An Interview with Watts Humphrey, Part 6: The IBM 360

Like this article? We recommend

Reflections on Management: How to Manage Your Software Projects, Your Teams, Your Boss, and Yourself

Learn More Buy

Like this article? We recommend

Like this article? We recommend 

Reflections on Management: How to Manage Your Software Projects, Your Teams, Your Boss, and Yourself

Learn More Buy

The IBM 360 Announcement

Booch: You were going to tell me a story about the 360 and its announcement.

Humphrey: Oh, yeah, the 360 announcement. This was in end of March in ’64 and at this time, after we got the FAA bid in, I had been promoted in January. I’d been promoted to Director of Systems and Applications Engineering on the corporate staff and I was now reporting to Dick Bullen, who worked for Dick Watson. I reported directly to Dick Watson originally, before Dick actually left IBM and went to be Ambassador in Paris. But I got called to this meeting and it turned out it was the meeting where Bob Evans and Fred Brooks were presenting the 360 to get approval to announce it.

Booch: Was this your first exposure to Fred or had you run across him before?

Humphrey: Oh, no, I’d run across Fred before. He’d gone around all the divisions and talked to all the people about 360 and his vision of it. Fred had been marvelous. He kept everybody up to speed and tried to get everybody to buy it and he was a marvelous salesman. He got everybody convinced this was the way to go and he had a marvelous vision. And of course, trying to pull together all these disparate divisions at that time was extraordinarily important because they were all on different wavelengths and viewpoints, but I’d say a large part of getting everybody to work together was Fred. And then getting the architecture behind it and getting people to realize you needed all of these standards for interfaces and needed a standard machine architecture and you had to have compatible programs. And he had this vision, but he was sharp enough to know that a vision without backers isn’t going to go anywhere. And he had put it together and got all the support for it, and Bob Evans who was the bulldozer, so that fortunately Fred Brooks was the guy who did it. I mean Bob [Evans] was bulldozing the senior management. He didn't have to fight the technical stuff. And so that was a marvelous team. They did a great job. As you may know they got a National Medal of Technology for that.

Booch: Yes, Fred told me the story when I think he went for that. I guess it was in 2000 something was it not? Was it a bit later?

Humphrey: I don't remember what year they got theirs but I think it was a little before that. It was one of the early ones.

Booch: Yeah, yeah.  Fred told me the story that he was there at the same time Steve Jobs was, and Fred said he had a problem with his Apple, and Steve handed him his card and said, "Send it out to me I'll get it fixed."

Humphrey: Oh, is that right?

Booch: Small world.

Humphrey: So I sat through the meeting where they were reviewing the proposed announcement, and Fred got started with the architecture of all the technical discussion, and Tom Watson got up and walked out of the room.

Booch: Wow.

Humphrey: They were all sitting there, and Learson was sitting there and he said, "Go ahead keep going." They went through the whole discussion. He described it all technically what they were doing and everything else, and about the time he wrapped up, Tom Watson came back in, and sat down for the financial and marketing discussions. How he knew when to come back I'm not quite sure -- maybe somebody snuck out and told him.

Booch: So what did you think of him leaving during that? Were you all concerned, or how did you feel?

Humphrey: I guess I wasn’t too surprised. It was a very good tutorial discussion, but I learned by then that executives weren't interested in tutorials. And I mean Tom had a lot on his plate. He was a great multi-programmer. The day I met with him for my very first interview in New York at 590 [Madison Ave, IBM headquarters], the next day was the annual meeting that he was presiding over.

Booch: Oh my, busy guy.

Humphrey: Yeah. So this guy, he took an hour to meet with me, and I thought this guy can compartmentalize stuff pretty damn well. So that's what he was doing in the System 360 meeting. He knew what he was doing. He was satisfied with the technology, and he'd gone through all of that stuff, and he wasn't going to waste time listening to us. So he just went out and did something and came back when he was needed.

Booch: Good for him.

Humphrey: And so they approved the April 4th announcement, and they went forth with the announcement -- changed the world.

Booch: Indeed they did.

IBM Time Sharing

Humphrey: Yeah. So I was on corporate staff as Director of Systems and Application Engineering, which had systems engineering in the field and computer software. I was director of both of them. And the following November, November of '64, I got a call to a meeting. It was from Learson’s office, and he wanted to meet with me on Saturday in White Plains. I told his office that I could but I had a date with my wife in New York for dinner and the symphony and I really didn't want to miss that. And he said, "Well when do you need to leave?" This was his assistant that I was talking to. And I said, "I need to be able to leave at five." He said, "Oh, okay." So the meeting was at two or something like that Saturday afternoon. So I arrived in Learson's office at two, and there was a crowd there, and they were starting to present a story. And I didn't know what the meeting was about or anything. I was in the dark completely. And so Learson kept pushing. He said, "Now hurry up. Watts has got to leave at five." And he had all these people there, and no one knew what I was doing sitting there. And so they went through all of this stuff and Learson finally got them to finish by four. And then he said, "Okay." So they all trooped out and I was the only one left.

And so it was Learson and I and another guy who had become Group Executive, John Gibson. He was a Senior VP of IBM and he -- and he and Learson -- Learson had, at this point, marketing and the FSD people. He did not have development and manufacturing. Gibson had development and manufacturing and research, the research division and other stuff. And so it turned out that what they wanted me to do was to report to them and to take over a proposal again. But this time what had happened was that MIT had come up with a timesharing approach that required virtual memory. And they had been negotiating with IBM to get IBM to put virtual memory capabilities into the 360 line, and IBM had flat out refused, arguing that the virtual memory wasn't needed, memory can be much bigger. Amdahl was the key behind all this. He said, "We don't need virtual memory. We'll just have bigger physical memory." And memories were getting bigger, and so we didn't have that problem. And MIT wasn't buying it, and so they had come up with their own design. And they got GE -- remember we talked about GE -- they got GE to build it.

And so MIT had bought the GE Multics machine and system -- the GE system which had the multiprocessing with virtual memory -- and they were working jointly with GE. MIT and GE did the design for the Multics System, and Corbato at MIT was sort of the key guy to run the design and development of that. And they were bidding that system now, GE was, to Lincoln Laboratory and a bunch of other people. And the nervousness from the marketing people was, and that's why they were so upset about this, and the top marketing people were in the meeting with Learson, by the way, before they walked out. But we got this whole story, and Bell Laboratories was leaning toward getting the GE Multics System, as was General Motors. Essentially IBM's top 10 accounts were all tending towards going with Multics. And the marketing guys were scared to death.

This was becoming a hell of an erosion. It would get GE really in the catbird seat and IBM 360 would be in real trouble, because these were the top people and Multics would be the wave of the future. So there was a proposal due next to Lincoln Laboratories. IBM had already lost the proposal at MIT to GE. And so the Lincoln Labs proposal was now due and we had a couple of months or something for the Lincoln Labs deal, it wasn't very long. I guess it had to be in like the beginning of the year or something. And they had been working on it and nobody had figured out what they were going to do.

Booch: Sure. And correct me if I'm wrong, but around this time there was the phrase about IBM and the seven dwarfs because GE was one of them, and there were lots of competitors nibbling at IBM's heels around that time.

Humphrey: Oh yeah, NCR was there and we had the Unisys, of course, and CDC and Philco and, oh there were a whole mess of people. So it wasn't a small band at all, and there was a lot of ferment going on in the field. So in any event, I was brought into this thing and again same kind of story, "Whatever you need Watts," and you know, "go do it." The problem was the 360 was taking up all the horsepower we had. And the problem we had was we had to build the programming system. And so I got agreement that the people in the Advanced Systems Development Division up in Yorktown Heights who had actually build a time sharing system, an early one which had been worked on with MIT and was the precursor to Multics. And so they had this early version. And so I asked for that team. So I got that team and it was a very good folk, marvelous folk. And I got a guy as the marketing director -- Orville Wright.

Booch: What a name.

Humphrey: Yes.

Booch: From Buck Rogers to Orville Wright.

Humphrey: Orville Wright, it turned out, later left and was president of one of the big telephone companies, and so he got a bunch of big jobs, good jobs.

Booch: Any relation to the famous Orville Wright from way back?

Humphrey: Yes, a grandson.

Booch: A grandson. Wow.  Small, small world.

Humphrey: I think Orville has subsequently died. But he was very nice guy and a marvelous marketing director. And so we had this thing, we had this bonfire going, we had to win the Lincoln Labs proposal and I got to know the people up there. Very good crowd and very logical people. They were very sensible, marvelous customers. So we were putting together a bid for them, and I spent quite a bit of time with them to understand what they really wanted and that sort of thing. It was pretty clear we had to have virtual memory, period. So we got together with Jerry Blaauw and a whole bunch of people and I put them in a building. I think it was called Fluellen House on the Yorktown Heights property where the Yorktown research lab is, down in the lower corner there, bottom of the hill at the entrance level. It was just an old white house. So I commandeered the main area -- well, I commandeered the house. We got this crew down there. They all commuted in; they weren't living there. I met with them first and I told them -- this was on a Thursday -- ”you’re working on weekends.” I mean, this wasn't one of those nine to five deals at all. And so I went through the criteria that this bid had to meet and the architecture had to do the following. And so I said, "You guys figure it out and I'll be back occasionally but here's what we’ve got to do." And I had the top architects and designers. The marketing people were there, as well as the top architects and the engineers. And so we had a long weekend to design the system so we could make a bid. And we had the guys from ASDD group -- the programming group -- pulled in, and those were the only programmers we had on it.

Marketing the IBM Model 67

And so I come back, they put together a marvelous proposal -- the Blaauw Box. Jerry Blaauw came up with the design and it was a modification. We couldn't start building new machines, obviously, so we had to take the 360 and bang it up. So we decided to use the model 65 as the base for the system 360 model 67. And we had the 65 engineers there with it, and they went through and figured out how to do it and how to put this stuff together. So they came up with a hardware design, and the other guys came up a software strategy, and so we basically had it and that was after about four days. And when I came back to go through it I told them, "I want you to show me how we're going to be best in each of these categories compared to Multics.” And so they did, I mean they put the design together and it was marvelous-- it was really a great job.

Booch: Now as I remember Gene [Amdahl] was still on the scene, wasn't he? He was still at IBM around this time because it wasn't until '70 that he took off.

Humphrey: I don’t remember when he left. But he was not fighting this. I mean, you knew this wasn't the first time where he was going to fight this. He still strongly disagreed with virtual memory.

Booch: And I know that was one of the reasons he moved on.

Humphrey: Yeah exactly, exactly. But in any event we put this together, we put in the proposal. It was a very simple design for the virtual memory, but it was a good one. And we put it in and we won the bid. We got the Lincoln Labs bid, and the marketing guys were going off, and Orville Wright and his team and they were putting out fires with this system. The Model 67 turned out to take the market by storm. I mean, people loved it. And it had multiprocessing -- the 67 was the first multiprocessor of the 360 line.

So we had to have that in place so we could have multiple computers come together with a virtual memory, which is a very attractive system, and they had all kinds of expansion capabilities and a big deal with some of the real-time communication that you needed and everything else. So it was a great system. When we put that in, we won and we were going great guns. The programming guys did extremely well. They were up in this lab in Yorktown Heights and then all hell broke loose. The Multics people and the GE folks had decided to leapfrog the technology, and so they had come up with an expanded virtual memory approach and they sold it to General Motors.

Well, the GE people had come up with a bid, GE and MIT together to put a much more expanded virtual memory into the Multics System, and it actually had a great deal of flexibility. The reason it was attractive to General Motors was, General Motors wanted to use this timesharing system for the graphic design for their automobiles. And they had a big graphic design system, and the marketing people got me out to GM to see what they were doing and why it was interesting. And they were working with the University of Michigan and Bernie Galler and folks out there. And very nice folk. I got to know Bernie quite well and he was pushing this stuff real hard. And he was sort of the intellectual push behind all of this stuff for this expanded virtual memory.

Booch: I have to ask. Were they doing their CAD work on it?

Humphrey: Well, they had big IBM displays. I think it was called the 2250 or something, but it was a big display running off the 360, and they had them on earlier systems but they were a big part of the 360 proposal. And so it they were damn good systems. They were doing amazing things with them way back then. So it was quite something that GM had a lot going on, GM Research -- that's who we were working with. I didn't tell you also the Bell Labs people we worked with Ed David and a bunch of those folks. So I got to know all of those guys. It was quite an interesting bunch of folks. In any event GM was really pushing us hard on this. They had to have this added memory and they argued they couldn't -- literally couldn't -- do without it. It was pretty obvious we were not going to win the GM bid unless we could build something substantially more then what we were doing. And the marketing people were all upset because they concluded that, if GM went with GE and Mulltics, we'd lose Bell Labs; and if we lost GM and Bell Labs, we were going to lose the momentum pretty completely. And so the GM win was a big deal. And so I got together with the programming guys and the architects to figure out how we would do it, and that it would take a hardware change, which we were told was straightforward. But the programming guys went through it and after like a weekend's -- a long weekend's -- worth of study….we were strong on long weekends in those days.

Booch: It sure sounds that way.

Humphrey: But in any event they concluded that it would add about three months to the schedule. Well like a dummy I bought it. And so we put in a proposal. We won the GM bid, and Lincoln Labs was very upset with the three month delay, but we talked them into it and all the other customers -- we had a three month delay for everybody. Everybody finally bought it we got it sold in the market. They’d all do it. And so we got started on that. We had that bid. And so this was in the fall of '65, I think it was.