A Software Maverick in Champaign
Chicago Software Newspaper, June 1996. Stephen Wolfram, a gifted and original mathematician fled academia to do software HIS way...
Although Stephen Wolfram needs no introduction thanks to the success of Wolfram Research, Inc., and its product, Mathematica, he does deserve one. Consider some of his accomplishments. He published his first scientific paper at age 15 and completed his Ph.D. in theoretical physics at Caltech by age 20. Then, in 1981, he became the youngest recipient of a MacArthur Prize Fellowship. Since the early 1980s, he has continued the pace with more original research and two different software companies.
CSN: You've spent a lot of time in business and working in pure science areas. Do you prefer one environment to the other or have any observations on the two worlds?
SW: There is science, and then there is academia. When I was a professor I used to say that business was my hobby and being a professor was what I made money at. Now I say that if one wants to do basic research and you have no other choice—go be a professor. But if you have a choice, don't be! In the 10 years since I left academia it has decayed even further than when I was there. That whole university world has expanded and increased its mass of mediocrity. Similarly, its repute has gone down. I remember that it used to be common here in the US to address people with a Ph.D. as "doctor." That happens very rarely now because most people have lost their esteem for academics.
The problem is that universities have no clear cut goal. If you are a professor what is your point—simply to be a professor? In business I have found that it is the opposite. In the academic world people may have good ideas but most of them, at best, end up in the form of a paper published in an obscure journal with a circulation of 500. In business, on the other hand, you can take those ideas and move them forward rather than just putting them out in the ether....You can take them to the market and explain them to people.
People in academia view business as nonintellectual. They are wrong. It is an intellectual activity where you use a broader kind of intelligence. For me, I end up using a lot of different kinds of thinking combining the intellectual with exercises in figuring out what the point of the activity is. In fact, many academics have tremendous troubles making the transition to business. They think they should focus just on the technical issue and not have to engage themselves seriously with other kinds of thinking, say figuring out how to motivate someone to buy something.
For me, what I like about business is that you can think about a whole variety of things. And in my business if I have an idea I have lots of leverage for turning it into a reality. I have a great team of people and they too come up with lots of ideas. We can make things happen. It isn't like hoping some unpaid graduate students will do you a favor. These are people who want to make a career of making things happen!
CSN: What prompted you to develop Mathematica. Did you perceive a market need?
SW: I started science when I was very young—about the same time I first got introduced to computers. I wanted to try to use computers for science and I began to try to do more and more sophisticated programs.
By the late 1970s, I was unhappy with the programs that existed so I started a project to do a particular kind of science. That was my trial run. I finished that effort around 1981 and started a company out in Los Angeles, where I was living, to sell it.
It was my first company, Inference Corporation, and I made many mistakes. I was a 21-year-old physics professor and had an idea that I didn't know enough about business to do a company properly. So I found some pedigreed business folks to do it for me. The fact is that with some intelligence and common sense you can do a lot better than pedigreed business people. What happened is that the company went forward and I kept saying I think this decision or that decision is wrong or stupid....Well, it turns out I was right most of the time. That gave me more confidence.
Subsequently, I left and the company has gone through all sorts of changes. In fact, to my surprise, it even went public about a year or so ago....
After that I went on doing basic science and then in the mid-1980s there were two things that happened to me. I needed to do computer experiments more easily and I found that all my time was spent doing programming. I thought if I could write one big program that was flexible I could have it do most of what I wanted. That was the first motivation.
The other motivation was that I could see the personal computer getting powerful enough to run serious, large environments for technical computing. Up until then, most of it had been done on mainframes and minis. It was coming to the point where many people would be able to afford the technology and many would want to use it.
So, around 1986, I started building Mathematica. I guess the second time I did things very deliberately in terms of designing the product and setting up the company. For one thing I had no outside investors and no senior business people until the business was well off the ground. Mostly those two decisions have worked out well. The intellectual underpinning of Mathematica is that is provides a comprehensive environment for technical computing. It pulls together a lot of algebraic capability and user interface functionality. All of that was made possible by my ideas for making a new programming language.
If you look at most of the dominant programming languages, they date back 20 or 30 years or more. They were devised when computer time was expensive compared to programmer time. Making it easy for the computer is the central idea. Even with Java or Visual Basic you still have that basic paradigm. I thought that if we wanted to make a language based on how people think of a problem it would have to be more direct. If you compare programming for a given task in Mathematica and look at similar work in C—it is about 20 times longer.
Although what we created was a powerful general language we haven't concentrated on that. We knew that people would say, "I know my own language, why switch?" So we concentrated on saying this is a great way to do calculations—something useful that people wanted to achieve. Now, gradually, more and more are using it as a programming language. I would say seven-eighths of our customers use it a little and half do so in a serious way.
It turns out, too, our language can do nice things on the Web. You can build a program that is also a Web site, very nearly. So this is now another area we are working hard to develop. It is still the same intellectual base but applied in a different domain. What is frustrating for me about working in the Web is that we are not unique; everyone is working in the Web. That's not really where I enjoy being.
More broadly, the thing that has been frustrating is that we know people could find this more useful but it has been a difficult marketing problem, figuring out how to present it. They will say we must spend too much money to convert. It if was free they might think about it but if you want to charge real money....
However, we've had our hands full up until now with the technical market.
Of course, programming languages are not short lived. Indeed, the serious ones seem, so far, to live forever. We think Mathematica will live for a long time and we may eventually develop the energy to package it for other domains.
CSN: There are now dozens of firms developing addition products using Mathematica. Are many of them in Illinois?
CSN: Are some of them spin-offs, formal or otherwise, from your firm?
SW: Yes, at least half a dozen. Most of them, though are on the West Coast or in Colorado or something. Mostly it has been a case of someone not wanting to live in Champaign—though in a few instances I've said to people they would be better off working on their own rather than for me.
CSN: Have you had difficulty finding an appropriate staff?
SW: The way the company has evolved and the kinds of people who work for the company.... we have lots of "intellectual" types. Many are very bright "misfits" from academia who were too entrepreneurial. We have people who range from distinguished former professors to those who dropped out of high school. I challenge people to tell the difference!
CSN: Was the MacArthur grant what enabled you to start Wolfram Research?
SW: No, that wasn't really enough. I made some money from the previous company and I made money doing consulting work, too. Also, the costs of launching a software company aren't that great. I tell people that if a software company costs more than a million dollars to launch you are doing something wrong. Wolfram Research didn't cost that much to start.
CSN: And you indicated there were no outside investors?
SW: One of the things we have done that is somewhat unusual for a software company is avoid venture capital and going public. Many companies seek venture capital and then work to puff up the company for four or five years so they can go to the market....My company has been built up as a stable, long-term entity. That would be hard to do with outside investors. Similarly, I could probably have gone public and I would have a big net worth on paper but I decided not to because I want to do things I think are valuable.
CSN: You are now again dividing your time between company business and pure research. What is the focus of that research?
SW: The basic research I am doing is a grand project...it is mostly me and my computers and a few people who help. We are not being funded by a grant from the NSF. They might be willing to give me a grant but that kind of funding is no way to live.
What I am trying to do is set up a new direction in science. If you look at the history of science, for the last 300 years, when people wanted to describe or model something in the natural world they looked to mathematics for an equation. That worked pretty well for some things, say for describing the movement of the planets, but less well for things like biology. What I am interested in is what we are going to do if we can't adequately explain things with mathematical terms.
Assuming the universe runs on rules is there any reason they should conform to mathematics, something invented by humans? I'm interested in other arbitrary but logical rules. Specifically, I am arguing that the universe is in some ways like a computer program. I study how programs typically behave, particularly very simple programs. Looking at simple things, like dropping two balls from the Tower of Pisa, is how science began hundreds of years ago. No one has done anything similar with these simple program actions. I have been looking at this since the early 1980s and what I find is that those simple things are remarkably like phenomena in nature.
Now my task is to present my ideas. I don't want to just put them in a technical journal. I have been struggling for five years with this magnum opus—a book project which I hope to finally complete this year.
But of course it isn't easy when I'm also trying to get the next version of Mathematica out the door.