26th November 2010: a GenStat Anniversary
It was 25 years ago today that Roger Payne, VSNi’s Chief Scientific and Technical Officer, took over as leader of the Genstat team at its then home, Rothamsted. Back in 1985, Genstat (note – no capital S in those days in the middle of the word!) was designed for batch use. You entered a file of commands and, later in the day, you hoped to get back a file of output, often in the form of a wide sheaf of printed paper. This was all that could then be supported effectively by Rothamsted’s ICL 4.70 computer, and the similar computers elsewhere. However, computing technology has improved dramatically since then – and Genstat has improved with it!
Roger was aware that interactive computing would soon become feasible, and knew that the quality of statistical analysis would be greatly improved by putting statisticians into closer touch with their data. So in 1983 he had convinced John Nelder, the Head of Statistics at Rothamsted and originator of Genstat, to let him become the architect of the new radically revised version that became Genstat 5. Nelder retired at the end of 1984. John Gower replaced him as Rothamsted Head of Statistics in 1985, and confirmed Roger’s leadership of Genstat on 26th November.
The first full release of Genstat 5 took place in 1987, and it soon became very popular on the new VAX 750 computers at Rothamsted and its sister AFRC (Agricultural and Food Research Council) Institutes. In fact Genstat 5 became so popular that Roger was summoned before the Committee of AFRC Computer Managers to be told off for providing the statisticians with this tool for overloading their computers! Fortunately, some of the Committee were aware that the real purpose of the computers was to support research – and the key role that statisticians played in this.
Genstat 5 had a redesigned, simpler and consistent, syntax which was very much easier to learn and remember. Another important innovation was the procedure structure, which has enabled us to implement, investigate and then distribute new ideas in a very effective way. Procedures have an identical syntax for their use to that of the standard GenStat commands. They can be stored in libraries, and accessed automatically as required during an analysis. The official GenStat Procedure Library contains procedures from the users at large as well as from the GenStat team, and an Editorial Committee checks that the accepted procedures are useful, reliable and accompanied by clear documentation.
The GenStat that we all now use is still based on the Genstat 5 language, but the “5” was dropped when the “S” in the middle of “GenStat” became a capital letter. So you can still write Genstat 5 programs and procedures but, for most users, the programs are written for them automatically by the menus in GenStat for Windows. As a result, GenStat is no longer a tool just for statisticians, but has become the statistical system of choice for researchers in biology and the life sciences all over the world.