They are an elite fraternity, the programmers who create the heart and soul of computers-- the operating systems that build bridges from the intricate hardware to screens full of words and symbols that a child, or a corporate vice president, can understand.
In that fraternity, James Allchin PhD ICS '83, is a legend.
At 41, he has already lived through several generations of computing technology. He made his reputation before there were personal computers, before screens were filled with the elegantly accessible images that allow unskilled users to harness a computer's power with the click of a mouse. In an earlier era, programmers like Allchin spoke to computers in the only language the machines understood - 1's and 0's.
Allchin, colleagues say, is the ultimate computer hacker, with a style and intelligence that blends the mathematical, logical and intuitive. He has carried with him for more than a decade a passion for inventing a new kind of computing. Called distributed processing, it is a vision of coupling the power of hundereds of large and small computers
The Call From Gates
Several years ago Allchin's work caught the eye of William Gates, and in late 1990 the chairman of Microsoft Corp., the nation's largest software publisher, recruited him away from a rival.
Gates had his own vision, dubbed "information at your fingertips," in which each desktop computer would be seamlessly woven into a vast tapestry of data. He picked Allchin to make it happen.
Computer users, of course, can already retrieve all manner of information, by connecting to remote data bases and issuing complex commands. But Gates and Allchin are planning to embed far simpler and more powerful storage and retrieval capabilities into the basic command set of each personal computer. Rather than being able to retrieve information only by document name, it will be possible to simultaneously browse through vast data bases stored in many separate places, sorting by similarities in the meanings of words hidden inside a document even by likeness in graphic images from photos.
Putting such power in the hands of every personal-computer user is seen as the holy grail by many in the computer industry; Competitors like Apple, Novell, Next, Lotus and Sun Microsystems are all working on similar projects.
But the popularity of Microsoft's Windows program gives Gates a significant lead. He has put Allchin work weaving his distributed computing software into what will become a future version of Windows.
Code-named "Cairo," the project has attracted wide interest - and skepticism. Even at Microsoft people recognize the gamble involved. "There are skeptics," said Jonathan Lazarus, Microsoft's vice president for strategic systems. "Jim is the one who is promising to build the electric car."
Yet many computer-industry executives agree that while Windows NT (the operating system that Microsoft irtroduced last month to compete against Unix and other systems) has the spotlight now, it will ultimately be Cairo that has the greater impact.
If Cairo does succeed, it will be because Allchin delivers on his long-held vision. In the early 1980s, his exploits were described in a document widely circulated over computer networks. 'Titled "Real Programmers Don't Write Specs," it included 24 "rules" that purport to describe the heroic ventures of the first generation of hackers -- rules like "Real programmers don't eat quiche. They eat Twinkies and Szechwan food," and "Real programmers never work 9-to-5. If any real programmers are around at 9 a.m., it's because they were up all night."
The report also detailed a number of heroic sagas, including one about Allchin, then working at Texas lnstruments. One day, the story, goes, he got a long-distance call from a user whose system had crashed in the middle of saving some important work. AlIchin assessed the damage over the phone by getting the user to use toggle switches to enter instructions controlling the disk drive into the front panel of the computer. He then repaired the system -- without ever seeing it -- by instructing the user to enter cryptic numbers that only computers understand and reading the contents of individual registers back to him.
All in the Family
Math skills were a family trait.
Although neither his father nor mother was college educated, AlIchin remembers watching his father standing in the barn doing in his head the math necessary to sell the products produced on their Michigan farm. "My father was always right," he recalled.
Today, like many computer scientists, for relaxation Alichin plays jazz or blues guitar, frequently jamming in the evenings with professional musicians in the Seattle area. His wife, Bonnie, is a painter.
His talents were apparent early.
"He swept through exams in record time," said Eugene Spafford, a computer scientist at Purdue University who was a graduate student with Allchin at Georgia Institute of Technology in the early 1980s.
At Georgia Tech a small group of researchers was working on an advanced operating system known as Clouds, designed to be spread across a number of computers connected on a network. The idea was to distribute computing and data to gain performance and reliability.
After leaving graduate school in 1983, Allchin helped found Banyan Systems, a Microsoft competitor.
Allchin was not easy to recruit.
"My view of Microsoft was pretty dim," he recalled. I said, "You guys brought us DOS," referring to Microsofts flagship operating system, which today is still the standard for office personal computers but is viewed with contempt by many of the technically savvy.
Gates, however, is notoriously persistent and persuasive. He spent more than a year pursuing Allchin. Finally, after four meetings, Gates won his allegiance with devastating logic.
"He told me, 'No matter what you do at Banyan, you won't have the impact you'll have here,' "Allchin said.
lndeed, Gates, in his drive to consolidate his control over the computing world, has in the past five years acquired a remarkable stable of the world's best operating-system gurus. In addition to Allchin, his star recruits include David Cutler, a Digital Equipment veteran who came to Microsoft five years ago and is now completing work on Windows NT. Another is Richard Rashid, a Carnegie Mellon computer scientist who designed an advanced version of Unix.
All have come to Microsoft because Gates has managed to convince them that he planned a huge investment -- it is now about $400 million a year -- to build the world's most advanced software.
The Computer Teams
Cutler and Allchin both run groups of more than 100 programmers, but their styles are remarkably different. Cutler has a reputation fo no-nonsense military-style leadership and a willingness to put his fist through a wall when things are not going well. He is also notorious for all the niceties of corporate life that don't relate directly to his technical goals. Allchin, by contrast, is said to attend executive staff meetings regularly.
Allchin, who with his shock of white hair was jokingly referred to as the company's "great white hope" when he began, has a cooler - if no less intense approach.
"I don't have a military style, but people don't want to see me in their office, either," he said. "I'm just as intense and passionate as Dave. Which style is more effective? I don't know." Allchin is involved in broad planning, but tends to the nitty-gritty, too, reading every program his team writes.
Allchin can still bristle when he is challenged. At a computer-industry conference last year, he got into a nasty spat with Steven Jobs, the chairman of Next Inc. Allchin told Jobs he was wrong in many of his technical claims. In the end, though, Jobs was impressed - and later tried to lure Allchin away from Microsoft.
The Road to Cairo
Cairo's most important feature will be to allow a network of thousands of computers, each with large data bases, to appear as a single web of data to a desktop-computer user.
The project got its name when two Microsoft executives decided that there were too many American project code names - Lazarus tossed out several foreign names including Cairo.
Industry executives say Cairo is now scheduled to arrive sometime in 1995, after the release of another version of Windows, code-named Chicago, probably next year.
One feature will be an information browser called an "explorer," say those who have seen demonstrations of Cairo. Unlike conventional browsers, which simply organize information alphabetically or by document size or type, Cairo will help the computer to find deeper relationships among different documents based on language and word meanings. Someone could, say, order a search for all animals, and the computer would know which words refer to animals.
Also, just like an efficient secretary, the Cairo user interface might offer the capability to organize electronic mail, sorting the most important documents for immediate attention. To achieve these goals, Allchin's group is going beyond simple filtering of information. They are also making use of a rudimentary understanding of word meanings and synonyms.
It hasn't been easy, Allchin acknowledges that he has already "torn up the binder" --the specifications for the project -- and started from scratch twice.
Last summer, for example, the Cairo group realized that their work was duplicating that of another major Microsoft development group known as OLE, for object linking and embedding.
Microsoft is racing, with the rest of the industry, to rewrite its software around a design approach called objects. These are pre-built software called modules that can easily be hooked together, making software development faster and making it easier for users to customize applications.
But Cairo and OLE had different approaches, and Allchin said he realized there could be only one style of objects at Microsoft. The Cairo work was thrown out. It was a painful decision but one he says he had to make. "We were farther along than OLE," he said, "but it was my decision to say one object model. Period."
Today, Allchin says he feels as if Cairo is on track. "We now understand everything and either have the code written or understand how to do it," he said recently in an interview at Microsoft's Redmond, Wash., corporate campus. But he is coy about saying when Cairo will emerge from the laboratory.
Last year he rewarded his programmers for meeting a key milestone by finding a camel and bringing it into the office. The camel was an immediate hit with the Cairo team, who petted it and had their pictures taken with it. "Now they keep asking when the camel is coming back," he said with a grin. "I keep telling them that it will be back, but I'm not saying when."
Copyright 1993 by the New York Times Company. Reprinted with Permission.