Monument Square Association, Colonial Inn
Concord Oral History Program
Renee Garrelick, Interviewer.
Science & Technology Segment
Ken Olsen, a pioneer in the computer industry, reflects on his career and the days when computers were designed for the problems they were to solve. Today he observes the race is for faster computers and networks, more software and a large database that is unnecessary for many users. Ken Olsen began his career at the former Digital Computer Laboratory at MIT Lincoln Laboratory and his remembrances supplement oral histories that have already been done for Concord's program with Jay Forrester and Robert Everett.
Several weeks ago I was talking at the Sloan School at MIT and I started off by saying "We've lost something and we've forgotten the scientific method." And afterward the professor said to me, "You know that's right. We don't have time for it any more." That's true for a large number of parts of many of our universities. When many of you were in school, it was still the basis of education.
Scientific methods in the dictionary are described something like this, you define the problem, study it, analyze it, and then make a hypotheses of what the solution is. Many of you tested by experiment or by analysis and from that you conclude, always tentative, always learning to get better, but you never have the absolute. Today the scientists, the academics are politicians Â¾ everything is absolute and it's my way. Now in the computer business, the cover story or one of the cover stories of several computer magazines, and there are about 200 of them, said "Is science serving computing today?" This is when in a company you have all the present use tied together to one machine servicing them. They came to the conclusion some say yes, it's dead and some say no, it's still got hope. Why would something as obvious as everybody having a computer and organizing so that it works together, why could that be dead? I think the answer is we forgot what we should have known by the scientific method. Where's the problem?
If you listen to people in the computer science department at MIT, who happen to be some of the better people on this earth, or if you read almost any magazine or almost any book, they say the answer is faster computers on every desk with faster networks and more and more software to do everything automatically and a large database. Everything in the whole corporation is on one database and everybody has access to everything. With these fast networks and these fast database machines, they gather all the information and plot graphs. I'm not joking at all. When you ask them what's the problem, they say we have the answer, we don't have to know the problem.
Now I ran a fairly large company and people were shocked that I didn't have a computer in my office. I use a computer all the time. I tell my secretary what to do with it. When I want to find something that would be very useful, I call my son and say "I paid for your Ph.D., find it for me."
You look at any small corporation and you have people that take catalogue orders, people who write letters, people in the library, do they all have to have access to all the data of the corporation? Do you have to have faster and faster computers?
One of the things I'm doing in saying this is tell you what I tried to do at Digital and try to do at my new company called Advanced Modular Solutions. You can accuse me and rightly so of adding a philosophical meaning to something that hasn't a philosophical meaning. What we started to do at Digital 40 years ago was to do those things with computers to solve problems. When the world was whole hog after chasing speed for no reason and no understanding, they fired me because I wanted to solve problems. In general there is no interest in trying to solve problems with computers because they are so complicated today. Typically a network in a corporation is absolute.
Back in 1950 I went to work at the field lab at MIT with a Whirlwind computer. It had 10,000 vacuum tubes and it was different from the other computers and there were only three or four computers in the world. But all the computers were made to program mathematical tables usually for artillery. They'd sit and grind out artillery tables on page, after page, after page in order to reach a decision. When we started in World War II, they wanted a wind tunnel which is a physical phenomenon. You couldn't measure anything very precisely and they started off with an analog computer that was very fast. The war was over by about five years, and it was one of the ideas for aircraft control. In September 1949 if I remember correctly the Russians set off the first atom bomb and took the Americans by surprise. The Americans were terrified because they had no defense. It was to do intercepts in the sky above and the fighters never got close to the bombers. At jet speeds they just couldn't do it. What were we going to do? The Russians could come over here and we couldn't do a thing about it. What a panic! The obvious thing was the Whirlwind computer. In September they exploded the bomb and in March they had a telephone link between Cambridge and Bedford airport which they called slow down video, we would call it a modem. You couldn't get through an analog computer pictures on telephones so they made a ditto form on the telephone and then the Vax started the Whirlwind computer for air defense. Computers ended up with 60,000 vacuum tubes in them. They had 23 of them around the country and in Canada and they had some interesting characteristics. The tubes were designed to have 500 hours of life with 60,000 tubes. It didn't last very long. A number of things they learned at MIT. One was how to make computers work reliably and another one was how to design magnets.
The old scientists knew a specified problem and they had a broad education and they could analyze it from a broad perspective. These people, the scientists in the ‘30s, had something that we've lost since then. The interesting thing about the world of computing in regard to the air defense system was that all the characteristics of a personal computer was 16 bits, a cathode ray tube and a keyboard. It didn't have a mouse but it did have a joy stick. It didn't have a disk drum and it did all the things we do on a personal computer. It didn't have much memory. There were two banks of 256 words of memory. I don't think both banks ever worked together. But it was a personal computer and it was a military device that MIT did have and they still had an openness and a trust which is very important and they allowed some students and some staff to use that machine to learn what to do. They learned how to use a personal computer which was absolutely different from a number crunching slow 64-bit computer.
We started Digital for several reasons. The old computer lab of 100 people was run by Jay Forrester and Bob Everett. They had a group that was motivated, worked as a team, was intellectually challenging, had tremendous pride and got an enormous amount done. Trying to build on that and despite the personal computer we started Digital. The idea was to build a small inexpensive computer. The first computer we built was $120,000. When computers cost millions, those were real dollars. We had the idea that computers were supposed to be sacred. You don't have fun with one, it's immoral. They're supposed to be protected from people.
We ran it like you would expect academics would. After a few years of 20 people in a room and it didn't work very well because they all had wonderful ideas for spending money and I was the only one that had a P&L statement in mind and I would have to say no. They went to the Board of Directors and said "Ken is a dictator; he won't let us do anything." They had really wonderful marketing ideas like let's print matchbook covers. They had really good ideas and you don't want to be the only one to say no. But we started a $14 million company building computers much too big for the size of company.
One man cannot run everything. He can't understand everything and it is interesting how unnatural this way of organization is. We're taught by every authority at MIT who say "We're going to teach you to make decisions." Well, what that means is that they're going to teach you to tell other people what to do. I thought the idea was again give people responsibility and then people are motivated to do a job, excited about it, and have fun doing it. I was more shocked than ever to how unnatural it is to do as to how Alfred Sloan preached which was give responsibility to each group, and don't interfere all the time. Everywhere we're taught - interfere - tell them what to do, you're boss. An author said a few years ago and the Chinese philosophers said something like this Â¾ a leader is at his best when he is hardly noticed. When the job is done, his role is fulfilled and people would say we did it ourselves. That's what the Chinese said years ago. And it is mathematically so obvious to concentrate on their business albeit small. The boss can concentrate on his job as leader, keeping things going, but he can't do everything. It is almost impossible to get that across, but miraculously we did.
Before this I had a bunch of people and they were so dumb. They couldn't understand anything. It's amazing how smart they became. We either never communicated or we weren't that smart. Every one of them did very well because they had responsibility. One of the secrets was they had a budget. I innocently would say "Every quarter you meet your budget, ah, make it the budget for the year, make it 3/4, that's all." I knew I was pulling their leg. That budgeting, the only word I could use to describe it, was travail. You never have enough money to do all the things you've got to do. I don't know how it is in your church, Father, but in every other church I know, there is never enough money. But travail is what made them work. They had to analyze everything. And one more thing, they had to share resources with others. They had to share products, share sales people, share facilities because their budget would never work without it. With that they really grew and developed.
When China opened up, all the business leaders went over there and made deals with the Chinese, everybody except me. I was over there at the end of World War II and so I would have felt at ease there, but I never went. You don't have the top man go to the Chinese bureaucrats, you know nothing would come out of it. We sent our Chinese people to Hong Kong two weeks at a time and quickly we were the largest computer seller in China. It is obvious the Chinese didn't trust them, we needed to improve that trust to do well. If I'd gone over there and I did eventually and took all the credit.... This story I'd like to say is a serious story. It is the way you should manage things.
Every department knew what their job was. They don't worry about anybody else and nobody at the top of the ship ever said they were going to run all those departments. They had a magnificent attitude. It ain't going to be our fault. Another interesting thing about the military is most communication doesn't go between generals, doesn't go between divisions or the head of the companies, it goes through the sergeants or the lower level. And you have to think that that doesn't work. That's an obvious way to run a business. People at higher levels have a responsibility. They take care of the decisions.
Another story, after I was fired from Digital and before I left, I visited a group that does special systems up in the New Hampshire plant. They were complaining to me about all the rules that they had to deal with. They couldn't get things done because a good job in the military is the military didn't do standards so they couldn't take a military job and the company wanted to integrate some of their things into our computers. We couldn't do that because they didn't follow our standards. I knew where it was heading. One way or another you get the job done. This is what made Digital, people taking responsibility.
The thing I never figured out is how to survive success. You know sports figures, politicians, even preachers, often and most of the time fail at success, and even this way Al Sloan said to run a business failed at least underneath. The late ‘70s were years of great success and people started to think you're wonderful. They delegated their budgeting to a spreadsheet or a financial type to no avail. They thought they were the greatest people on earth, just being there was a success.
Then I had our senior technical people who mostly sit hour and hour, day after day in meetings here and there develop a common strategy. After all that work the strategy we decided on was exactly the one the senior vice president had announced five years earlier but dropped. We would have one computer, one operating system, one networking system. After he pronounced it, he ran off in all directions. He got nothing done in five years. He spent that year developing process but he never did anything wrong. He never did anything. Everybody working on a common goal.
The other part of the history of Digital which I was suggesting as the first part of the philosophical meaning or scientific approach is this high flalutin title on how to solve a problem. It is so unusual today in a new company. One of our people came out to visit a potential customer and said "Well, what is your problem?" The customer said, "What do you mean? You're the only computer company that has asked me what my problem was." I probably enjoy being contrary, and the Globe enjoyed beating me up for it. If you really want to create interest, you've got to do something different and you've got to do something with a little audacity. Find something the rest of the world has dropped.
Back in the early years of Digital, the analysts and reporters and the financial people were so insistent that we do the latest fad. I got a call once from a financial analyst and he said, "Do you use positive logic or negative logic?" I said "We use negative logic." He hung up. Hopefully it means trying to solve particular problems. The computers we did in the beginning were different. We gave one of them to MIT for the students to use. Those two machines were sitting on the second floor of a building that had a main frame on the first floor. Any student could use it for anything, sign up months ahead, 24 hours a day. And weird things happened. Kids stopped eating, they stopped washing, they stopped going to class, they got so involved. Hacker then was a good word. Out of that group came a feeling for what that kind of fast, simple computer could do. You could walk in there at 2:00 in the morning and they would type in gibberish and out would come gibberish and they would all roar in laughter. They invented word processing, a more expensive typewriter, they invented computer games, and the best one was Space Wars. Space Wars took in the concept of physics of space travel, space craft with a tight bank and a sharp turn in an atmosphere-less world. Most of us couldn't get used to it because we were riding with velocity and acceleration. One day somebody came by, sat down and played it beautifully. Also came out of that was time sharing. Somebody types something, he gets an answer back, but it takes him a while before he types his response. It had time for something else to do, so they put number stations together and that was time sharing because many people were using the same machine. Now that became networking and in effect that computer was the first server. That made a huge contribution.
By the way, my first view of those time sharing machines was everything worked. Nothing fancy, black and white screen, anyone could sit down at any secretary's desk, take up their mail, answer their mail, go back and read it. And it always worked. I could dictate something early in the morning and by 9:00 it would be all over the world if I wanted it to. Always worked. Now nothing works anymore. It's fast, expensive, and maintenance is enormous.
Remember just a few years ago something called artificial intelligence came out of MIT. They knew those words would obviously intimidate people. They said it would revolutionize the world. Everything will be artificial intelligence. The Japanese believed it. It's good and it does unusual things, but it doesn't do everything as promised.
Remember also a few years ago everything was Unix. And all software on the Unix would run on anybody's machine. I once said very carefully, to phrase it carefully, if you believe putting the name Unix on something makes it standard, you get snake oil. The Lowell Sun said "Olsen said Unix is snake oil." Another time I said, "I don't believe people want their lives run by personal computers." If you steal something from the refrigerator at midnight, you want to enter it on the computer so it won't be on the menu tomorrow, we just don't want to do that, and the fad was everything is run by computer and people said, "Ken Olsen is against computers at home."
At Digital a few years ago we were in one very real sense a service company that also made computers. The overwhelming part of Digital was training more hours than any of the largest universities Â¾ teaching ourselves and teaching customers. In a sense you can't beat the marketing. They pay you $2,000 to come and listen to you for two weeks 8 hours a day telling them about your products. Now that's good marketing. That wasn't traditional marketing. Traditional marketing was selling something he doesn't need and doesn't want. And that was a big operation. Then when customers had to use computers to do miscellaneous things, it took a lot of consulting, an enormous amount of consulting and they'd charge exorbitant rates. They support our people. In most of the large companies there is a fairly large contingent of Digital people supporting and they have access to the place just like an employee. In the sales operation we assigned a team in every large company and they were part of that company. They worked for the success of that company, and they sold them Digital products. The trust they had was complete. They had the run of the place. If somebody wanted to know what was going on in the company, they would call the Digital people because they would know everything. The interesting thing about Digital when things were going well was it was a service company.
Computers were competent and good. They were never designed for manufacturing. Computer scientists don't like electricity and they don't like mechanical things and they don't want to be telling you how to do it. But our products were very big money because of this. But the interesting point of this is it was a service business and you're trusted by the customer. The computer scientists said all of that is in the way, let's get rid of all that and make computers. One of the directors used to head up Ford. He said "At Ford we never even talked to a customer, we never even saw a customer. Somebody else sold them, somebody else fixed them, that's what Digital ought to do."
The important thing here is people and how they use their computers. One should obviously, even though you may be called stupid and dumb and irrelevant and not in the real world, but you need to look at every desk and decide what do you need for that desk. Fill that need and a little extra, but don't do like so many people who say "We'll have the very fastest and the very best at every desk." There is no logic in that and the cost is enormous. People discover the cost of computers and how you can be swamped in just one year by the support. I've been preaching this for seven years now, on the desk is what they need and no more, keep it simple. The other idea we are pushing in the new company is you use only company software. The last thing you want is to have somebody pick something off the Internet and say "I'm going to try it out." That is dangerous. We bought a box of new floppy disks, brand new, wrapped in plastic and viruses were already in there. A virus could wipe out a whole company, it could destroy everything. So only company software. There is an argument going on, don't people have a right to work with other software during working hours? Sounds stupid, doesn't it? Don't you trust them? No. I tell people if they had access to anything else except company software, make sure that tube is always facing the corridor. It might be nasty things or it might just be Solitaire. We have a little box, it's heavy, it's solid steel, it's got no disks. You can't even enter software into it. The software comes with the company. Whatever software you want comes with company software.
This idea I have been pushing causes emotions you wouldn't believe. I was criminal and I was called senile and old and beyond repair with this kind of idea. You have one computer and you put all the applications on it. The result is you have a huge computer with many applications. If they want to change the software, they have to prove it a year ahead of time because they put it in for the others and it might ruin the whole company. The terror they go through. The obvious thing to do is have a separate computer for every application. You cannot believe the emotion that this idea can generate. We always have had everything on one machine. Heaven must have meant it that way.
I don't know if you remember but we had a murder here in Concord. It's important to tell you how emotions are. When the fellows were first making our computer in a room about half the size of this, they made every one. They had someone else put the wires in. But I finally said we should put these up in production and let the girls make them. They were devastated. They had to make every one. The idea of having somebody else make them they couldn't understand. One fellow quit, the other never recovered. That second fellow eventually shot his boss with a shotgun slug sitting at the dinner table with his six kids at night. Over this vague issue. It was his fault, mine really, that they lost the opportunity to make every one of those computers themselves. I tell that story to show you the emotions involved in someone's work. When it comes to ideas like anyone's computer, I can't imagine anybody being full of hatred over the issue of anybody bringing up a separate idea.
This is something you read about, the year 2000 problem. Remember when I was fired, they said Ken Olsen lost it, he doesn't know mainframe computers are dead, PCs are the thing. Four years ago everybody knew that mainframes were dead. Now the mainframe computer has a bunch of applications on it and it started a long time ago when they used two digits for the year. 19 was two digits. How anybody could be so dumb now it's hard to figure out, but everybody did it. A catastrophe is going to happen in a few years. The people who wrote the software are gone, what they wrote it on is gone, and they fired all the people in downsizing who knew what it was. They obviously weren't needed. This is a real serious problem. I would say to people if you can't even take care of what year it is, how are you going to move that over onto a PC. One of the other things about computers is they say they never fail. The company is dead if they did.
If you want to take $100 out of an automatic teller machine, you wait a while and it comes back and says how much do you want. And it prints out your receipt. The only arithmetic it did was to subtract $100 from your account. All the rest of the time it kept your files, decided if you were worth it, then put the files back. So the fact that personal computers are $100 faster than a mainframe is irrelevant. Remember how ridiculed I was when I spent a million dollars on a mainframe computer.
In summary the interesting thing is for anybody if you're buying a computer for your home or to do anything at work is to be scientific enough to say what is the problem. Now there is a certain emotion as a buyer sitting in an airplane with his 386 laptop and a girl next to him with a pentium with twice the speed. "I will never sit next to a girl with a faster computer." Neither one could type that fast. That's the kind of problem we face. Sometimes your manhood has to be satisfied a little differently.