SCIENCE

THE HISTORY OF SYSTEMS SOFTWARE: WHERE DO WE GO FROM HERE?

Introduced by Andy Pepperdine on 27 November 1998

When you are sitting at your computer using your favourite word processor, have you ever wondered how much software has been written to enable you to see a letter on the screen when you hit a key? Sixty years ago there was no such thing as a computer, now we have systems occupying hundreds of megabytes of programmes on every office desk. How did this happen?
With the advent of the idea during the second World War and then the technology of a machine that stored its instructions, enabling it to do different things on different occasions, the computer was born and with it the software challenge - how to write all those instructions and get them right. The first machines loaded their memories from patch panels, like an old
telephone exchange; later, paper tape and punched cards were used, but people wanted numbers and letters to read so ‘assemblers’ were wrtitten to translate readable text into machine-readable strings of bytes. The first had to be written laboriously in the form the machine wanted, but it was a breakthrough because it could be used to write more complex assemblers, so by the end of the 40s the first commercial computers appeared.
During the 50s and 60s there was an explosion of computer languages - Fortran, Cobol. PL/I, Lisp, APL, BPL and C amongst countless others. Each found a niche. At the same time machines were getting faster and more expensive, so there was a demand for more people to use them at the same time, which led to the development of operating systems, each developed by the company selling the computer.
In the 60s computers started to be used to control processes in steel mills and chemical factories. These applications required new languages, like Coral and RTL, and systems, but then the problem arose that changing hardware to a faster model meant expensive re-writing of the software. In 1968 the US Department of Defence proposed a ‘better’ language, Ada, which helped their problems but did not tackle the industrial need to combine flexibility with re-usability.
From the 70s, applications and systems software drifted apart. Applications built on existing programmes, but they needed compilers for languages to help write the programmes and operating systems to run simultaneous applications. The number of systems decreased because there was no money in writing systems, only in selling hardware. These pressures continue, but , in parallel, microprocessors have expanded into other areas. A modern car may have two or three dozen, needing an operating system for each.
The present state can be summed up like this. Manufacturers of microprocessors want to sell their silicon chips and would prefer not to supply the software, but customers want a complete box-of-tricks off the shelf. No one can work out how to make money from the software that is needed to write the applications (compilers) and to run them (operating systems), so there will be fewer companies and people doing so, but customers will be more dependent on them for their developments. This situation will only change when there is an acceptance of the true value of software and prices begin to
match the true costs.

Andy Pepperdine