acm - an acm publication

A conversation with Herbert R.J. Grosch

Ubiquity, Volume 2001 Issue December, December 1 - December 31, 2001 | BY Ubiquity staff 


Full citation in the ACM Digital Library

Reflections on the early days of computing, the importance of standards, and The Old Man.

Reflections on the early days of computing, the importance of standards, and The Old Man.

Dr. Herbert R.J. Grosch has done pioneering computer work with IBM, General Electric, the Federal government, and other organizations. In 1945, he was drafted into the new IBM Watson Lab at Columbia by Los Alamos to provide backup for bomb calculations. He is known as the author of Grosch's Law, which describes the relationship between computer processing speed and cost. At different times in a long and illustrious career, he has been active in celestial mechanics and optical engineering, managed IBM's space program, taught some of the first computer courses, and worked as an international consultant. He is a charter member and former president of ACM, was a consulting editor of Datamation, was on the editorial board of Management Technology, and was editorial director of Computerworld. After living and working in Europe for many years, he recently returned to the United States, where he was named Adjunct Distinguished Professor of Computer Science at the University of Nevada Las Vegas.

UBIQUITY: You're certainly among those people in the "who's who" of computing. And you clearly also have many friends and associates who collectively form the "who's who" of computing. Tell us about some of them.

GROSCH: That's a good starting point. I like to talk more about people and less about abstract languages and things of that sort. I have opinions about the latter but they're not popular ones. Way back when, I remember writing a piece essentially denigrating COBOL in its formative years, which has come back to haunt me a good many times since, as you can imagine.

UBIQUITY: Why has it come back to haunt you?

GROSCH: In the sense that COBOL is the second most durable language. During the millennial flurry, an awful lot of people scurried around looking for old COBOLers. Unfortunately for many of those people, they found the COBOL programmers and spent a lot of money with them. It turned out that there wasn't much to worry about after all.

UBIQUITY: Were you surprised that the so-called Y2K problem didn't turn out to be anything?

GROSCH: I was in Europe during the time when the big flap was supposed to come to a crisis. The two stories that I get are, on the one hand, many people made precautionary changes, reviewed their antique spaghetti code, fixed it, and as the result when the bad time came, they survived. The other story is that the whole thing was a put-on -- that there wasn't a problem -- and crude commercial forces operated to sell cures to things that weren't really broken. My guess is that it is a mixture of the two; that indeed many of the fixes were worth the money spent on them, but that there were many people scrambling around trying to make a buck in the inimitable American fashion.

UBIQUITY: As you think back on all the illustrious folks you've known, pick one that you think was particularly interesting, important and/or colorful.

GROSCH: To me, the great one was IBM founder Thomas J. Watson. What made Watson Senior a hero in my eyes was his decision to back the original IBM mainframe for mass production. It's clear that Watson's decision to do this started the whole ball rolling. Obviously, technology as a whole, and the pressures of our Western culture, were going to produce something like computers in a few years, but Watson certainly stepped it up. Jim Rand, the president of Remington Rand, was jealous of Watson Senior and tried to catch up to him at the beginning by buying Eckert-Mauchly.

UBIQUITY: You're referring, of course, to J. Presper Eckert and John W. Mauchly, designers and builders of ENIAC, the first completely electronic computer, who went on to develop UNIVAC for Remington Rand.

GROSCH: Right. But it wasn't really Rand's initiative. It was Watson Senior's foresight, which Rand was anxious to catch up to and, if possible, exceed. Which, of course, he did for a little while. But then the Watsonian vigor soon overtook him. The Rand empire, which included a bunch of funny old round-hole, punched card guys in Norwalk, Connecticut, as well as Eckert and Mauchly, wasn't up to the task of beating out IBM.

UBIQUITY: Should we be surprised that the first person you mention as important is someone who normally people would think of as a business person?

GROSCH: But he isn't my hero because he was a businessperson. In fact, much of the enormous growth of IBM came after he handed the reins over to Watson Junior. That growth, by the way, was pretty much automatic. Watson Junior was simply the guy sitting in the high seat when it happened. Let me put it this way. The Old Man practiced -- without consciously stating or perhaps even recognizing the idea -- what Tony Blair, the prime minister of England, used to call "stakeholding." Blair dropped the concept like a hot potato as soon as he got elected and started commercializing himself for reelection. But the original stakeholder concept was that the privileges extended to large organizations, and especially profit-making corporations, by the broader society must be paid back to a variety of destinations. You had to pay back the shareholders who furnished money. That, of course, is the dominant factor today and almost the only thing that anybody talks about. But you also had to be good to your employees. You had to be good to your customers. You had to be conscious of the arts and the sciences, which the old man was pushing at the time that he made these decisions. Instinctively, although it was not a big thing back in the '40s and the '50s, you had to support the environment. And you had to support the organizing forces like government by paying healthy taxes. A great deal of the success of IBM was due to the fact that he sent highly motivated young employees out to work on highly motivated customers and bring back lovely orders; that fitted the whole concept remarkably well.

UBIQUITY: Were there any other technology business leaders who bought into the stakeholder concept?

GROSCH: Most of the others were narrower than that. People like Seymour Cray, for instance, cared only for the technology. They were perfectly willing to plow corporate profits back into more R&D and better componentry. But they didn't give a damn for the other parameters. Notably, Seymour didn't care at all about customers. He regarded them as a big nuisance that diverted him from his vision of what the supercomputer of the future should be like. And this in spite of the fact that a very large part of his success was due to one or two customers, like Lawrence Livermore National Laboratory, which survived in his esteem largely because they did what he wanted them to do instead of the reverse.

UBIQUITY: So you're suggesting that Watson Senior had more than a technology vision and more than just a business vision but also a social vision?

GROSCH: His broad vision was that of a prideful citizen of American culture. He wanted the Metropolitan Opera to be the best opera in the world. He wanted the Metropolitan Museum to be the best museum in the world. He wanted Columbia to be the best university in America. He also belonged to international organizations. He was head of the International Chamber of Commerce for a few years. It was a broader view than just profits. The profits came in vast quantities. But profit wasn't the driving force with The Old Man that it is today for other business leaders. He wanted everybody to do the best of everything and especially IBM to be the best of everything. He thought the money would take care of itself. And it certainly did, until close to the end of Watson Junior's regime.

UBIQUITY: And you don't think that's the way it is now?

GROSCH: Watson Senior used to announce these feelings in, for instance, Think magazine. Of course, a lot of that was written for him. He had a whole floor full of people grinding out everything from directives on how to run the International Chamber of Commerce to replacing the manager in Anchorage who let a good customer get away. Among them were people who wrote things for him that he would not have originated himself, although he certainly approved every single word that was published.

UBIQUITY: Compare computing as it was in the days of heavy metal, the large machines, and what it is now. Is it basically the same business and the same technology?

GROSCH: In the early days, there were a few hundred people who were real honest-to-God professionals. They exchanged information freely, much of it informally by telephone calls and meetings and in slightly later years through the pages of Datamation. They knew what they were doing and they knew when they had had a success and when they hadn't. They varied from tremendous figures who are more or less forgotten today, like Bill Bell and John Lowe, who were highly professional and interested in the realities of things. They kept IBM honest. When IBM tried to sell them a bill of goods, they'd jump up and down and usually get it straightened out, and in the process educate IBM a great deal about what their machines could do. These people operated two classes of big installations. People like Bell, Lowe, myself, and two or three others, operated closed shops. That means that we furnished most of the programming. We did all the operation. We used tightly managed and well written applications programs to do the work. We charged our services by the hour, including our human adjuncts.

UBIQUITY: And this group of honest-to-God professionals compares to what other group?

GROSCH: The other class, which was somewhat larger and which was exemplified by places like Rand Corporation, had very clever user people who were not necessarily expert programmers. They ran what were called open shops. In an open shop, you used an overlying language. One of the earlier ones came out of Los Alamos, for instance. The customer himself was responsible for the application. That led to things like COBOL and FORTRAN. It meant that the number of useful operations performed per hour dropped off a lot because these magic languages used up a great deal of computer power. On the other hand, it meant that you could have a smaller professional staff. And if the customers were clever -- and certainly they were clever at Los Alamos -- everything went pretty well. The big jump to electronics generated so much enormous speed compared to relays and punched card machines that the fact that you could have jumped still further if you were a John Lowe or a Herb Grosch, was pretty much hidden from the top management.

UBIQUITY: And what is the present situation?

GROSCH: What you have now is an enormous bunch of open shops -- millions of PCs that are used badly by inexpert users, in most cases. The average person is just plugging away on e-mail and clicking on those cute little icons.

UBIQUITY: What happened to all the experts?

GROSCH: They've been drawn into places like Microsoft. What you've got now is a kind of closed shop at the top end in the sense that the customer comes and petitions Microsoft and its few rivals to do good things. Then the thousands of programmers hidden away in the Northwest do it if Bill permits them to.

UBIQUITY: The result being?

GROSCH: The net result is that the sheer power of the chip inside is diluted a thousand-fold or ten thousand-fold. Every time you click on the blue band and move the thing around on the screen, you're wasting millions of operations. However, if you didn't do it that way, you wouldn't have millions of customers because they wouldn't be clever enough to run the PCs the way that we used to run the IBM 701s, 704s, 709s and so forth. That's an enormous transformation. I don't think of that as being a single person's decision in the sense that Watson decided to build 701s. I think of it as a cultural and business phenomenon, one that I regret but obviously can't imagine an alternative to.

UBIQUITY: Aren't there enough computing cycles now readily available to run these millions of operations? Is there still a need to be parsimonious?

GROSCH: It certainly isn't clear that we could use all the power that's available if we went back to something like the professionalism of the early days. If you could transform every one of the millions of PC users around the world into an expert overnight, I think the industry would probably come to a halt because they wouldn't need better computers for a long time, they'd have so many extra cycles available.

UBIQUITY: What are your thoughts on a broadband Internet for major users?

GROSCH: Presumably the users would link themselves together and have super power and super bandwidth. That will probably come in a year or two. I hope that it will reach my university so that I can have access to it. But the fact of the matter is that most of the stuff they want to send over such a network isn't worth sending. The problem with excessive power and enthusiasm is that there is no arbiter elegantiarum saying that it isn't good taste to waste time on computer games, especially the crunch-and-bash, let-'em-bleed-to-death-on-the-screen computer games.

UBIQUITY: Are you suggesting that bad use drives out good use?

GROSCH: It's a question of how much you want to steer this broader society. If you have strong feelings, as I do, about right and wrong, economical and wasteful, and clean and dirty, then your vision of what you can do with computers is different from that of people who see it as an opportunity to sell books or dirty pictures or what have you. Not that I have any objection to dirty pictures, you understand. Pornography has pioneered half of the major technology in use today. In fact, it is probably the only overall commercially successful business on the Internet today. But putting aside any question of culture and just talking about wastefulness, let me tell you a story. This has not been in any of my biographies and stuff so far because I haven't gotten to it yet.

UBIQUITY: What's it about?

GROSCH: Traffic. I had a visit when I was a big shot at the Bureau of Standards from two guys in England. I took them to see what was the early beginning -- Arpanet. This was when Larry Roberts was at the Pentagon. The year must have been 1968. At that time, there were about eight or nine nodes, universities plus the Pentagon. We had a nice time and I kept my mouth shut about some of my doubts. And Roberts charmed us. The visitors went away absolutely dumbfounded at all these wonderful things that were happening. As we walked through the Pentagon parking lot to my car, one of them said to me, "I didn't quite understand one thing, Dr. Grosch. And that is, what is this 'traffic' that's going to move over these lines?" I laughed and said, "You put your finger on a very vital point. In fact, 99 percent of that traffic is going to be information about the network. They're going to be upgrading it, changing it, adding nodes to it, and exchanging information about its performance. That'll keep the network pretty busy. There may also, incidentally, be some actual work done over it, but not much."

UBIQUITY: And today?

GROSCH: That's still what it's like today. If you look at this enormous power, these zintillions of cycles generated by millions of machines all over the world, most of it isn't doing anything very important. There are some very useful things like e-mail, for instance, that clearly are vital and will continue. But a lot of stuff like game playing and dirty pictures and so forth could slip away without anybody really worrying very much. It won't but it could.

UBIQUITY: You mentioned a time when you were at the National Bureau of Standards. If you were in the government today and were in charge of technology policy, what area would you want to change?

GROSCH: I would still want to be focusing on standards. Mind you, I am somewhat prejudiced, having been a big shot in that area for a few years. But when I look at these incredible wrangles that go on between opposing corporations and opposing societies and opposing forces that make it four to six years before you can bring out something like the G3 television-on-a-handset kind of thing, it's terribly wasteful. If there was some way in which you could exert government forces on that without destroying the vitality and prosperity of the whole process, I would certainly advocate it. I was trying to do that in a minor way when I worked at the Bureau of Standards. Of course, it was a complete failure. And in one case where I managed to have some tiny fourth-decimal-place effect, the industrial response was so strong that I was removed from the office.

UBIQUITY: Why was it a failure?

GROSCH: Specifically what happened was IBM announced a thing called the System/3, which was the most nonstandard computer system that they could devise. Everything about it was different. It was to use a nonstandard physical punched card with non-familiar holes and an unfamiliar code. As originally announced, it was to have no input/output mechanisms that would read ordinary cards or magnetic tapes or anything like that. They wanted to sell 50 million bucks' worth of these over the life of the system to the federal government. I stopped it with one letter. And three or four weeks later, IBM sent down a delegation from Armonk and had me removed from the job -- not, I must say, to my surprise. The fact of the matter is that there are very few places in which one can exert that kind of force. And when you do, the backlash is tremendous. A follow-on to that story is that the System/3 was a failure. IBM discontinued it, and the federal government benefited enormously by not having purchased any.

UBIQUITY: It's just a little surprising that you give "standards" as your choice for a new initiative -- even granted your long-term interest in standards -- because we had been talking about your disdain for the inefficiency and frivolity in the use of computing.

GROSCH: It's inconsistent of me. I'm perfectly aware of the fact that I'm saying standards will be good because they make the machines function faster and on the other hand, saying they're too fast already. What we need to do is to use them more effectively, which has nothing to do with standards. I'm aware that those two are in conflict. But this is true of a great many of the forces in our trade.

UBIQUITY: What are some other examples of conflicting forces?

GROSCH: For instance, we have this magnificent technology opening out in front of us. The hardware is incredibly good. This is something that we can be proud of as human beings. It's just absolutely tremendous. But what we do with it concerns me a great deal. We didn't feel that way in the beginning. Let me cast back to the open shops, the closed shops, John Lowe and the 701s and all that stuff. It was a golden time. These people were enthusiastic about what they were doing. They had absolutely no doubts about the value of either the technology, the tools that they were developing or the projects they were using them on. They were designing airplanes and working out orbits to go around the moon and stuff like that. Even somebody who was developing the first versions of Sabre, the airline reservations system, was doing something that needed doing. It was right at the edge of what applications people could do. Even with all of the effort that American Airlines and IBM contributed, both for money and for prestige, it just barely squeaked through.

UBIQUITY: And have things changed so drastically since then?

GROSCH: I think so. Today, thinking about that the requirements of those first airline reservation systems, they were fairly simple. When you talk about the Reagan-Bush Star Wars shoot-down-enemy-missiles, that's a hundred times harder to do. In fact, both Dan McCracken and I, as presidents of ACM, and many other people equally well placed, said over and over again, it cannot be done. It is not possible to guarantee results because the problem doesn't hold still. While you are solving the problem, the enemy is changing the specs. They're redesigning their weapon. They're using different kinds of decoys. They're doing this, they're doing that and they're doing the other thing. By the time you hear about that and change your design to allow for that, they're off on something else. When the button actually gets pushed, chances are it will be a failure. When you think about how hard it was for very skilled people to create an airline reservations system, you can see why the FAA can't revise air traffic control methods for the United States. The pattern of activities changes too rapidly for them to keep up with it.

UBIQUITY: In 1945 you were recruited by the then-new IBM Watson Lab at Columbia to provide back-up for bomb calculations. You worked with John von Neumann and Richard Feynman and others. What was that like?

GROSCH: The Watson Lab is a fond memory in my mind. In its first few years it was wonderful, just terrific. We had a good link with Columbia and this, of course, was helped by the fact that Watson Senior was the most important trustee of the university. We had a knowledgeable director, Wallace Eckert, who not only had the Columbia link but had many acquaintances up and down the physical science communities, the physicists and the chemists and the astronomers and so forth. We had bright people like Hilleth Thomas and Leon Brillouin. Right around the corner we had guys like By Havens designing the NORC. So you had, on the one hand, my computer shop doing applications, some wonderful people thinking deep thoughts like Thomas , and Havens & Co. turning them into hardware for the benefit of the future IBM. It was a wonderful scene.

UBIQUITY: What happened?

GROSCH: Gradually, we all just went away. I left to do applications work outside the New York area for IBM. Some of them, like Wallace Eckert, got old and retired. The interest in physical science, laboratory bench science, chemistry, metallurgy and so forth caused the lab to move to a bigger building. And then the rivalry with Yorktown Heights and other places grew. As I understand it, in the end, they were all sort of absorbed. But it was a golden time while it lasted. We had people coming through with problems in their hands that were just wonderful to deal with -- people doing biochemistry, x-ray crystallography, oceanography, my optics, Eckert's astronomy.

UBIQUITY: What was unique about the Watson Lab that allowed it be so successful and inventive?

GROSCH: That was due almost entirely to The Old Man's vision. It was a vision that was broader than just commercial vision, the sort of thing that Jim Rand had. In early 1945 he saw the moment when Los Alamos came to him and asked for help as an opportunity to do three or four things at once. In the first place, he could get some priorities to have his own machines because IBM couldn't get a keypunch out of the priority system without something like Los Alamos or the Air Force backing it. It gave him an opportunity to influence Columbia, which he was keen to do, as a trustee. It gave him an opportunity to do good things for science in general. All of these things came together at a time when he was champing at the bit, so to speak, with wartime restrictions. So -- boom!-- he pushed a button and pretty soon Eckert and Thomas and others were milling around (and, of course, he himself coming fast down the road) to create the SSEC, the second giant IBM machine and the second giant electronic machine. The first IBM machine, of course, was the ASCC (Automatic Sequence Controlled Calculator), also called the Harvard Mark I.

UBIQUITY: Was that group particularly collegial? Was it a fun group?

GROSCH: Yes. We were in a very small building and it was a very small group of people. The collegiality was promoted by, for instance, the fact that two or three of the seniors among us ate regularly at the men's faculty club at Columbia. We interacted with the physics department, the math department, the chemistry department and so forth in that casual environment. We had warm relationships extending to Columbia -- warmer than might have been expected considering that we taught some specialized courses that didn't resound very deeply with most of the university. We were members of the university community in the good pre-1960s sense of the word. We were a happy nucleus within the larger rather happy university.

UBIQUITY: Bringing it up to the present, you're back in the university now.

GROSCH: Yes. Any day now, I'm going to be made an Adjunct Distinguished Professor of Computer Science. Now, that and 35 cents buy you a phone call, the key word being "adjunct." I'm being rather warmly welcomed. Whether they're preparing to put me in a glass case and exhibit me at the fancy new library or use me perhaps a little more effectively isn't clear. Anyhow, I'm very pleased to be back. I'm reluctant to have left Europe. I prefer the European scene considerably to the American scene. But I just couldn't cut it.

UBIQUITY: Why not?

GROSCH: I had two multi-year NSF grants, for which I'm very grateful, but they came to an end. The fact of the matter is that, without them, I couldn't afford to live overseas. Whereas here, I still have links. I can exploit my historical presence more than I could in Eastern Europe, certainly. I was a failure in the thing that drew me over there. I regret the failure but I've tried to pass it on so that it will be done better by somebody else.

UBIQUITY: What were you working on in Europe?

GROSCH: Most recently I was at Latvia University in Riga. Latvia University has an enormous collection of antique Soviet information technology hardware. By enormous, I mean several big rooms full. There are teletype machines and tape typewriters and punched card machines, vacuum tube machines and a few transistorized machines. The year 1991, when Latvia declared independence from the Soviet Union, was a terminal point. After that the collection tapers off. I tried to open that collection up to the Western museum world. I told people at the Science Museum in London and the Heinz Nixdorf Foundation Museum in Paderborn, Germany and the Smithsonian all about it. They were waiting for Latvia University to say, "Come and get it." I went to make LU say, "Come." And it didn't work. Their impression of me was that this white bearded old bastard is trying to steal their machines. The fact that they're doing absolutely nothing with them and they're just rusting away in this attic is less important than the Latvian pride of ownership. So I was a failure on that one. But the interesting ACM-related story about it is that I passed this on to an ACM member named Imants Freibergs, the "s" on the end being the masculine ending, who is married to a woman named Dr. Vaira Vike-Freiberga, that being the feminine version. Vaira is the President of Latvia! Freibergs is now a member of the computer science faculty, the outfit that kept turning me away. So I turned it over to him and said, "Dear boy, do what you can with this."

UBIQUITY: Well, one never knows. What are you working on now?

GROSCH: I've been doing two things. I had support from the Smithsonian and a grant from the National Science Foundation to record the early years of Datamation in detail. I've done some of that but only half as much as I had hoped. In addition, I have done the second third of my autobiography. The first third is in hard cover but out of print. The second third is available on CD. Now I'm working on the third part. The third part includes my Bureau of Standards experiences and my combats in ACM Council and the ACM presidency, and helping spring Natan Shcharansky, who is now an evil force in the Israeli government, from Lefortovo prison in Moscow, and a few things like that. They're good stories. But I think, from the point of view of NSF and the history community, the early stuff is the most important because there were so few of us and we're dying like flies.

UBIQUITY: Are there many stories of the early days of computing yet to be told?

GROSCH: The Annals of the History of Computing is full of great stuff, including many things I never heard of, coming out from old timers who write a letter to the editor or a department or something like that. The trouble is that it isn't strung together. Every once in a while, there's a special issue and its articles interrelate and give a wonderful picture. The late Cuthbert Hurd did one, for instance, that I thought was admirable. There have been some old manuscripts published like the von Neumann notes for the 1946 summer conferences that were so influential. I have dreamed about the possibility of stringing this all together in a better way. My boss here, Hal Berghel, is enthusiastic about this and, in fact, had already made some moves in that direction before I appeared. So I may get to contribute to that. It's the beginning of the beginning. I've only been here three weeks. So I certainly don't have a lot of detailed plans.

[The End. Alternatively, The Beginning]


Leave this field empty