several others who would later become part of the Infocom inner circle. Credit for actually materializing the company goes to Albert Vezza, who was then the chief of their research group. He had long cherished the idea of channeling some of the talent at the lab into a commercial venture, and in 1979, Vezza brought together nine of the men who had worked with him at the lab a few years before.

Joel Berez was in Pittsburgh, where he was working for his father's company, Action Industries, and planning to go on to business school. Marc Blank had just finished medical school and was beginning an internship in Pittsburgh. When Vezza's invitation came to join him in Boston for the formation of Infocom, Berez quickly decided to attend MIT's Sloan School of Management, and Blank dropped his two-week-old internship in order to go back to Massachusetts. Infocom was founded in June 1979, and Berez soon became its president and company spokesman.

Blank and Berez had been getting together for dinners when they were both in Pittsburgh. One of their favorite topics was Zork, which by then was quite a popular item among mainframe programmers. As Berez remembers it, they were at the Grand Conqueror restaurant, spooning up fish chowder, when a bright idea passed between them: Why not try to rewrite Zork for a personal computer and market it?

It was obvious why not. The program was immense, taking up a million characters of storage space on the mainframe; a floppy disk for a personal computer held about 100,000. But, in the next few days, both men became captivated by the idea, and they had the feeling they could do it. "It involved writing several new computer languages, a compiler, an assembler, and interpreters to run a lot of the machines," Berez told us, "but we thought it really was possible."

It was. They finished Zork in 1980, after recruiting former MIT comrades Michael Dornbrook, Stuart Galley, and Steve Meretsky as testers. A fledgling company called Personal Software took over the marketing and Zork quickly became a best-seller. Then, with the phenomenal success of its financial program, VisiCalc, Personal Software changed its name to Visi-Corp and divested itself of products that seemed to be beneath its newfound dignity. Naturally the grotto-and-monsters game had to go.

This was 1981, the year Berez graduated from MIT's business school and became Infocom's first full-time employee. Soon Blank also signed on full time, Zork II was released, and Deadline, their first mystery, was under development. In early 1982, sales trailed off, but when Deadline appeared in March, the charts soared and they've been soaring ever since. Sales in 1983 were $6 million, an increase of about 300 percent over 1982.

A New York Times Book Review writer described Deadline as "more like a genre of fiction than a game," declared that the narrative forms favored by Infocom have "archetypal power," and compared the Zork adventure to the Odyssey.

Most successful computer-related companies have a single star, or at most, two co-stars. Infocom is different. When newspapers and magazines write about the company, they favor group pictures: Berez, Blank, and Vezza; writers Galley, Meretsky, Lebling, and Michael Berlyn; Blank, Berez, and Dornbrook. They all have interesting stories. Berlyn, for instance, is the only non-MIT man in the leadership group. He was a recognized science fiction writer and head of a Colorado company called Sentient Software when he got hooked on Zork I. "I need more Zork!" he cried out, but there was no more. So he went to Boston and joined the group writing Infocom fiction, combining his skills as novelist and programmer. And when computer lab alumni Michael Dornbrook went to business school in Chicago, he initiated ZUG, the Zork Users Group, which he reckoned would take a few hours of his time each week. But when membership reached 20,000, he found himself obsessively producing maps, posters, T-shirt transfers, and bumper stickers. In 1983, he joined the company's management.

The fact that no single star outshines any other at Infocom must be due, at least in part, to the group's genesis in the MIT computer lab. The atmosphere of that lab — camaraderie, humor, intense intellectual effort that is both individual and collective — has carried over into the company. Ownership is fairly evenly distributed, with a number of people holding 10 to 15 percent of the stock and all employees receive stock options (which many have exercised).

Something else that has carried over is the emphasis on creating "state-of-the-art development tools" before plunging into specific applications. "If your tools are powerful enough," says Berez, "the applications drop out fairly easily." In the case of Infocom's interactive fiction, this means inventing the languages and codes needed to translate programs from the mainframe to more than two dozen personal computers. It also means developing and refining the parser — that special part of the program which makes complex sentences and interactions possible. Michael Berlyn says of Infocom's parser: "It's clearly the most complicated, convoluted, disgusting piece of code that's ever been written." Yet the company admits that its parser is still primitive compared to those that will make way for a truly sophisticated genre of interactive literature in the future.

Infocom claims to be the only major software company that writes its programs on a mainframe and then converts them into languages that can be understood by personal computers. To perform this esoteric act of translation, Marc Blanc and Joel Berez created ZIL (Zork Interactive Language). A rather idiosyncratic promotional story booklet called Our Circuits, Ourselves! — The Heroic Struggle of Micro-Americans to be Free likens this process to the dehydration of oats into oatmeal. A kind, archetypal old Infocom steward explains to the timorous hero, oatmeal company employee Delwood Bland, "Our Infokins do the same thing — taking the vast amount of information that goes into making up a world, then condensing it to a floppy disk you can slip into your Micro-American, without losing any of the 'goodness.'"

Infocom's mainframe is the DEC 2060, which a Washington Post writer describes as "the biggest byte-bender in Digital Equipment's fleet, compared to which your IBM PC is dumb as a toaster." The 2060 costs $750,000; the company bought its second one in mid-1984. With this machine, Infocom says its programmers can do in three months what might take three years on a personal computer.

But the makers of Floyd, Frobozz, and the cryogenically preserved hero of Suspended aren't satisfied. While $6 million may sound good by itself, it doesn't sound quite as good when compared to the sales figures of leading business software companies. The collapse of the arcade game industry in recent years has given all personal computer game makers cause to stop and reconsider; the Infocom team has lately been working hard — and secretly — on business programs. According to the Wall Street Journal, they will say only that they expect to produce something "more sophisticated than current data base managers." They also hint at the possibility of customizing the new programs for individual industries, and of adapting their games' conversational ability "to make business software less daunting." Naturally, they express confidence that they can make their way successfully into the already glutted business market.

Here again, as the relatively stable past gives way to the imaginary future, Infocom becomes a piece of fiction in potentio. Many times the dauntless crew has rushed across the universe in the mother ship Mainframe. Many times they have zipped in and out of the great ship's ports in their small and sporty micros. Long ago they left behind their home nebula, and now once again they are venturing into dangerous, uncharted worlds. There are scores of possible endings. Will there be a crash? Or will the VisiWizard and the Lotus Enchanter soon be trembling?

No one knows who writes the code for this story.

---

---

These historical, out-of-print articles and literary works have been GNUSTOed onto InvisiClues.org for academic and research purposes.