Leland Wilkinson Interview

Leland Wilkinson is Executive VP of SYSTAT Software Inc. and Adjunct Professor of Computer Science at the University of Illinois Chicago. He received an A.B. degree from Harvard in 1966, an S.T.B. degree from Harvard Divinity School in 1969, and a Ph.D. from Yale in 1975. Wilkinson wrote the SYSTAT statistical package and founded SYSTAT Inc. in 1984. After the company grew to 50 employees, he sold SYSTAT to SPSS in 1994 and worked there for ten years on research and development of visualization systems. SPSS eventually sold SYSTAT to Cranes Software International and Wilkinson rejoined SYSTAT in 2008.

Wilkinson is a Fellow of the American Statistical Association, an elected member of the International Statistical Institute, and a Fellow of the American Association for the Advancement of Science. He has won best speaker award at the National Computer Graphics Association and the Youden prize for best expository paper in the statistics journal Technometrics. He has served on the Committee on Applied and Theoretical Statistics of the National Research Council and has been Vice Chair of the Board of the National Institute of Statistical Sciences (NISS). In addition to authoring journal articles, the original SYSTAT computer program and manuals, and patents in visualization and distributed analytic computing, Wilkinson is the author (with Grant Blank and Chris Gruber) of Desktop Data Analysis with SYSTAT and The Grammar of Graphics.

Dr. Wilkinson, you might be called an “academic entrepreneur” or an “entrepreneurial academic.” You have had a very solid academic career in psychology and statistical computing. At the same time, you founded a software company, SYSTAT, that at one point employed 50 people and netted you millions. Over the years, you’ve kept feet firmly planted in both the academic and business worlds. Yet your early academic experience would not have suggested such a career trajectory. With undergraduate and graduate degrees from Harvard in religion, you were for a time a chaplain. Could you explain how you got from point A (as a Harvard undergrad focused on religious studies) to point B (a psychology professor writing the state-of-the-art statistical computing program SYSTAT)?

That route was rather circuitous. I had taken a lot of statistics and mathematics in secondary school (thanks to the Sputnik-inspired New Math curriculum) and had begun as a math major at Harvard. Two weeks into my Freshman year, I decided I had had enough with being a geek, so I switched to English. I decided to study religion after visiting Harvard Divinity School and finding the place intellectually challenging, with a brilliant and inspiring faculty. I left Divinity School hoping to get a Ph.D. in Psychology at Yale so I could become a full-time chaplain. When I got to New Haven, however, I soon found myself at the Yale Computer Center doing statistics homework. I was seduced almost immediately by the computer. I probably should have switched to the statistics department at that point, but I was already in the Ph.D. psychology program with a quantitative advisor. And, most important, John Hartigan and Frank Anscombe in the Yale statistics department were supportive of my efforts to learn psychometrics and mathematical psychology. I could have made good use of a statistics Ph.D., but did my best to make up for that in the following decades by programming and teaching myself graduate statistics. In the end, however, I have no regrets because what I now do (visual analytics) requires a knowledge of computer science, psychology, and statistics. And my wife, whom I met in divinity school, is now a Lutheran pastor. We talk about theology, biblical studies, and the church almost every day.

You wrote SYSTAT in the early 1980s during your sabbatical year as an associate professor of psychology at the University of Illinois at Chicago. Sabbatical years free professors from teaching and other duties to pursue research. At the end of that year, you had produced a computer program rather than the usual stack of research articles. That was an unusual move on your part — some would even call it gutsy. How did your fellow faculty at the University of Illinois at Chicago react to this use of your sabbatical time? What was their reaction when you turned this sabbatical project into the commercial venture SYSTAT? Since your income during your sabbatical year was paid by the University of Illinois at Chicago, why didn’t they own the rights to SYSTAT?

Some were not very supportive, but most were proud to have a faculty member succeed as an entrepreneur. Nowadays, of course, faculty in computer science, biology, and statistics form companies all the time. I was very careful to separate my SYSTAT work from my academic work. I continued to publish and do research during that time. And I did the SYSTAT programming on my own time. Also, I shut down my university computer account and wrote SYSTAT on a microcomputer I built with my own money. It was the same vintage machine, predating the IBM PC and Apple, that Bill Gates used to write Microsoft BASIC. When SYSTAT grew too large for me to give full attention to my teaching, I gave up my academic position and my tenure. Although the university did not have intellectual property rights to SYSTAT, I gave the university a free site license to the program.

You faced several pitfalls in getting SYSTAT off the ground. The publisher McGraw-Hill offered to buy your program (at minimal cost to them) and distribute it, but would essentially have reduced you to an indentured servant in keeping the program updated and free from error. How did you recognize this pitfall (and others) and what lessons did you learn that might help other academic entrepreneurs in getting their projects off the ground? How did you decide on the business model through which SYSTAT ended up thriving?

Interesting question. It was John Wiley & Sons who wanted to publish SYSTAT. My editor there was a rather brassy young woman who wanted me to sign right away. Another senior editor there cautioned me, however, and subsequently an editor at McGraw-Hill asked me why I wasn’t making a go of it myself. I owe the McGraw-Hill editor the biggest thanks because I hadn’t thought of going through all the work to set up a company. My wife, a lawyer at the time, also supported me in that decision and later handled the legal work for SYSTAT. The one caution I would give to entrepreneurial academics is that the work needed to found and grow a company will eventually force them to give up teaching.

From the get-go, you decided that SYSTAT should be compatible across platforms. In particular, Mac and PC versions of SYSTAT were available from the mid-80s. Why did you emphasize cross-platform compatibility and portability of underlying source code, and how did expanding SYSTAT’s applicability in this way pay dividends?

I loved all kinds of computers and was obsessed with making the same code run on them all. It gave us markets that competitors in the 1980s and early 1990s, notably SAS and SPSS, had ignored.

Your dual roles as an academic and as an entrepreneur have proven mutually reinforcing. SYSTAT remains a premier program in statistical computing for research statisticians and would never have gotten off the ground without the academic community’s input. At the same time, your academic work on graphical representation of statistical data has been abetted by your work on SYSTAT and has also found its way into SYSTAT’s graphic package. Perhaps you can elaborate on how these two sides of you have worked together.

I still consider myself an academic. I also work alone, spending almost every morning programming. When I got to SPSS in the mid 1990s, the head of development asked me how I was able to write so much code. I replied that I never went to meetings. What I have done for many years is to analyze real data and work with students and faculty who have interesting research problems. There is no better way to get research ideas than to teach and advise. I did no market research (and still don’t). I wrote software for myself and was pleased to discover that there were many people out there like me.

Your book “The Grammar of Graphics” (Springer) shows, on the front-cover, Minard’s famous graphic of Napoleon’s Russian Campaign (where shrinking line-width corresponds to Napoleon’s shrinking army — see below). In an age where scientific data sets can be huge (giga-, tera- and petabytes), constructing illuminating graphics that highlight key features of the data is becoming more and more necessary for scientific insight. How do you broach this problem in your book? How did your work on SYSTAT shape your ideas in that book? By referring to a “grammar” of graphics, are you suggesting that illuminating graphics are less the result of art than method? Isn’t Minard’s graphic of Napoleon’s campaign also a work of art? When grammar has done all it can with graphics, what, if any, is the remainder?

I used the word grammar as a metaphor (following Pearson’s Grammar of Science and Jones’ Grammar of Ornament). But I did intend to show that it is possible to formalize the rules for constructing graphics (statistical charts). I added proofs to the second edition and formalized the system to the point of enabling the development of a compiler for a Graphics Production Language. I vigorously opposed the discursive, subjective opinions of most of the people writing about visualization up to that time. And I’m pleased to see that my formalizations have now found their way into several commercial and open-source graphics systems. As I said in the book, the grammar of graphics isn’t about beauty, persuasiveness, or other subjective features. The artist/designer has a key role to play in the development of good graphics, but if they do not follow the correct mathematical rules, the result is nonsense.

SYSTAT faced three main competitors when it hit the scene in the early 1980s: SPSS, BMDP, and SAS. What distinguished SYSTAT from these other statistical packages? In what ways was it better? All three of these other packages remain to this day. Does that surprise you?

SPSS excelled at surveys. BMDP was for experts in medical research. SAS was the heavy-duty, ubiquitous engine. SYSTAT excelled at graphics, interactivity, and simplicity. I spent a lot of time designing a language without semicolons, arcane exceptions, and esoteric statistics that few needed.

You had a competitor (not one of the three just mentioned) who copycatted aspects of SYSTAT and then attempted to claim credit. You appealed to the court of public opinion, but then found yourself up against a devious countercampaign that attempted to turn the tables on you. In general terms, can you described what happened, what you learned from it, and what you would have done differently? We ask not to revisit a painful episode but to help readers who may be facing similar infringement of their intellectual property (a problem that has become increasingly prevalent).

I learned what every software developer now knows. You have one year before somebody copies what you’ve done. The only defense is to innovate continuously. Lawsuits and patent disputes are pointless. I never sued anyone and was never sued. I must admit, however, that I had a team of lawyers from one of the biggest firms in the country (thanks to my wife) to ensure that never happened.

You ended up selling SYSTAT to SPSS (“Statistical Package for the Social Sciences”). In the early 1980s, SPSS’s statistical algorithms left much to be desired whereas SYSTAT’s were cutting edge. Why then sell to SPSS? What was in it for you? What was in it for SPSS? Was it a good move for both companies?

Yes and no. I was tired of competing for business and wanted to get back to research. Being at SPSS taught me a lot about good and bad management and good and bad development. Ironically, the development department at SPSS made a good-faith effort to carry SYSTAT forward. It was the sales department that undermined this effort by converting most of the SYSTAT campuses to SPSS after telling them that SYSTAT would not be supported in the future. I was warned by advisors before the sale that this would probably happen, so it was not a huge surprise.

What happened after SYSTAT was sold to SPSS? Why did SPSS in turn end up selling it? Through its changes in ownership, what has been your continued involvement with SYSTAT? As your intellectual offspring, how has SYSTAT changed over the years? Are you pleased with how it has grown up? Is SYSTAT still at the cutting edge of statistical computing? What do you see as its future? What has been its impact on statistical computing generally?

SPSS sold SYSTAT to an Indian development company. Cranes Software in Bangalore invested a huge amount of programming and statistical talent in the product — more than it deserved, I have to admit. The current Version 13 is far different from what I originally envisioned. It is a powerhouse that is justifiably a competitor to SPSS. Fortunately, sales are growing once again so that we expect to recoup the investment. My role at this point is mainly in sales and marketing, although I have written some new material for the program.

You ended up resigning a tenured position in psychology to devote more attention to SYSTAT. When did you decide to make this move and what finally prompted it? Was it hard to leave that position? Since then, you’ve held adjunct positions. Where have these been and how have they worked out for you? Adjunct positions usually entail lots of teaching with low pay and no benefits. Yet in your case, adjunct positions have given you all the benefits of an academic appointment without all the time-consuming duties. How did you swing this?

There are two types of adjuncts in most universities — young worker bees and older honorifics. It’s quite common for computer science departments to add adjunct professors from high-tech companies to facilitate internships and cross-fertilization. These are unpaid positions, but people like me don’t need the pay. I don’t teach courses but I advise several graduate students on their dissertations. And I get grants to support them (not me). Everybody wins.

What advice would you give to bright young faculty who see a way to turn a profit from their academic expertise — in other words, how would you advise upcoming academic entrepreneurs who find themselves in circumstances similar to those of your early years? Are there any big changes over the last 25 years on the business and academic fronts that they need to keep in mind? How would you encourage them to pursue academic entrepreneurship?

Don’t do anything without tenure. Assistant professors should concentrate on their publications, teaching, and service. When you get tenure, don’t plan to start a company unless you can’t help yourself. I believe there is no life more satisfying than that of a tenured, productive professor. In any case, plan to leave once you are successful. You can return, the way I did, but don’t expect to get the same privileges full-time faculty have. I have no regrets about giving up tenure. I am now 66, but I feel like a new assistant professor hoping to have the time to pursue all the research ideas still in my head. And it is a special pleasure to have returned to the University of Illinois at Chicago, where I started. My grad students are in their early 20s. I learn as much from them as they do from me. Also, I’m ignoring my own advice and starting another company this year. It’s called Advise Analytics and is based on a program I’ve written in my spare time over the last few years. The program is called A Second Opinion. It re-analyzes your data and tells you if the analysis you did in SAS/SPSS/SYSTAT/R/Stata/JMP meets the required assumptions.

The best ideas in the world.© 2011 SuperScholar.org