First-Hand:A Historical Cobol Note

A Historical Cobol Note

Submitted by Robert Patrick

The Smithsonian has a Cobol exhibit located at http://americanhistory.si.edu/cobol/introduction. The curator seems to be impressed that Cobol worked so well for so long. The following note gives one possible explanation.

In the 1950s there were several manufacturers competing for the growing data processing segment of the mainframe computer business. For a while these manufacturers made conscious differences in their product offerings so their competitors would have a harder time stealing their customers.

While IBM and Remington Rand (later Univac) were the language leaders, they were followed by Burroughs, CDC, GE, Honeywell, NCR, and RCA. A Datamation wag referred to this collection of competitors as Snow White and the Seven Dwarfs.

In 1958, Fletcher Jones, Roy Nutt, and I had discussed forming a corporation to produce computer software. Later, Fletcher found that Dick Clippinger of the Datamatic Division of Minneapolis-Honeywell wanted to buy a business compiler for their new Honeywell-800 computer. Things moved rapidly and in the Spring of 1959, Computer Sciences Corporation was formed to do a natural language compiler for the H-800.

All of the early CSC employees had scientific backgrounds on IBM 7090s. We had not done a lot of business work. We didn’t get a lot of guidance from Clippinger either. We were aware of Flomatic (Hopper) and Comtran (Talmadge), but these were stylized languages.

As CSC was getting organized in Los Angeles, I did a literature search and ‘discovered’ a book called “Basic English” by C. K. Ogden. Being an engineer and not a linguist, I was not aware of the serious debate that had taken place during the 1930s to produce/standardize on a single language for international commerce. The two principal contenders were Esperanto and a simplified dialect of English. Once I found Basic English, we adopted its simplified constructs for the Honeywell compiler (called FACT, for fully automatic compiling technique). Not very far into the project I wrote a primitive FACT users manual.

When the Codasyl committee was formed, Nutt was selected by Clippinger to represent Honeywell. When Roy went East he took our FACT manual with him. From there the story is well known. The committee produced a specification, compilers were written, and programs could be moved between machines of various manufacturers with relative ease.

About ten years later, I came across a then current Cobol manual. I was pleasantly surprised to find some of the programming examples I put into the original FACT manual still intact. Cobol lasted many years as a business programming language. Some of that longevity I ascribe to the foundation of simplified English laid down by the linguists in the 1930s.