Winner of the 1993 Price Waterhouse Leadership Award for Lifetime Achievement,
Computerworld Smithsonian Awards
Location: Microsoft Corporation, Bellevue, Washington
Interviewer: David Allison (DA)
Division of Computers, Information, & Society
National Museum of American History, Smithsonian Institution
Part 1 Page 2 Page 3
Page 4 Page 5
Microsoft as a Separate Company
DA: So, you were convinced that you should setup your company to deal with a variety of different computer vendors?
BG: We had Microsoft as a separate company. In a certain sense, if things hadn't worked out, I could always go back to school. I was officially on leave. I didn't have a family to feed or anything. But I was doing the payroll, writing the taxes, doing the contracts, figuring out how to price the software. In fact, I was business-oriented enough that I wrote a letter about software piracy, sort of complaining that a lot of these computer groups weren't paying for their software. That really became a cause celebre at the time: "Is it fair that this guy is asking for money? Should we pay for this stuff?"
MITS was very controversial because some of the memory boards they had been shipping didn't work. They'd been late with a lot of things. So, some people felt like it was a way of getting back at MITS to take the BASIC.
We had the first computer convention. Their people came in. So, Microsoft was a business from the beginning. Not that we had any clear view that it would ever be a large business, but I had to pay these friends that I had hired. At a minimum, I had to make enough money to write their paycheck, and if I got enough confidence that we could sell a lot more, then I would be able to hire even more people to get ahead, to be the leader in doing lots of products that could share code with each other, and take the market.
DA: Before we move on, there is something you might want to show us, for the record, on the terminal there to get a sense of what BASIC was, and how it was used?
BG: Okay, this Altair is actually a running Altair. And what we've done is we've loaded the BASIC into its memory. So, this has an 8080 chip in it and 16K of memory. So, at this time, this would have been a big machine, 16K. It is connected through a serial card that is in the bus here. This bus is an S100 bus. Up to this Lear Signal terminal. These were fairly cheap terminals of the day. But you can tell that it's not really very dense because this was before they had cursor keys on these terminals. And so this is the manual for this BASIC.
We did different versions of BASIC. We did a 4K BASIC that was the most modest. We did an 8K BASIC. We did an Extended BASIC. This is actually Extended BASIC here. Later the most advanced version we did was a Disk BASIC, which is what you see here as the early floppy disks -- this is actually a hard-sector disk -- these little holes you see there. That is a sector marker. This is how IBM originally did these floppy disks. These did not work well at all when they first came out. You couldn't take a disk from one machine and run it in another.
But the thing that I wanted to show you -- we've actually got it running here. I can list out this little program I typed in and run that. That is my little sign wave, printing program. One of the nice things about this BASIC is it has this so called direct mode. So you can PRINT 2 + 2. It prints the square root of ten. By now, we had things like this edit command, where you can say, EDIT, like and sit there and step through and change what is in the line.
This BASIC, although it was very much modeled on what had been at Dartmouth, which was what I encountered when I used that first GE time-sharing machine. We'd gone beyond it in a number of ways to let people add machine language sub-routines. Even new words in the language like PEEK and POKE to let you read and write memory locations, were starting to come up here. I'm a big believer in interpreted languages, not only from the beginning of computing, but the future of computing. It was really the right approach, because you could just type the thing in and immediately see what was happening. And yet you could add new capabilities very easily.
Another machine in that time was this IMSAI machine. In fact, that was the next company to come out. It was sort of the first kind of clone-thing going on in the PC world, because they used the same bus that this computer here the Altair used. And so the same add-in cards would go in. So we were able to license our BASIC over to this company.
When MITS got that kind of competition, they actually went back and looked at their contract and were kind of disappointed to see that they hadn't tied up exclusive rights. In fact, they made it very hard for us to license BASIC to other people. And there was an arbitrated dispute where they were told that they were completely wrong to have done that; and in fact, had their license terminated. They were bought by a peripheral manufacturer called Pertec, and actually ended up disappearing fairly quickly, as did this company here. Most of our early customers were out of business very early on.
The People who formed Microsoft
DA: Bill, let's talk a little bit about the people that formed the early Microsoft.
BG: Paul and I were the founders. During the time we were in Albuquerque, which was 1975 to 1978, we ended up with about sixteen people. This is a picture we took towards the end of that time [Bill refers to the group photo taken in Albuquerque], myself and Paul. This is Gordon Letwin, who had worked and did the Heath BASIC, Benton/Harver BASIC. Then he was upset when they were licensing my BASIC. So, he came to work for us and did some incredible work.
Marc McDonald was actually our first employee. Other than Andrea, who wrote the manuals, and Marla, who helped keep the books. I was the Sales Department, Contract Department. Everybody else here were programmers. We all wrote an immense amount of code. These were exciting years. The number of new machines coming out were pretty dramatic.
Our offices were here in fancy Albuquerque, up on the eighth floor of this building here. Albuquerque was great. There weren't many distractions there, but it was hard to recruit people as we tried to grow.
DA: What was the culture of the company like?
BG: Well, I would program sort of night and day, and tell people, "Hey, we promised this thing to be done in a few months we've got to get these things done." We were so aggressive at just getting things done. Like committing to write a ROM BASIC that would fit into 12K of ROM. It was fun because, I at that time, got to look over all the code that people did, and talk to people about where we would go with things. It was just a very small group, and yet between the new machines being done in Japan and the U.S., every week something new was happening.
DA: Did the other people share your intention? Were they there as many hours as you were?
BG: Well, certainly most of the people did. And we were all quite young. I may have set the most extreme example. But, the work was really fun. We always had deadlines that we ended up committing to that ended up being very challenging.
A great example of that was the work that we did on this TRS-80. The Altair BASIC comes out in 1975. In the next big wave is a set of three machines that came out in 1977. The TRS-80, the Apple II that we have over here that came out, actually, without the disks at first, and a machine called the Commodore PET. And those were low-cost and yet, they weren't kits or anything. They came out prepackaged. And they looked like they would really ignite the volume in the market. And all three of them went out and did very well.
Several of them, when they first shipped, had BASICs that the company themselves did. This had a Level I BASIC, that actually Steve Leininger enhanced off of Lee Chin(?) Wang's Tiny BASIC. But they knew that it was pretty inadequate. And so they licensed the BASIC from us that was built into all these machines thereafter that they called Level II BASIC. We even left some hooks in there so we could sell a BASIC called Level III BASIC, that went even further.
So, Radio Shack, with its distribution and its name set the market on fire. Apple, because they really went out to computer dealers and did a good job, far better than people like the MITS guys and the IMSAI guys. They really thought of this as a market where they had to develop the channel and do new things. And the Commodore machine, the PET, was actually the most aggressively-priced machine. It had some very innovative things. These machines drove the market and eventually, a year after they were out, all of them had our BASIC built-in. So you could even move programs back and forth between these machines because of the compatibility that we had built in there.
What Distinguished Microsoft Basic
DA: You mentioned that often they started with their own BASIC and then they came to you. What was it that made your products stand out and made them come to you?
BG: Well, our BASIC was fairly deep. The BASIC they first put in this machine was really, really limited. It just wasn't going to be expandable. And they wanted to put on graphics. They wanted to put on a disk. They wanted to have sound. And we knew how to do those things. We went in and showed them that we could help them design new machines, really work in partnership with them. And do it even less expensively than what they could in trying to manage software development themselves.
In the case of Apple, Wozniak had done the Integer BASIC. And he was playing around thinking about doing a floating point BASIC. I don't know why he never got around to it, but they knew they needed one, and so they had Jobs and Randy Wiggington came out and talked to us. And I put the cassette extensions in integrated into their ROM. And that became what was called AppleSoft BASIC.
In terms of Commodore PET, they started with us from the very beginning. Because we helped Chuck Pedal, who was at Commodore at that time, really think about the design of the machine. Adding lots of fun characters to the character set, things like smiley faces, and suit symbols. That was the first machine we did that had this wild extended character set. All these machines started out using cassette-based storage, where we could store about 1,200 baud worth of data on these cassette tapes.
So, this was really a generation of machines. These were the popular 8-bit machines. There was another type of 8-bit machine that was a little higher end, that were the CP/M-80 machines. All which ran this operating system from Digital Research. We wrote our languages to run on top of that. That was Gary Kildall's company.
That really defines the 8-bit era of computing. I mean there were some refinements. Like we can see here that Radio Shack built a portable version of their computer. And this is the Osborne computer, that's actually a CP/M-based computer. This is an attempt at portability in this 8-bit era.
We're actually running, on this machine here, one of our first applications. We were broadening out by now to a lot more than just BASIC. We had a portfolio of languages. But we also had a spreadsheet and we were about to come out with a word processor.
So, this is the screen appearance of this spreadsheet called Multiplan. It is kind of an obscure footnote in history, because VisiCalc that came before it, the real creator of that category did incredibly well in the Apple II. We designed our spreadsheet to work on 8-bit machines and the next generation. But a 1-2-3, which was only designed for the 16-bit machines got ahead in terms of features. Multiplan, although contributed a lot of good ideas and, actually, was essentially passed by with the work that Lotus did on 1-2-3.
Defining Microsoft Corporate Strategy
DA: Bill, you skipped quickly through a notion of change and strategy in your corporation. First you started with languages and then you started branching out. Can you talk a little bit about the thinking and talking that you and Paul and the others did about what strategy Microsoft as a company should have?
BG: We always knew that we didn't want to have a single product that was a dominant product. We wanted to hire in more software people and have a full product line. In a sense, one of the earliest things we decided to do was to make available on the microprocessor everything that had been available on the mini computer. And that is why we did the languages, COBOL, Assembler, all the normal tools that you would expect to be able to do with native software development. But, it was clear that as an individual machine there were some things that you would use that you didn't need on a mini computer or mainframe.
And already there were other software companies, although, we had been the very first micro computer software company. A company called WordStar had emerged. One of the marketing people from IMSAI, after IMSAI wasn't doing so well, went out and focused that company around the WordStar word processor. Actually, the company was originally called MicroPro. And Seymour Rubinstein ran that. There was Michael Shrayer with Electric Pencil. And we had come out with games very early on. Here is the so called, Adventure, game running here on this TRS-80. This is an early box that we used in our packaging. We called it, that group, Microsoft Consumer Products, at the time.
One of the products that really got us kicked-off in doing new things, was we made a card that plugged into the Apple II with the little Z-80, a microprocessor that succeeded the 8080 on it. So, it let you run all of the business applications that were coming out in this CP/M-80 world, on this machine which was very high-volume. That was a big success and let us grow our retail group, and even go out and license some products like Flight Simulator and many others that defined a whole new avenue for us.
We never saw ourselves as limited, as in terms of what kind of software we wanted to do. As long as it was software where development talent was the key to doing it well. And that it could be sold in fairly high volumes, which in those days weren't nearly what they are today. We decided that might be a good product for Microsoft.
DA: Do you feel like you saw the movement towards the business market out of the hottest market soon enough?
BG: Well, the key to the move to the business market was getting a machine with more capability, and getting a standard, although you could exchange some basic programs between these 8-bit machines. If you went and used any of the advanced features then you were just tied to that one machine. So, it wasn't all that economical to write lots of software for these machines. And the 8-bit machines maxed out at 64K of memory. And even though I wrote BASIC in 4K, 64K seemed like a lot, but as we started taking on more and more challenges, it just wasn't going to be enough. And since we were tracking what Intel was doing very closely, in building the next generation chip, their 8086 family, including the very low-cost part called the 8088, we knew that 16-bit computing was on its way. We saw that it could be a good business machine and we decided to focus a lot of early work on that Intel chip.
In fact, it was that decision that forced us to do the SoftCard, because we had so many products for the Intel chip, and the question was, "Should we spread those products over to other 8-bit chips, like the 6502 that runs in here? Or, should we immediately move up and do 16-bit software?" And I said, "No. We are going to do 16-bit software." Everybody was a little bit disappointed because it meant that we wouldn't be able to sell onto these machines. That is when Paul invented the idea of the SoftCard, so that we could actually take our Intel software and run it on this machine, and, at the same time go ahead and devote our resources to being way ahead of everybody else in developing software for the 8086.
DA: You mentioned that you were tracking the developments throughout the hardware side of the industry. How easy was it to get information on what was coming available? Were people open in those days? Or were they pretty secretive?
BG: Well, there was no secrecy at all, at least as far as I could detect. Everybody in the industry would be at shows like the West Coast Computer Fair. And there was so much to do -- we were overlapping each other some. There was some good rivalry, but not in a sense that people were keeping lots of secrets about what they were up to. It was a very exciting business at that time.
Intel was a customer for our BASIC. They came out and asked us to do some custom work. I remember telling them that I could do it in two weeks. And they said, "Don't say that, don't say that -- say four months -- say something reasonable." And it turned out that it took four weeks to do, because configuring their system was so hard. But we'd gotten to know Intel. And we were talking about where the chips were going. It was a very small group that sort of shared this secret. In fact, Ted Nelson gave a speech at one of the West Coast Computer Faires about how we were going to overthrow all the big computer companies, and we really knew that it was "power to the individual". There was a little bit of a minority feel, that we had to advance this cause. And, eventually everybody would realize that we were right about what was going on with these machines.
DA: Was that shared by the people at Microsoft?
BG: Oh, absolutely. Microsoft at this time, I think our average age was 22 or 23 years old. People enjoyed the programming. Some of the people didn't ever go off to see customers. They just stayed and did the work. But, it was a very small group. And when I came back from a trip, I'd always talk about what I'd been up to and what we were seeing out there in the marketplace.
DA: Talk a little bit about how your company grew through this period as you started moving around into the business applications, different machines, you added employees. How did that work?
BG: We were in Albuquerque until the end of 1978. And that takes us up to where we really completed all the languages we wanted to do in most of our 8-bit software work. So, as we were moving up to Seattle in 1979, we had some custom extensions to do for some Japanese machines of that time, like a NEC machine and an OKI machine. But, a lot of our resource was already focused on 8086 development software. We just saw that as the coming thing. But moving to Seattle let us expand our personnel quite a bit. That is when the Multiplan, the spreadsheet development started. That is when Microsoft Word started.
In particular, we hired a number of key people like Charles Simonyi, who had been a founder of the Xerox Palo Alto research labs, and had shown us the Alto computer and talked to us about graphical interface. We shared a common view that low-cost microprocessors would be doing that, and got him on board to help us write applications that would eventually become very graphical applications.
So, when I moved from Albuquerque in the very, very beginning of 1979, there were the sixteen people. Then, about a year after that, I hired a friend of mine from college, Steve Ballmer, who is very good at hiring people. He could see that we had more projects that we wanted to do than we could. He was able to almost double the size of the company and people every year for the next five years. So, it really started to change in character where I had written the high percentage of the code myself until we got to Seattle, and reviewed everything that people were doing to the point where we were setting up a lot of autonomous teams, and having to do a lot more in terms of what was our methodology, and how did we interview people. Just to stay up with all the projects we were going after.
DA: Did you continue to look for the same sort of people?
BG: Well, certainly in development we did. I mean, for the first time we hired a salesperson. We hired a finance person. So, we had some new categories of skill that were things that I had essentially done, but that were more demanding now. We hired in Vern Rayburn to run this consumer products effort, to think about packaging, pricing, promotion, and how to get those things out there. He hired quite a few people to help him work in that area. The Development Group, which has always been the core of Microsoft, there it was the same kind of person coming on board, but the company had more aspects. In fact, we started our activities in Japan and Europe. So, we were getting people in many different places. Even though Development we kept all here in Seattle.
DA: Was everyone in the company happy about the move to Seattle? Or was that a matter of concern for many people?
BG: Well, it took a while for me to sell the idea to everybody. But, in fact, everyone, except my secretary, decided to make the move. Albuquerque has its advantages; warm weather, a nice place. So, it took quite a bit of selling. But, everybody was very involved in what they were doing, and there was the excitement behind where we were going. So, I was able to get literally everyone but my secretary, Miriam Lubow, to come up with us. And that became the core group. But Seattle was a lot better for hiring, particularly once Steve got in and really increased the rate a great deal.
Our hiring was always focused on people right out of school. We had a few key hires like Charles Simonyi who came in with experience. But most of our developers, we decided that we wanted them to come with clear minds, not polluted by some other approach, to learn the way that we liked to develop software, and to put the kind of energy into it that we thought was key.
Part 1 Page 2
Page 3 Page 4
Page 5
David K. Allison, Chairman, Division of Information Technology and Society. National
Museum of American History.
Research specialties: Computer technologies; military technology; social history
of technology.
B.A. (1973) St. John's College; Ph.D. (1980) Princeton University.
09/27/1994 Final Edit Kris Kaeding
06/28/1995 Webification by David Allison and Mikel Maron
Copyright 1994