I just learned that Roman Smolensky passed away in 1995, at age 35. His paper on "Algebraic methods in the theory of lower bounds for Boolean circuit complexity" is a classic; it was one of the ten most important papers I read in preparing my own D.Phil. thesis.
I got to meet Roman at the Durham symposium on Boolean function complexity in 1990, which turned out to be a watershed event for the field. I was as amazed by the elegance and power of Roman's work in person as on paper. (In the photo, Roman and I are both in the back row; Roman is standing directly under the "1961" date marker, and I'm three people to the right of him in the green shirt.)
Allan Borodin (with Leonid Charny) wrote a tribute to Roman in the journal Computational Complexity in 1997. He recalls Roman as "the most gentle, friendly and unimposing person" who "radiated a sense of internal peace and warm, positive and creative energy," words that perfectly match my now twenty-year-old recollections. Having left the theoretical computer science community for public policy and law in 1994, however, I could not help but find the following passage personally chastening:
"Finally in November 1994, Marek Karpinski was able to obtain a long-term position for Roman with his group in Bonn. This was a time when jobs in theoretical computer science were getting more difficult to obtain and many individuals were trying to redirect their work (or at least dress up their work) so as to be more 'practical.' This thought never entered Roman's considerations....
"Curiosity was the main motivation for Roman's research, so neither fashion nor practical considerations of job hunting were able to shift his interests from theoretical computer science to more 'practical' things."