Technological evolution of SAS
In this excerpt Goodnight recalls some of SAS's early competition and the additional features SAS offered to make it desirable. He recalls the evolution of SAS over time, including the modifications it required to run on different operating systems. Goodnight did not predict the PC revolution.
Citing this Excerpt
Oral History Interview with Jim Goodnight, July 22, 1999. Interview I-0073. Southern Oral History Program Collection (#4007) in the Southern Oral History Program Collection, Southern Historical Collection, Wilson Library, University of North Carolina at Chapel Hill.
Full Text of the Excerpt
JM: Can you describe the approximate character of the market you were working in those years, in the late '70s? Was there any comparable product out there that you could identify?
JG: I guess our number one competition in those days was SPSS. There was also another package called BMDP. I think in the late '70s those were our main competition. They were a hundred percent stat packages. One of the things that we felt made SAS more popular was the fact that we had things to print out reports and do things that aren't just purely statistical. That gave SAS more flexibility than just a pure packaged program for statistics. So as we began broadening SAS to do other things, we added modules for economic and time series data. We added an OR package later on. We added graphics in the late '70s early 80s, which really made us quite a bit different than other packages. Graphics back in those days were fairly primitive. You had these box, these little pin plotters, with pictures with pins. They would change. They had four or five different color ink pens and the little arm would go and grab an ink pen and go out there and draw. It was quite fascinating to watch in those days because it was so new. Now you've got laser jets and all that stuff, so it's quite a bit different now.
JM: Tell me a little bit more about the early marketing effort and -- you mentioned hiring Bill Gjertsen -- how collectively you defined the marketing task and went about it.
JG: Well, marketing to me is something that I just as soon let somebody else do. We were basically programmers and we don't enjoy selling. I never did. Jim never did. John certainly never did. What we were interested in was doing development work. I think we hired Bill and we hired another fellow about a year later. The two of them began divvying up the country and hiring sales people. We began hiring more programmers to shed some of the work that we had. By that time we were maintaining about a million lines of code. When we left the university we had about 300,000 line of code and over the next few years we had grown to over a million. Today it's over eight million lines of code and still growing. We supported one single operating system, called MVS back in those days, on one computer. So when we left the university it was one computer, one operating system that was it. Once we left, we decided we needed to run on one of the other operating systems, CMS. So we did that. Eventually we hired someone that knew IBM's DOS operating system to work on that. We began to expand the number of platforms that we ran on. In the early '80s, we completely rewrote a large part of the core of SAS to give it more portability. That's when we began running on mini-computers like VAX's and VG's and Prime's. By the mid-'80s when the PC came out, our users began asking us if we could run SAS on a PC. That was quite a challenge at that time because we were up to several million lines of code and to try to cram that into 64K or 128K, or whatever the size the computer [was] back in those days, was a major, major task. What we did was to decide if we were going to rewrite SAS again, we were going to choose a new language. We chose "C" because "C" was a more popular language. PL/1, which we had been using, was pretty much an IBM only language. Some of the other mini-computers supported it, but it was not as robust as "C". We decided to move all of our development language into “C.” We made the decision that we would not write a separate PC program, but rather rewrite the system with the idea that it would run across all platforms. We call that our multi-vendor architecture. We're still building upon that architecture today.
JM: Were you anticipating the PC revolution or just [thinking that] “PCs are another part of our market. Who knows what will become of them.” Did it seem that you had already approached this watershed?
JG: The thing about PCs in the mid-eighties is that they were not very much use for anything, but people had their spreadsheets on them. Then they did a little word processing on them. That's the main thing we thought back in those days. Gosh, now they're more powerful than our mainframes. I've got a box sitting over on my desk that runs faster than my mainframe does. We certainly didn't foresee the revolution. If I had, gosh, I'd have bought a chunk of Microsoft.