Home > Articles > Programming

An Interview with Watts Humphrey, Part 31: Expanding TSP Use and Lack of Academic Interest

  • Print
  • + Share This
In this transcript of an oral history, Grady Booch interviews SEI Fellow Watts Humphrey. In Part 31, Humphrey discusses how TSP caught on at large corporations but why it has never made inroads in academia.

This interview was provided courtesy of the Computer History Museum.

See the entire interview.

From the author of

Expanding TSP Use

Humphrey: The very first TSP was with an Embry-Riddle team in 1996. We gradually developed it. We learned a good deal more about it. We began to put together coaching and coaching classes and stuff. I have written a whole mess of books on it since. I've written now three books on the PSP. I wrote a book on TSP coaching, coaching disciplined teams, and I wrote one on leading a TSP team.

I wrote a book on winning with software, which is sort of for executives. It's a fairly short book -- couple hundred pages. It doesn't come with crayons but it's sort of close. It sort of tells stories about how this stuff works. What's interesting is that book -- I actually published it in, I think, 2002 -- one of the Microsoft executives, called a business unit I.T. manager (BUIT), when he went on to Amazon, it said, "If you're interested in that book, why don't you try this?" So he bought my book and he read it on a plane somewhere. It's that short. You can read it quickly. And he called us up and said, "I want to do this."

So we got started with Microsoft. We’ve got over 1,000 people using it there. It's in and out. We're not at a senior enough level with Microsoft to be sure it'll stick. Managers keep changing and so it turns on and off and, you know, things like that. But we got it going with Intuit, and the way we start a lot of these, by the way, it's either a book or a paper or a talk I've given, or someone calls and wants me to come give a keynote address. At Intuit, they called me and wanted me to give a keynote address out at their technology conference. I said, "I'm happy to do it on condition that I can have an executive meeting after the address." So I went out and gave the talk and the senior V.P, Bill Eerie, a marvelous guy, introduced me. I gave this talk to about 500 of Intuit's top engineers in the audience.

At the end of it, Bill got up. As I say, he's the senior V.P. of technology. He got up and said, "Who'd like to do this?" About half the hands went up. Well, I'll tell you, that felt great. So we started with them, and their whole QuickBooks division is now using it. It’s had marvelous results. They ended up, I believe, during one period, after they were shipping this code, they had a reduction in calls to their help-line in India, of course, of 800,000 fewer calls in an eight-month period. These calls cost $25 each. They pay out of pocket $25 each so it saved them a lot of money. Microsoft, also -- one of their guys gave, briefly, a comment at one of the conference talks, and he said they've spent $3 million in all the TSP training and installation and what they pay for coaching and instructors and stuff, and their savings to date were $68 million.

Booch: My goodness. And, of course, you get a percentage of that, right?

Humphrey: Oh, yeah, right. The Intuit people did one thing which I was quite interested in. They did opinion surveys, and the opinion surveys actually came in much higher with the TSP teams than the non-TSP teams, and they added a whole bunch of quotes of what engineers say, and one quote was, "After I'd used the TSP, I'm not going back to do it any other way." So we're getting that increasingly. We got Adobe started, we've got, you know, people making video games in a division, there's a division of ActiVision, I think. They used TSP on the last 13 games they produced, including Spider Man II and you've seen the Guitar Hero game?

Booch: Yes.

Humphrey: Now, that was produced with the TSP.

Booch: Marvelous.

Humphrey: And basically I asked the CEO what made them pick TSP, and he said, "Our turnover had hit 17%. People were just burned out. They couldn't handle it, so we decided to try the TSP and now engineers refuse to work any other way."

Lack of Academic Interest

Booch: Watts, what universities would you name that are also contributing to this and teaching courses on this process?

Humphrey: That's been the biggest disappointment. Embry-Riddle did some work. They've been very helpful, but I've got essentially nothing else. They got some CMU courses. It has not been adopted as a standard at CMU at all. They have got a class here and there and that sort of thing. The masters of software engineering course is a little bit better. They've got a little bit more, and they're just beginning to do more with someone who was on our team who is now over there helping with that. So I think that they may be doing better. The problem with the teaching, it's a little bit like in a hospital or medicine. It's very hard for people to get this across and to understand how to get it across or how important it is until they've actually practiced it. I mean, learning surgery from somebody who figured out about surgery by reading an anatomy textbook is not a real way to learn how to do, you know, high-quality medicine.

And that's the case here. It's very hard to get across the power of the disciplines and what it'll do for a team member and a team unless you've really practiced it. That's the issue. So the academic community, by and large, many of them have no industrial experience. They're very bright folks. Their typical approach is to read a textbook and then teach it. And it's very hard to get the real disciplines that come out of it and be able to get it across in that way. So it is not sticking at universities. It's doing a little here and there. It's spotty. There is no academic involvement from a technology point of view. No one is looking at this as an exciting technology, which is unfortunate.

So that's been a real disappointment for me. I've spent a lot of time trying to convince the universities to do something with it. I started with CMU because I figure, if I can't get CMU involved, it's very hard to go to anybody else. The CMU people are great, they're wonderful, very nice folks, but they don't have the time to do it. They're willing to listen to me if I want to teach the stuff. They're happy to have me do that, so they're very responsive and helpful, but in terms of them actually picking up a spear and saying, this is the technology we're going to carry forward, that is not happening and I don't see it happening. So my sense is the academic community will come on board when they have to.

Booch: Got it.

Humphrey: Either because other countries are doing it or because industry insists. I'm afraid that's where we're going to have to go.

Booch: Very well. Thank you for sharing that.

Humphrey: But that's a frustration that's a little bit early on my frustrations.

Booch: Okay.

Humphrey: But we need them. I mean, the problem is, when the White House or anybody else turns to say, "What should we be doing with software in the future?" they put together academic committees. And they, you know, they come back with better test tools and all this kind of stuff and that isn't going to get there. So that is a major frustration. Somehow, we've got to figure out how to get through that. My concern is, am I going to live long enough? I don't think so but I'm trying to get this far enough along so maybe it can survive without me.

Booch: Very well.

Humphrey: So that's the issue there. We've got a lot of stuff with industries, many, many companies, and I could probably bring some names, a mess of them, but there are whole lots of them that we've got. The team is going strong and they're doing lots of, you know, going out helping teams and this sort of thing, and so it's really very impressive what they're doing. I think they've made a lot of progress. One of the things that is happening, which is really kind of exciting, is that we've got a number of countries starting to get interested.

The first one was Mexico, and they had gotten quite interested. They had me come out and give some talks and stuff, and they got the state and federal government support. The federal government, through their Ministry of the Economy, has established a group called Pro Soft to actually get Mexico [to] build a competitive software industry, outsourcing industry, in Mexico. And they decided to do it with the TSP because they wanted to be higher quality, more predictable. They could not go for lower salaries than India or China, but they thought that perhaps this was something that they could do. So we've started with them.

Their strategy was to get faculty trained in this stuff and use the universities, but they decided to do it in what I consider to be a really creative approach. So we went in and trained a whole mess of faculty on the PSP and we trained them to be instructors and coaches, and so they are teaching essentially our industry grade courses to students. They're also now working and have been working with companies to go out and get the companies starting to use the TSP. One of the companies doing that is called Softtek. Softtek turns out to be the largest software company in Latin America; it's headquartered in Monterey, Mexico. I work extensively with the University of Monterey.

[Softteck is] headquartered in Monterey, Mexico. They've got branches in Mexico City, in Canada, in Brazil and they're growing. They've got some offices in the U.S. but not much. They got several thousand people.  As a matter of fact, they did one project with a big department store in Mexico City, and what's interesting about this project is that this big department store, they needed this project to run their business,but what they wanted to do was to be able to take, when a customer comes into the store who wanted to buy a refrigerator or whatever else, they wanted to be able to go ahead and take the order and give a delivery date wherever in their system that refrigerator was. So they could find it. If they had it in a warehouse somewhere or in another store somewhere, they could find it anywhere in their system. They could get their shipping stuff and they could schedule a delivery right then.

Apparently, with all of the stuff that went on, there was a lot of, apparently, a lot of complications in putting this together, but they had actually had three separate contracts to do this. One was with a big company that you and I both know very well <laughter> and others. They had three contracts, each one of which ended up failing. Now, why they failed, I've never really gone into that, but my guess is their costs started to go out of bounds, they began to run into schedule problems, et cetera, but, in each case, they had all kinds of problems. And so Softtek went to this company and said, "We can do it with the TSP." Well, they ended up doing it. The customer really desperately needed the product and, when I was down in Mexico last, I met with the team, and they were delivering the first release on schedule. The company has said, "We're not going to do software work with anybody else from now on." So, I mean, it's amazing. They've done a great job. They also told me that the turnover on their TSP teams is a quarter of that on the other teams. So their people love it, they stay there and it's very effective.

So that's the kind of thing we're running into. I've got to say, we've got Adobe, EDS is working with us, Oracle. We've got IBM Mexico, by the way, is working with us. We've got a Sun Microsystems effort, Fuji Film, Toshiba. I mentioned Intuit, Microsoft and I've got a list. I won't count the numbers, but it must be about 50 companies here, some in Europe, in the U.S, some people in India, LG Electronics in Korea. We've got a NASA group going. So we've got, you know, Siemens, an enormous list. So, in any event, that gives you a sense of what we're doing there.

So basically, in terms of where we are, when you look at the data we've got results with companies, as I say, dozens and dozens. I think it's well over 100 companies now we've worked with. I had data in my last book. I was going through a data set to go and analyze stuff, and the data I had was on 30,000 programs written in the PSP. So when people started questioning, you know, what do you mean? What are defect levels and things like that, I've got an enormous amount of data here. I mean, I've got data on thousands of students who have taken the PSP course and we've got before and after data. You can see how their quality work helps and what happens to their productivity, the accuracy of their estimating.

  • + Share This
  • 🔖 Save To Your Account