Burrell Smith on Design

from The Macintosh Reader by Doug Clapp

February 1992

 

 

The Macintosh logic board—the heart of the computer—was magic. Never had there been a computer that did so much with so few chips.

 

Compare the Macintosh to the IBM PC. The PC had a higher chip count on its video card—on the video card alone—than the entire Macintosh.

 

Or compare the first Macintosh to Lisa, as Jeffrey S. Young did in “The Journey is the Reward”:

 

“The digital design that Smith had come up with by January of 1981 was extraordinary. The Lisa, which by then had been in development for two years, was a planned product with a team of 24 hardware engineers and countless software programmers. A single working prototype of the Lisa machine existed … to work, it needed five circuit boards and a number of custom components. Over in a tiny suite of rooms in another building … a long-haired, 25-year-old engineer had made a new computer in a few weeks that was twice as fast and could be sold for one-third the price. It relied on a single circuit board and contained nothing but off-the-shelf parts.”

 

Without a magical design, Macintosh could not have existed.

 

 

 

 

DC: So, could you give me a few minutes on the Macintosh?

 

Burrell Smith: I’m not really interested in being, ah ... what would you want to do with it?

 

DC: I’d put it in a book I’m doing that’s an anthology of all kinds of things about Macintosh. The really interesting thing to me is why what you did [designing the Macintosh logic board] was so much better than what everyone else was doing at the time.

 

Burrell Smith: You know, it’s been so long ago, it’s sort of like, gee whiz, Mr. Smith, designer of the Macintosh ... it’s a little dated, for me. But what I’d say is ... that it’s having a team of people who have interdisciplinary knowledge.

 

It [creation of Macintosh] was essentially what Dr. Land [of Polaroid] did with his product. His little girl was using one of his cameras one day, and said, “Daddy, how come the picture doesn’t come out immediately?” And he realized that he had solved the problem, because he treated the answer as the problem and the problem as the answer. So that’s the answer: you make the film come out developed!

 

DC: That’s the way it should work.

 

Burrell Smith: And then he worked backwards to figure out “How you do that?”—but that’s easy. The basic idea is: what do you want it to do? Essentially, I think with the Macintosh that was the main thing: that what we wanted the computer to do was infinitely more important than how we were going to do it, and that’s really true for any discipline. The decision of what to work on is infinitely more important than how to actually do the work.

 

DC: So you aren’t be limited by anything. You just say, “This is what should really be done now.”

 

Burrell Smith: Well, no. What I’m saying is that you have to distinguish that from the normal engineer, whose job is really to perform a function. To a normal engineer you say, “Okay, Sam, I need a 16-megabit board instead of a five-megabit board,” and the engineer never challenges the notion that maybe we don’t need a memory board at all! We could do it optically, using Kodak’s new film holography or something—and have a gigabyte! So it s a matter of resonating the problems, as opposed to having the problems linearly defined.

 

And since no one claimed that they were able to completely define the problem space [of Macintosh], that really created the environment where we could resonate to the answer to the problem, and from there to figure out what the problem is.

 

It’s a completely different idea than saying that it’ll have this amount of RAM, this amount of ROM, and then going ahead and linearly implementing that. You determine “What is the answer?” as opposed to “What is the problem?”

 

DC: How did you even get to work on Macintosh?

 

Burrell Smith: Moxie, basically.

 

DC: Because today they would never let you work on it!

 

Burrell Smith: Well, you know, it’s hard to say. I don’t know if that’s true or not. I think the way it works in any corporation is that you get to do the work once you’ve proven that you’ve already done the work that they would give you. So if you design a computer first, like me, and get it running, then ...

 

The managers at Apple would come by and look at the original Macintosh computer running and they couldn’t say, “Can you do the computer or not?” They would be asking questions the next level down, which are, “Can you make the video invert the screen, so the characters are black on a white background? Can you make the screen bigger or smaller?” You know what I’m saying?

 

DC: Yeah.

 

Burrell Smith: See, once the answer’s fully defined, then the problem is solvable instantly.

 

DC: In one of the books about Apple, you’re pictured sitting in front of a piece of paper, which is covered in chips and components, and just moving them around until “it worked.” Is that true?

 

Burrell Smith: That’s essentially how it’s done. What I generally do is ... I have the chips ... I personalize the chips—which is not unusual. By personalizing, I mean that I ascribe certain personality characteristics to the components. Then I determine how the compendium of components would interact, essentially on a social basis or a biological basis, almost. And then I look for places where the interactions are unfriendly.

 

And that’s called ... metamorphosis is one term. Personalizing, I think, is a better term.

 

DC: So you’re trying to create a “good person,” in a sense.

 

Burrell Smith: Well, no, not exactly. On a chip level, like the serial communications chip and the state machine, it’s a matter of getting them to work together in a group. It’s just saying, “How do these components get along with one another?”

 

And the engineer is the manager of the logic board. So I became the manager of the logic board. Okay, so, you, serial chip, talk to this chip, and so on. It’s treating microchips more as biological systems that relate. It’s just a very right-brained thing. And I’m a very sensitive person, so I can tell very easily when the board, in total, doesn’t have the right balance.

 

DC: It’s an intuitive thing, rather than an analytical thing.

 

Burrell Smith: Well, right. I’m trying to pull together as much of the circuits and concepts simultaneously—you’ve got multiple levels, don’t forget. You’ve got the user interface level: how will this effect the person using it?

 

Then, below that, you’ve got the engineering level: how will this affect the reliability and manufacturing of the boards?

 

And then below that you’ve got another level, which is “Can these components actually be purchased, how will the various submodules of the machine relate?”

 

So those are at least three levels. It’s like the Star Trek chess game, where you’ve got the three tiers of the chess board.

 

DC: I get it.

 

Burrell Smith: That’s essentially what it is. You’re trying to solve the chess game on three different levels simultaneously. And there isn’t a good way to do that analytically, because it exists in at least three orders of infinity, if not more!

 

It’s more of a combination of intuitive and analytical, and I place enormous stress on myself. I concentrate very, very hard for eight, ten hours, even 24 hours in a row, just on a single problem. In that case [designing the Macintosh logic board], that’s what I had to do ... make those different levels work. Because the manufacturing people have to like it, the software people have to like it, the purchasing people have to like it. All these different levels have to be satisfied.

 

DC: Was there anything you remember as being really hard, or the thing that took the most time?

 

Burrell Smith: It doesn’t take any time. The hardest problem is communications between management and engineering, and getting the actual computer defined. Once the problem, which is really the answer, is defined, the implementation is really instant. I don’t take any time to design my products, virtually. Sometimes it takes a few weeks, but normally two or three weeks to do anything. I won’t work on it if it’s much harder than that.

 

That was true for the Macintosh logic board, that was true for the LaserWriter—which took even less time. That was only two weeks. It was true for the Radius Full Page display.

 

The Radius Accelerator took a bit longer ... it was very difficult to debug. That took more than six weeks. That’s the very hardest one because it has associative memory built into it.

 

I did the multiple versions of the Mac. I did the 6809 version of the Mac, the 64K with the original 68000 chip, the 128K, the LSI chip ... and I had strong feelings when doing the custom chip that they were going to fail, that the team was not integrated well enough and that I’d have to do another version. And so I knew it. When it happened, I just ran home and did it!

 

Once I’m comfortable with the general feeling I get from everyone, in terms of their finally having defined the problem well enough, then the actual design ... it takes me no time at all to do that.

 

So I do it with—the only way to say it is, with concentration.

 

DC: How did you learn this? Did you take any classes or just pick it all up from books?

 

Burrell Smith: Yeah, I took ... I started building projects on my own, you know, a Heathkit radio receiver project ...

 

DC: That’s what I did when I was in high school.

 

Burrell Smith: And ... lets see, I built a blue box after that, to do research in telephone communications systems—strictly, of course, legitimate. And then I took two courses in computers at Foothill College and took courses in electronics that I liked. A total of about ten courses at Foothill College—they have some very good professors there—and I just built projects.

 

The most important thing I tried to learn was, what was the psychology of being a good engineer? Their language—engineers have their own language—the way different chips are numbered, “Should I put a Schottky in because the power TTL is too slow?”

 

I do everything on the right side of the brain, whereas other people use the left side. It’s the biggest difference between myself and other designers, because I deal totally on how do I feel about the design. And I can do the detailed engineering on the left side of my brain, but only after my right side says, “Go do it.”

 

With most people, the left side dominates the right side when you’re in the engineering field. I think, “How do I like the way the circuit board looks? If I had that chip there, would it make it look prettier?” I just use a free-flowing set of design alternatives.

 

DC: So there’s an aesthetic? It really works better when it looks better?

 

Burrell Smith: No question. There’s no question it does. And now it’s becoming more of a science. And back then, that was one of Steve’s better notions: that the computer ought to look good.

 

If you start with that as a filter for any project—that the project has to look good—if you make that the highest level criterion, it will always work.

 

 

These days, Burrell is engaged in solitary, scientific research not related, directly, to computers. Someday, maybe, years from now, we’ll see his magic again.