John Backus and the IBM labs version of the war for talent

Media reported earlier the week the death of John Backus, ex-IBM researcher, who in the late 1950s, developed the first high-level programming language, Fortran, and thus transformed what was previously a tedious, largely mechanical task into a challenging intellectual pursuit. Despite fame and honours bestowed on him, Backus remained a level-headed man, who claimed that he invented Fortran out of “being lazy.”

Backus is one of several distinguished IBM researchers, who transformed not only the science of computing but also basic physics (with four researchers sharing two Nobel Prises and mathematics. Its historical scientific achievements are on a par with better known Bell Labs

By pure coincidence, I attended a week ago in Brussels, a lecture given by another illustrious IBM lab alumnus, Benoit Mandelbrot, who invented fractal geometry, while at IBM. Professor Mandelbrot, who now is in his 80s but remains professionally active and as feisty as ever, digressed for few minutes from fractals and their applications in finance and other domains, to discuss the evolution of basic research. Not surprisingly, he expressed concern about the loss of quality. His point was that as government withdraws from funding of research, the latter becomes more academic. By that he meant not a choice of topics detached from practical realities or conceptual relevance but the selection of researchers, based on criteria of academic credentials. According to him, such criteria are too narrow. IBM Labs, at their day, made sure that researchers they selected had various qualities, including practical bent. He specifically mentioned one of Nobel Prizes in physics, co-inventor of scanning tunnelling microscope, who would probably never been hired by a top university.

From what I read, it looks like Backus was a similar case. His academic record was solid but not outstanding. At the time he was hired by IBM, he had a MA, rather a Ph.D in mathematics (from Columbia, where he also studied medicine and radio engineering) and was chosen on an intuition about his future rather than the evidence of his grades and research papers.

As “war for talent” rages and new generation corporate giants, HP, Microsoft or Google, set up their research units, lessons of IBM labs approach, as presented by Mandelbrot, deserve careful considerations. So far, new labs appeared to have made little contribution. One reason for this may be that they are missing the vital tier of researchers, between developers and academics