The previous Bush administration used surveys to form a Cabinet. President, elder, Bush was in Moscow, and he met with Gorbachev. And he was having trouble with the vice president at the time, Dan Quayle. So he was impressed with Schevardnadze, and he asked Gorbachev, “How did you get this guy?” And he said, “I gave him a survey. I asked him who his father’s son was, who wasn’t his brother. And Schevardnadze said, ‘Me.’ ” So Gorbachev hired him. So, elder Bush said, “Look, I can’t wait to get back to D.C. so I can use this survey!” [laughter] So he called in Dan Quayle and said, “I have a question for you, which is kind of important to your career. Who is your father’s son, who is not your brother?” And Dan Quayle said, “I think I need two days on this.” So Quayle went to see Dick Cheney; Dick Cheney was Secretary of Defense back then. And he said, “Who’s your father’s son, who’s not your brother?” And Dick Cheney says, “Well, it’s me!” So Quayle went running back to Bush and he says, “I’ve got the answer!” And Bush says, “Well, what’s the answer?” And Quayle said, “It’s Dick Cheney!” And Bush said, “Well, that’s wrong—it’s Schevardnadze!” [laughter] So surveys are used a lot in D.C., here … and in the spirit of that, I’m going to talk about software.

I want to talk about visualization tools and metrics. And I’ll first talk a little bit about the business—what it’s like to be in the business of analyzing a lot of software, in different languages, on many computers, with metrics. Then I’ll talk about algorithms, and I’ll give an example of how to quantify complexity. What’s struck me is that there’s been a lot of discussion about complexity—I think maybe intuitive complexity—but yet [finding] the point at which things get uncomprehensible, untestable, unmaintainable … [answering] that question, certainly, is pertinent to software. And that happens to be the particular work that I’ve done.

I’ll give first a definition of it and show how it works, and then I’ll show a lot of examples of algorithms, in increasing complexity. There’s a point at which most of us [lose] sense of what these algorithms are doing, and I realize that this is a long stretch here—so we’ll give a quiz, and see where people lose the comprehension.

And then we’ll talk about using similar metrics for testing and maintaining as well. It turns out that, in software, maybe 20 percent to 30 percent of the effort is in development and 70 percent is in maintaining. And I’d bet that’s probably the case with surveys. But yet all the attention is on development when a lot of the effort is in maintaining. So we’ll talk about how to use metrics in a maintenance activity. And then we’ll talk about architecture—how to take a very large system and see a lot of software on a screen, and interact with that, and how you get a sense—beyond documentation—of the shape, the quality, the texture, the layout, the landscape of a very large application.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement