Home > Articles > Programming

An Interview with Watts Humphrey, Part 5: Early Projects at IBM

  • Print
  • + Share This
In this transcript of an oral history, Grady Booch interviews SEI Fellow Watts Humphrey. In part 5, Humphrey discusses his early years at IBM in the 1960s, where he built the first automated system for the New York Stock Exchange ticker system and the FAA's air traffic control system, which are still in use today.

This interview was provided courtesy of the Computer History Museum.

See the entire interview.

From the author of

The Brokerage Studies at IBM

Booch: So tell me about the things you did and what were you assigned to do [at IBM], because I know you got the title later on of Director of Programming, but was that your initial title or is that something that you came into?

Humphrey: I came into that later. There’s a couple of interesting stories that led to that. The first job I had, they put me in charge of the brokerage industry studies in the Advanced System Development Division andI basically had to build a group. We didn’t have anybody. They gave me a group that was doing some work on the stock exchange. And so I remember first day I got the job, I got a hold of the branch office manager at Number 1 Broadway who ran the brokerage office, went down to see him. His name was Buck Rogers, a wonderful guy. He later became a VP of IBM. A marvelous fellow.

Booch: How can you go wrong with a name like Buck Rogers?

Humphrey: I think that’s true, but in any event I went and talked to Buck. He took me over to see some people at Merrill Lynch. He said, “That’s where we want to work.” I went and chatted with them, and I went back up and talked to the management team, and we got a meeting together because I wanted to start to work with Merrill Lynch. So we got together a meeting with the division president and all the other top people, and I presented what I wanted to do, and so everybody said, “That sounds great.” So I left the meeting, sounded like I had an okay. I called the branch office. I said, “Let’s go down and talk to them.”

So I went over and talked to Merrill Lynch, and that afternoon we had an agreement that we were going to put together and do a joint study. So they basically agreed on it, and we agreed generally how we were going to do it. We just had to put together the documents. So I got back and told the management team. They were horrified. They said, “You can’t do that.” I said, “Why not? If everybody agrees we can do it.” They said, “But you got to go through and get an approval from the lawyers and everybody.” I said, “Let’s do that, okay? That’s fine.” But, enough. So in any event we just went ahead and did it, and that’s one of the things that I learned way back at Sylvania before that, and that is that the old Jesuit theory: It’s better to ask forgiveness than approval.

Booch: I’m curious. Did you grow up Catholic? We hadn’t talked about that side of it. You mentioned--

Humphrey: Oh, no. I grew up agnostic. My dad was an agnostic. I did, in fact, become a Catholic a couple of years ago after all these years, after seven kids brought up in the Catholic church and my wife a lifelong Catholic.

Booch: Very good.

Humphrey: My step mother was also Catholic, but we didn’t go to church at all and it’s just something I ultimately decided to do. There’s probably not time to go into that, but I am now a Catholic. And I am a lector and a Eucharistic minister and all that sort of thing, so it’s great.

Booch: Marvelous, marvelous.

Humphrey: But in any event, I started off in the brokerage industry and we actually put together the first automated system for the New York Stock Exchange ticker system and their floor trading system. We put together a proposal for Merrill Lynch on how they could make an electronic trading network. We also worked with Bache and Thomson McKinnon and we had a bunch of others that we were talking to. We were designing a system to work with them and to help run electronic trading, basically automate the way they did it, pretty much what they’ve ended up doing.

I remember at one point during this point they had a meeting with the board of governors of the New York Stock Exchange in Poughkeepsie, and they asked me to come give them a talk. And at this time the biggest market trading day they had, as I recall, was in 1928; like, 20 million shares or something like that, 19 million shares that day. A typical trading day in those days was three million shares of stock, and so when I gave my talk to the New York Stock Exchange board of governors, I talked about trends and that sort of thing and I basically started by saying, “How would you handle a 100 million share day?” That kind of blew their minds. They’d never thought about a 100 million share day, but I presented enough evidence to show them that it was going to happen.

In fact, it happened before a whole lot longer. We’re running about 2 to 3 billion share days now. So it’s way up. They later told me the stock exchange people changed their whole perception and all of a sudden they began to think you had to go electronic.

Booch: Because back then the trade reconciliation was still largely done by humans, was it not?

Humphrey: It was all by hand.

Booch: This reminds me then of the story of what became the Depository Trust Corporation. Were they in the picture here around this time, too?

Humphrey: I didn’t have any involvement with them at all. All of the brokerage firms, they had their own secure lock up systems for all the certificates and stuff, because they still had certificates, and everything was done with paper and all that sort of thing. And so basically all the brokerage houses had that -- essentially a lock up secure area where they kept all that stuff, so that’s what we did there. And when I was running the brokerage industry stuff, IBM was at this time designing the 360 system and that sort of thing.  Let me step back. I guess we just moved to a house on-- #10 Barron Circle.

The FAA Bid

Humphrey: It’s the six bedroom house that we bought and it was about 1963. I got a call one day. This is the day before Thanksgiving in November of ’63 that there was a meeting with [IBM’s] Vin Learson Friday morning in his office at Yorktown. IBM was moving its headquarters over to Armonk, and they had temporary headquarters in Yorktown, and he wanted me there at 9:00 A.M. on Friday. Well, this is a day off, of course, the day after Thanksgiving. I didn’t argue about that and I said, “Okay.” His secretary wouldn’t tell me anything about the meeting, so I just arrived in Yorktown, got to his office, and there were two other gentlemen there.

There was the president of the marketing division, Frank Cary, and the president of the Systems Development Division, George Kennard, and me and Learson. And so they explained that they had gotten the bid request from the FAA for an air traffic control system, and they wanted me to take charge of the proposal effort. The development team had put together their approach for how to do the job, and the development division president, a VP under him, had put this together. A fellow named Bob Evans. And the marketing division guy, the Federal Region, fellow named Ralph Pfeiffer, his team had put together a proposal from the marketing people as to what they ought to do, and they were at loggerheads. They could not agree. At this point my whole brokerage group and half of the ASDD crowd had all been moved over to the marketing division.

So I was in a very strange position. I was a development engineer with all this computer background and stuff in the marketing division. And so they had scouted around to find who they could put in charge of fixing this problem and getting these things resolved, and they’d landed on me. So I was pulled from way down in obscurity somewhere, and I’m sure my contact with Tom Watson really didn’t hurt at all. So contacts don’t hurt in this field, and you take advantage of them when you can, and I was very lucky. So in any event they went through this and they asked me if I would run the job, and I agreed. I had to figure out what we were going to go. And it was the day after Thanksgiving, and the technical proposal was due before the end of the year -- I think before Christmas -- and the financial proposal was due before New Year’s Day, so we didn’t have a whole lot of time. And they had these two teams, and they said, “What do you want?” And I said, “I want to meet with the marketing team tomorrow in Washington.”

So Frank Cary said, “Okay, we’ll set that up.” And then I said, “I want to meet with the development team Sunday in Poughkeepsie.” George Kennard said, “Okay, we’ll set that up.” So they did, and I went down to meet with the marketing guys and find out what they really had to have, but I wanted to go there first. And so I really got to know what the heck they were talking about and why and they’d come up with a pretty impressive story. It was a multiprocessing system just like the ones we designed at Sylvania. They were having a big battle with the development folks because they said you couldn’t build such a system and you couldn’t program it if you did, and so I knew they were part way wrong anyway.

So I went up to Poughkeepsie Sunday and discovered that they had a real distributed crowd of people all over the place. I was introduced as the boss in both places immediately, and one the things that Learson, by the way, told the two division presidents, he said, “For purposes of this proposal you report to Watts.” So I was reporting to Learson, and I had the two division presidents working for me. So I was given carte blanche, and I thought it was wild and I might as well use it. So I talked to the engineering manager who was putting the proposal together and engineering it, and they were trying to build it out of what was going to be the Mod 50 360 machine the following April and the Mod 50 turned out to be far ahead of the other 360 machines.  So in any event they used the Mod 50 as the base for building the FAA machine, but they had not put together the multiprocessing the FAA wanted. The fail soft multiprocessing had to be a polymorphic system. There must have been half a dozen people bidding for this thing and I’d gotten the whole story of competition. It was the number one proposal, the biggest bid IBM had ever submitted at this point.

Booch: What dollar amount were we talking about back then?

Humphrey: $100 million.

Booch: Wow, that’s a lot of money even today.

Humphrey: It was then. But I told the engineering guy, “We’ve got to get this team together. We can’t do this unless we get the marketing guys here, too.” So they all agreed, and I said, “Where we going to do it?” And they didn’t have any room at all, and the engineering guy said, “Well, I belong to a volunteer fire department out in Red Oaks Mill and they have a big dance hall there. Maybe we could rent that?” I said, “Let’s do it.” So he called and rented the dance hall. It had a piano in the corner. It was just a big room. And I got a hold of the marketing guys and I said, “We’re all showing up there, like, ASAP.” So in about two days we got in there. We had furniture. We had tables set up. There was a blackboard we could slide into the middle. We had about 50 people. Even had the financial people come in and work with us.

Booch: With the piano, did you start off by singing the IBM company song, which I think existed back then?

Humphrey: No. Every so often somebody would go banging something on the piano. It was marvelous fun. I’d pull the blackboard in the middle of the room and bang on it. I said, “Meeting.” I told them what the meeting was. Anybody who wanted would come and so our first issue was we had to agree on a strategy and exactly how we were going to design the system. So I went through that. It took a little while. We hammered it out and decided because I’d gone through this stuff, could you build the hardware or not. I said, “First of all, this contract is not for programming. We are building the machine, and here’s the kind of machine they want and if we don’t build that machine, we’re not going to win.”

And so the guys finally bought that, even though the VP of engineering over the engineering guys, Bob Evans, absolutely disagreed with the design approach. He said, “You can never get it to work.” You couldn’t program it, and that sort of thing. So in any event I overrode that and so we went ahead and put together the proposal. As part of the proposal the Lincoln Lab had put together a set of four programs that you were supposed to take. Whatever your final proposal was, you had to write those four programs and come up with eight numbers. The numbers were the size of the program in bytes that it would take to store and the time it would take to execute for each of the four programs. That was eight numbers and the marketing guys had put together, with the 360 instruction set, the answers and they’d gotten their best system’s engineers to do it and they’d put the whole thing to together and they had eight numbers.

So I took the programs and the numbers and the specs over to see Gene Amdahl who was then the head of architecture for [System] 360 and I explained the problem to Gene. I said, “Gene, these are $100 million numbers,” because it was, like, a Thursday or a Friday. I said, “Could you just have your folks take a look at this and see if these guys have done a good enough job?” And he said, “Okay. When do you need it?” I said, “I need it, like, yesterday.” He said, “Okay.” So that weekend he and his two top architects re-did the programs and took 40% off the time and size. Boy, that was a $100 million weekend.

Booch: So if I may ask, do you remember the nature of those four programs you were being benchmarked against? What were they exactly?

Humphrey: I do not know, but they were something that Lincoln Labs had worked out. This will tell you how good a system this can be to do air traffic control. I’m sure it was basically air traffic calculations -- the intercepts and all the kind of stuff you’re after to run an air traffic control network of the type we were building. So in any event, we did get the proposal in. It was a hassle. It was printed on, I think, Christmas Eve and we had to go down to the printers and boy, we had everybody volunteering to go down to the printers.

They wouldn’t let me go. They said, “We’ll go do it, Watts.” So they did and we got the proposal out, got it in on time, but we didn’t have the pricing, and then we put together the pricing. Well, it turns out they’d done some fancy work with some of the memory stuff because we had standard memories and then they had a bulk memory that IBM was also offering. So our original proposal was with bulk memory, and so we priced it with the bulk memory but they had standard memories also, and the system could be multiprocessed. You got the four processors. You got the eight memories, a bunch of channels, and everything could communicate with anything and anything could fail. And, of course, we had all of these interlocks between them that were programmable so you could actually tell status and pass control and that sort of thing, so we had all that stuff in that we had designed before and I had some marvelous people working on this. They were really very, very talented.

FAA Bid Pricing

The architects were working with us as well, so we had a really very, very, very good technical crew. So we got the proposal together and submitted it and when we put the price proposal together and submitted that, the FAA took a look at it. I remember having a meeting with Tom Watson and Vin Learson. Vin Learson, when I’d been given the job said, “If we have to bid it at $1.00, we’re going to win this contract.”

Booch: And if I remember your pricing came in a little bit higher than $1.00.

Humphrey: Well, I was, kind of, surprised when I arrived at the pricing meeting in Armonk with Vin Learson and Tom Watson and a bunch of folk… and it was a pricing meeting. They laid out what the costs were and all that sort of thing, and so we went through all of that. The top financial guy, Hillary Faw, a marvelous guy, he was there. He described all that stuff and we went through and put together the price. We proposed what it was going to be and the marketing guys were driving for low number. They wanted to come in under $70 million, because they said, “That’s where the best competitor’s going to be. We have got to do that.” And that wasn’t profitable. And so it got around to Vin Learson who was a senior VP over marketing and development at this point. Vin was a big guy. Remember, he was my boss. And so Tom turned to Vin and said, “Vin, where do you think we ought to price it?” He said, “$100 million.” I thought, “Oops.” Way more than the marketing guys were thinking. He said, “That’s what it’s worth.” He said, “This is a marvelous system. It’ll do the job for them, that’s what it’s worth.” So actually they priced a little bit over $100 million, and that’s what they submitted.

And so the FAA went through that, and they came back and because of the pricing, the bulk memories turned out to be most of the memory in the system, but the way that we priced everything (which had been, kind of, you know, done quietly by the financial community and they didn’t tell people what they were going to do) it turned out that the memories in the computers themselves were priced much lower and that if the FAA ended up replacing the bulk memories with more basic memories it’d actually save money, several million dollars. And so they did. So the bid came in right under $100 million and the FAA accepted our proposal. Even though we were the highest-priced bid they took it. Well, the team we had together, this 50 person team, they were amazing. While they were all at loggerheads before this, they were all together and we kept them all there.

They were all involved. We knew what we were doing and the whole idea of building a coherent team, everyone knew what we were doing. Tom Watson never arrived there, but Vin showed up for several meetings.. He’d come in and I’d bang on the screen and introduce him, and so Vin would talk to the group, and so we’d have a daily meeting with everybody, and something would happen and there would be this excitement. They’d bang something on the piano and then they’d go announce it.

Booch: Oh, you were still in the dancehall during this time. You rented it for a long time.

Humphrey: We were at the dancehall through New Years, and they said that dancehall really paid for itself.

Booch: I’d say so. That’s certainly not a piece of IBM history I knew about.

Humphrey: No. But I mean, we had an integrated group. It was just so enthusiastic, and the excitement when we got that thing in was amazing. But I remember a meeting right after the win. There was another meeting called with Tom Watson with me and Learson and Ralph Pfeiffer and all the top marketing people. Frank Cary was there and the financial people, and they wanted to know how the FAA had been able to snooker us by getting the price down. And so Tom was really mad and he, kind of, looked around the room and he said, “Now, who’s responsible for that?” Dead silence. But finally I said, “Oh, Tom, I probably am. I was proposal manager.” He said, “Well, what happened?” So I explained it to him about the pricing and how it had happened, and I said it was, in fact, a surprise that that’s what happened. And he said, “Okay,” and so that was the end of it, but I was surprised. In that whole room not a soul would step up and say, “Well, no, I was responsible.” But I figured I’d better. And in a way I was, because I -- Hillary Faw later told me, he said, “You could have known all of that.” I said, “Quite frankly, you were pretty busy.” So I didn’t really understand it, but that’s what happened. So we won that, and shortly later they had a re-organization and Bob Evans, the VP, was pulled from his job. He was the guy that I’d had these battles with, but he’d been involved in this whole thing. He knew what was going on and he was put in charge of the Federal Systems Division, which was also reorganized at that point. So Bob became division president of FSD.

Booch: And this would have been what year? This would have been ’65 I think?

Humphrey: I think it was’64. Well, the [System] 360 was announced in April. I think shortly after they had the reorganization because they’d rearranged the whole division structure. They put one guy in charge of all development, and then they had marketing and manufacturing pulled out of the separate units. It was all rearranged and Learson still had development and manufacturing. Marketing was under Dick Watson, Jr., Tom Watson’s younger brother. Frank Cary was development group manager. He had stuff under him including a systems development division and systems development had the whole 360 and everything. So they’d gotten 360 announced, and there’s a story there, too, but let me finish this first and come to that. The Federal Systems Division was later asked to bid on the programming for the FAA system. Remember, “Bo” [Bob] Evans was the guy as division president now at FSD who said you couldn’t program this thing. So his division put in the proposal to program the FAA system and they won it. And so Bob’s team actually programmed the FAA air traffic control network, the en route air traffic control network which is used to this day.

Booch: Wow, well, as a frequent flyer I feel happy that you were involved with that.

Humphrey: Boy, it is the system. It was delivered and it’s been enhanced and upgraded and all kinds of stuff but it’s been working. That design has been working ever since.

Booch: Now, this was all still in assembly language, correct? Because I know FORTRAN was on the scene back then but that really wasn’t being used, was it?

Humphrey: No. No, that isn’t quite right, because part of our bid -- the one programming thing we had to deliver -- they had some utilities and stuff where we had to deliver a JOVIAL compiler.

Booch: JOVIAL, of course, wow.

Humphrey: A JOVIAL was behind that thing, and so that was in it. I’m sure they moved to C and C++ and everything since but in any event that was in it.

Booch: Did IBM actually build that compiler?

Humphrey: No, it was contracted out. It was part of our bid. We had a contract to do it. I don’t have any recollection but we got a JOVIAL. We were able to buy one essentially, modified to run on the 360.

Booch: Got it.

Humphrey: So that was amazing. So Bob Evans at FSD actually did the programming, and they did a marvelous job. It worked fine as we knew it would.

  • + Share This
  • 🔖 Save To Your Account