Friday, November 17, 2017

Calling computers names in Swedish

I wrote a correction to the very nice article "Carl-Gustaf Rossby: Theorist, institution builder, bon vivant". Rossby and Germund Dahlquist developed a numerical weather model for the BESK computer in 1953-54 and the article made the erroneous claim that "Rossby pursued numerical weather prediction in Sweden in an era in which there was no Swedish word for digital computer". I listed the five Swedish terms for digital computer that were in use at the time. My purpose was not to nitpick, but to make the point that Stockholm in 1953 might well have been uniquely fertile soil for developing novel applications for computational science because of BESK and Dahlquist.

Through luck and skill, Sweden managed to stay out of WWII and had a booming, technologically advanced, industrial economy in the post-war period. In 1947 the Swedish government initiated attempts to buy a computer from the US. A Swedish delegation of five young engineers got remarkable access to the principals behind the early computers in the US. However, ultimately the Swedish government was denied an export license for a US computer and BESK was rapidly designed and built in Stockholm (at my alma mater KTH) using know-how gained during the visit to the US. BESK became operational in September 1953 and was briefly the fastest computer in the world, capable of 16,000 additions per second. It was also the first computer to (partially) use semiconductor electronics (400 germanium diodes).

The Swedish government's primary need for a computer was to support their nuclear-weapon program, which could explain the denied US export license, and to decrypt intercepted Russian radio communication. The ability to decrypt all electronic communication between Berlin and the German embassy in Stockholm had been instrumental in keeping Sweden out of WWII. Continued access to encrypted Russian diplomatic and military radio communications to help navigate the cold-war political environment was a strong motivation.

In stark contrast to this political and military pragmatism was Dahlquist's scientific idealism. His goal was to disperse rigorous computation through all fields of science and technology. Together with co-author Åke Björck he wrote the world's first textbook on numerical methods, with an emphasis on numerical accuracy and stability. Dahlquist's collaboration with Rossby on numerical weather modeling was one of the main achievements of this quest. Their work enabled 24-hour national weather forecast in Sweden from September 1954, 4-5 years earlier than in any other part of the world. Rossby's decision to return to Sweden from the US in 1953 to pursue weather modeling on computers was a wise one, not the folly of trying to do state-of-the-art computing in a place that did not yet have a word for computer.

No comments:

Post a Comment