Allen Newell: The Lone Ranger
James H. Morris is a retired professor of computer science and dean of the School of Computer Science as well as the Silicon Valley campus of Carnegie Mellon University. In a series of blogs for Pittsburgh Quarterly he writes about some of the computing pioneers he encountered during his career.
After succumbing to Alan Perlis’s siren call in my sophomore year, I delved deeper into computer topics and encountered Allen Newell, who had somewhat more hair than Perlis, as much charm, and a more coherent research agenda: understanding human intelligence. In the opening class of his graduate seminar, the large, former football player said, “I found an interesting biology paper describing a model of how exertion, breathing, and one’s heart rate affect one another.” As he scribbled scary-looking differential equations on the board, I felt overwhelmed. Then, he admitted that he didn’t really understand it all, being a cognitive psychologist. I admired his courage and enthusiasm for diving into unfamiliar areas.
The courage to match wits with any professor came naturally to Newell, but was reinforced by his lifelong partner Herbert Simon. Herb discovered him at the Rand Corporation, brought him to the Carnegie Mellon University (then Carnegie Institute of Technology) business school, where he was awarded a Ph.D. and was made a professor.
One day in his class, Newell asked a student if he would make a presentation the following week. The cheeky student (David Parnas, who became a software guru) said, “Well, if you twist my arm…”, so Newell, always open to a joke, walked over to him and twisted his arm until he agreed.
On another occasion, when a Ph.D. student claimed the system he had designed would take too long to program, Newell said, “We could do it in a weekend,” and proceeded to spend the following 48 hours proving it.
When Newell met Fred Rogers, creator of the “Mister Rogers’ Neighborhood” children’s educational television series, he accused him of being cruel for not considering the feelings of robots. Rogers thought he was nuts, but Newell was delighted with his joke.
For Newell’s seminar, I wrote a paper reviewing work in neural networks as a computational model, including a paper co-authored by a young Manuel Blum. Newell criticized it fiercely, wounding my undergraduate pride and diminishing my belief in his congeniality. It was the first of many times that I saw how he took no prisoners when it came to intellectual matters.
By the 1980s, universities had grasped the idea that computing was as central to their mission as libraries. Urged by Newell, Carnegie Mellon raced to lead this trend by adopting personal computers, connected by a network. University President Richard Cyert negotiated a huge contract with IBM to underwrite the effort.
The project came to be called Andrew after university benefactors Andrew Carnegie and Andrew Mellon.
Unaware of any of this, I visited Carnegie Mellon for a sabbatical. Within a month of my arrival, Cyert announced the “computer for every student” project. Then, Newell and others were urging me to leave Xerox and become the director of the project. I refused at first because working for IBM, even indirectly, would be taboo for a person from Xerox’s Palo Alto Research Center (PARC). Newell said, “The thing about taboos is that, when you actually break one, it’s quite enjoyable.”
It all fit perfectly. Xerox was abandoning its office of the future dream and my Xerox colleagues were departing for startups and the Digital Equipment Corporation. There was plenty of funding. The computer science department was graduating many brilliant Ph.D.s who were itching to implement fancy systems and I was anxious to carry out PARC’s vision somehow. So I stayed in Pittsburgh while my PARC colleagues pursued the vision throughout Silicon Valley.
Our project was bedeviled with the question of which kind of personal computer the students should use. In my first month on the job, Steve Jobs arrived to sell Macintoshs. In his hotel room, wearing his hooded sweatshirt he excitedly showed us the yet-to-be released computer, editing text and drawing pictures. Most university customers might have been astounded by this, but we recognized it as very watered-down Alto computer in a little plastic box. When I asked about networking, Jobs mumbled something about Big Brother. In any case, Macintoshs were a non-starter for a project sponsored by IBM, which had just introduced its personal computer.
On the other hand, the project’s staff, mostly graduates of the computer science department, loved the UNIX computers being sold by SUN Microsystems, a much beefier and expensive version of the Alto.
The battle among IBM, Apple, and SUN raged, at one point causing the vice president of the university to drive out my boss, the vice provost for computing.
In the middle of all this, I went to Newell for help and he gave me some prescient advice: “The network, not the device, is the important thing to get right.” Indeed, the network file system we developed is still actively used in many places 30 years later and recently won a national award. In the meantime, the personal devices have gone from workstations, to desktops, to laptops, to smart phones.
That was one of many occasions where his wisdom and generous way of applying it has helped the university. Newell seemed like my childhood hero, the Lone Ranger. When there was a problem to be solved he would make himself fully present, do something brilliant, then ride out of town to return to his first love: research. He was a continual, background coach for Nico Habermann, the long-time head of the university’s computer science department.
When the department was chafing under the science college, Newell spent a few years helping Nico persuade the university to elevate it to become its own college—a major political achievement that was later adopted by other universities.
Late in the 1980’s, Newell called an ad hoc campus-wide meeting to initiate a research program in human-computer interaction. There was an overfull room and great enthusiasm. But I was leaving Carnegie Mellon to start MAYA Design, a commercial version of the same idea. Newell and everyone else went back to their individual research, so nothing happened. This was the only time I recall Newell being a little angry at me.
Newell was dying of cancer in the summer of 1992. I broke out of our professional relationship to say, “You know, I am really sorry this is happening.” Newell joked, “Not nearly as sorry as I am.”
After he died someone commented, “Our giants are gone.” The college had depended upon Newell for his superb leadership for so long that its future seemed uncertain. Raj Reddy, then the dean of the college, asked me to come back and I did, becoming the head of the computer science department.
My first move was to create and disseminate a list of principles Newell practiced as a small way to perpetuate his influence.
Do what you love; love what you do. His incredible energy and enthusiasm sprung from this.
Help others to do what they love, no matter how different their choices might be. Because he was happy and secure in the rightness of what he was doing, Newell was open-minded about what other people did, and often helped them.
Don’t worry about how intrinsically smart you are or anyone else is. I never saw him feel threatened by another person’s brilliance or offended by their lack of it.
Be intellectually tough – uniformly on everyone. He only judged performance.
It’s okay to take time out from research, but go all in and make it pay off. When he dropped artificial intelligence for a time to collaborate in the invention of a hardware design language, they wrote a well-received book about it.
When a technical disagreement arises, stop arguing and devise an experiment to settle it.
Set the research agenda, but let the students choose the methods. Often, Ph.D. students would be supported to pursue approaches that differed from those of their thesis advisors. Sometimes, they were right.
Don’t expect an institution to have a heart. Only people have hearts.
My second move was to begin the process of creating a human-computer interaction department, partly based on the model we had followed at MAYA Design. It was supported by several of his colleagues and students from different parts of the university. In the following 25 years, it produced graduates who led industry and other university departments.