Experiences serving on the Board of Directors of Sperry Corporation, 1983-1986


I entered the world and the workings of the BIG corporation by being asked to serve on the Sperry Board—an eye-opening experience. Sperry manufactured and serviced the once famous brand of UNIVAC computers; then became a multi-industry diversified giant. I was the first and only computer scientist ever selected for the Sperry Board. Sperry was acquired in a hostile takeover by Burroughs Corporation in 1986, ending my life as a Big Company Director…

I was asked to form and chair Sperry’s Technology Advisory Board. This Board found weakness in UNIVAC’s research and its product offerings. Over my few years as a Director, UNIVAC management (near Philadelphia) made good plans to shift from selling computers (vs IBM) to selling software and services. Excellent concept, but died at birth when Burroughs bought Sperry, forming Unisys. In my view, Burroughs CEO Blumenthal, with clumsy management, got rid of the UNIVAC planning (and product line and people).

I am not particularly proud of one important and ultimately costly piece of business advice I gave to UNIVAC management (after being asked).  I gave the same advice to the Texas Instruments (TI) CEO and have regretted it ever since. It concerns the manufacturing and sale of LISP machines, a type of computer particularly well-suited for symbolic computation (vs numeric calculation). Interest in LISP, the software substratum of artificial intelligence (AI) research at the time, was high. The Japanese National Fifth Generation Project and the widely read book The Fifth Generation, by McCorduck and me, had significantly raised the profile of AI, hence LISP also. With government support, TI had developed a “LISP chip.” The question was: would it be a good business decision to build and sell a LISP personal workstation using this chip? Being asked, my advice was “yes,” but I was wrong. The “AI market” was practically non-existent. In the time it would take to build up this market, the inevitability of increase of chip capability under Moore’s Law would overtake the underpowered LISP chip.

Sperry’s UNIVAC division got caught up in this. Like TI, UNIVAC was looking for a “breakthrough” product.  They decided to use their marketing and sales capabilities to sell LISP workstations that they bought OEM from TI.  Again, my advice to them was wrong. It’s hard to sell into a non-existent market, and harder still when a sales force proves incapable of learning symbolic computation, LISP, and AI software and applications. I can’t blame them. They were brought up at the other end of computing.

Comments are closed.