Home > Articles > Programming

An Interview with Watts Humphrey, Part 22: The Process Conferences and the PSP Course

  • Print
  • + Share This
In this transcript of an oral history, Grady Booch interviews SEI Fellow Watts Humphrey. In Part 22, Humphrey explains why he stopped going to process conferences and why he decided to write a book and teach a course on PSP.

This interview was provided courtesy of the Computer History Museum.

See the entire interview.

From the author of

The Process Conferences

Booch: In terms of influences, I remember there was a lot of discussion about what’s called process programming, and Lee Osterweil and crew were talking about that, and Barry Boehm was in the midst of it as well. Do you have any recollection of that being an influence, or you influencing what was going on there?

Humphrey: I participated in those workshops -- the software process workshops. They invited me to come, and I presented the maturity model at one of them, and everybody got all excited about it, and so I’d do a big talk on that. I used to go to ICSE (International Conference on Software Engineering) meetings, and I discovered surprisingly quickly that I was really off in left field. No one was where I was, and I kept describing what I was doing and why, and everybody was very interested in what Watts had to say, but nobody ever picked up on it. And I remember we had one of the software process workshops in Yokohama in Japan. Isn’t that that the north island?

Booch: Yes.

Humphrey: I think it is, yeah. Went out there, and a bunch of the guys were there. Of course, I’m the old guy. I was the old guy in all of this stuff, and the Japanese were probably closer than anybody to where I was, and they were pretty much reflecting what I was saying. They’re much more people-oriented. I was quite surprised.

By the way, I didn’t say I had put together a steering committee when I was running software process and quality at IBM, and I had Lee Osterweil on it. Lee Osterweil and Bob Balzer were both on the committee. I had a bunch of academics that were coming and reviewing what we were doing and giving us advice. I had a whole tool group and everything else in Poughkeepsie, and they were looking at the Ada design language stuff. So when I got working with them as well and we had a great time in this meeting near Denver and I’m trying to remember the name of it. Brecken--

Booch: Breckenridge, yeah.

Humphrey: Yeah, we were meeting in Breckenridge, and that’s where Lee Osterweil came out with his processes are programs, and I came up with this a _____eahhe name of it.  Brechbout next step.ing in is designing and programming has really basically repalced y needed essen CMM maturity model. Harlan Mills was involved then. I didn’t talk about my interactions with Harlan at IBM, but I did have some interactions with him there, and so Harlan was all excited about it. I remember at the ICSE meeting where Lee Osterweil and I both presented our talks, Harlan got up and made some comment about, “This is the best session we’ve ever had.” But he was a wonderful guy. I loved him. I used to meet with him. I’d go over to the east coast of Florida occasionally and meet with him. I disagreed with his Cleanroom stuff, though. Not because of what he did, but I disagreed with it because it didn’t measure. He didn’t gather data and measure stuff, and he and I were not together on the quality principles of where we were going.

But in any event, yes, I had met with these groups, and we were not on the same wavelength at all, and the meetings, they’d go through all this discussion. They’d want to automate this stuff, and they were trying to get tools to do this and that and a language for this and it was tech-y. It wasn’t focusing on what people do and that bothered me, because the way I thought about it was if I’m running a programming organization how was this going to help? And I didn’t see that and I still don’t. I mean, I think they’re still going in that same loop. I don’t know at all. I think it stayed mostly academic but I really don’t know. I haven’t gone back and looked. That may be unfair.

But in any event, we’d submit papers every year and they were reviewing them. And so I’d submit a paper every year and I’d get invited and I’d go, and one year they sent out the reviews. I’d never gotten the reviews before, but one year they sent out the reviews. This is for the one at Washington, and the reviews all said, “Why do we want to hear this stuff?” They had no interest in it but I was still invited to go. Obviously, I was going only because I was Watts Humphrey, not because what I would say was of interest to anybody, and I decided this is not something I’m going to continue doing, so I stopped.

The PSP Course

However, one meeting I did go to was the one in Germany. There was one in Germany and there was also a meeting in-- where was it? -- in Berlin a previous week. Some kind of a process conference, and Peter Feiler and I gave a paper on process terminology we put together. And when I was sitting in this process conference, it was boring as the devil and I wasn’t interested. I had developed the PSP and couldn’t get anybody to try it, and I’d talked to some people. I think Mahdevi -- not sure if you know him but he was a professor at McGill. I was chatting to him about it and he said, “You ought to teach a course.” I said, “That’s not a bad idea.” And so he said, “Well, I teach a software engineering course. Well, why don’t you come teach at McGill?” I said, “That’s a thought.” So I began to think about how I’d teach a course and concluded that to do that I needed some kind of a text.

So in the conference while all these people were talking, I was organizing my textbook, and this was early in the year, like, February. And so going to the conference, I had outlined my textbook, and I spent the weekend in Berlin because I had to go to this other conference elsewhere. And I had a wonderful time there going to look at the torn-down Berlin wall, and I went to a concert every night, and had a great weekend. Unfortunately, my wife wasn’t with me, so I was all alone, but it was a great weekend.

But I had this book to design, and I went to sit through the software process workshop the next week. I realized that I had no further interest in it. I was getting nothing out of these conferences in terms of what I was trying to do. So I didn’t quit in any way, but I had decided not to continue going, so I just stopped going to the process workshops or ICSEs. None of it seemed to have any effect. I’ll get back to ICSE in a minute, but I wrote the textbook. I got started in February. I had used PSP to make a plan for my textbook and I got a hold of my publisher who published Managing the Software Process, and he was very interested in it. I put together a schedule. My book would take me till January to get a manuscript out, and then I expected to submit it the following year.

Then I was deciding what to do to teach it, and at the same time one of the guys that I had known at the SEI who was there as a CMU student in our masters of software engineering program was Howard -- Howie Dow -- and he had actually spent some time with me and was very interested in what I was doing on the PSP. When he heard I was writing a book on it and was putting together a course, he got a hold of me and wanted to teach the course at the University of Massachusetts. He worked for DEC, and he wanted to use it at DEC and teach it there, and he wanted to start in September. My plan said the book wouldn’t be done until January, but I had been using a PSP on writing, and my PSP data showed me I would be through with the first draft of the manuscript in September. So I gulped and I told Howie that I’ll get you the manuscript in September for your course.

Booch: It’s known as eating your own dog food.

Humphrey: Exactly, exactly. So I basically said, “This is why -- you might as well prove it,” so I did. And it turned out to be a bit of a push at the end because I ended up having to reorganize the book a couple of times. I took chapter one and then made it chapter thirteen, and I decided I needed a whole new appendix because I was using a lot of statistics. But I didn’t want to write a statistics textbook. I had found statistics textbooks almost incomprehensible, so I put together an appendix A in my A Discipline for Software Engineering book and appendix A was, sort of, a you-drive-it statistics, how do you do this stuff. And the programming problems I put in the textbook 10 programming problems, eight of which were all just simple statistics, simple correlations and linear regression, multiple regression. I had chi-squared tests and approximations, a bunch of things.

So it was a lot of interesting stuff, but I had to explain it to people so the people who knew nothing about statistics could read it and that’s what appendix A in my book was. I learned a lot by writing it. Writing about statistics was fascinating. So in any event it ended being an 800 page book, and I got the manuscript together from February to September. Now, that’s pretty speedy, and so Howie Dow taught the course. I had decided not to teach at McGill but to teach at CMU, and so we put a PSP course together at CMU.

And so I started that in January and used my manuscript and taught the PSP course at CMU. I decided to wait on submitting the manuscript. I was going to rewrite it after I taught the course because the only data I had before I taught the course was personal data. So I got Howie Dow’s data. The first class he did worked well. He had lots of constructive comments. It was very helpful having somebody else do it. There were some guys at Embry-Riddle University who wanted to teach it, so they taught a course. I don’t think in parallel with mine. They taught in the fall.  And others -- some guy at Howard University wanted to teach a course and he got in there and it was a disaster. He didn’t really get it.

I guess about three or four courses were taught and I taught some courses. Some companies got quite interested and they worked extremely well. The course results were astounding. My first class at CMU I think four people in that course all changed their careers in the course to say, “This is what I want to work on.” And three of them are working directly with us still on the TSP and PSP. They just got so excited about it that they said, “It works. This is what process is.” A number of people have been working on process work on the CMM even. Once they took the PSP course, they concluded that, “Hey, now I understand it. This is really what it is.” And that’s what it did. You really learn what a process is and why it’s helpful when you actually use it personally to do your own work.

Booch: So thinking of other people around this time, did you have any connections with people like Walker Royce’s dad, Winn Royce, who wrote some of the classic papers in waterfall life-cycle or for that matter, the work that Filipe – I forget his last name – was doing starting around that time for what became the rational process? Did you have connections with any of those folks?

Humphrey: Yeah, I knew Winn Royce. I’d been involved in a number of things with the SEI and of course, Larry brought him in periodically. A marvelous guy, a wonderful guy. So I got to know him quite well. Of course, I got to know Barry Boehm very well and worked with him and yeah, there’s a whole list of people. I’ve been involved with just about everybody in the community one way or another. I never got to know the Rational folks, but I did get to know an awful lot of people when I got involved with him, so I’ve got an enormous array of people that I run into that I sort of know.

And there are a lot of people that I know that I don’t know, if you know what I mean. They say, “Well, I met you here when…“ I’ve given an enormous number of talks over the years and, of course, I’ve written so much stuff, but I’ve gotten to know an awful lot of people and it’s been a very rewarding period.

  • + Share This
  • 🔖 Save To Your Account