What is CODIL?

 

CODIL (COntext Dependent Information Language) is an early (1960s) experimental human-computer interaction language which was originally based on first-hand experience working on a 100% manual international  management information system handling research and development correspondence for a veterinary subsidiary of the Wellcome Foundation, followed by a period of programming and systems analysis on one of the largest sales accounting batch-processing computer system of the period, involving several LEO III computers handling sales for the oil marketing company Shell-Mex & BP. At the time many computer manufacturers were planning for the first generation of computers which hopefully could support integrated management systems with user terminals.

CODIL was intended to be  the user interface for such an integrated management system, allowing humans and an "intelligent electronic clerk" to work as a team on a wide range of commercial applications and one of the key requirements was that the electronic clerk should always be able to explain to its human teammates what it was doing in terms that they could understand. The idea was to replace the conventional black box systems,  with a transparent system which was flexible to accommodate the complexities of the real world market place.

 

I suggested the idea in 1966, in effect extending the ideas  in Bush's 19XX article "As We may Think" to commercial computing.  The research was initially supported by John Pinkerton and David Caminer, the commercial data processing pioneers responsible for the world's first purpose-built commercial computer, the LEO I, but moved to Brunel University because it had been in the former Leo research division which was closed down after the merger to from ICL. The approach was incompatible with the contemporary fashionable AI paradigms and I abandoned the research in 1988  when I was suffering from post-traumatic stress disorder following a family suicide. It was recently suggested that, because it was one of the earliest attempts to  design of a transparent  user-friendly computer system the project archives could be relevant to the history of the development of computing.

 

This blog is designed to identify the key documentation and review its significance in the light of computing and artificial intelligence research in the decades since the project closed. In particular it will examine how far the original research modelled the way that the human brain stores and processes information and its relevance to artificial intelligence research.

 

The initial assessment suggests that there are good scientific reasons to restart this project, perhaps starting by producing a more powerful version of the CODIL interpreter which will run on modern computers, and then by using the software to model a number of human and artificial intelligence tasks. Because I was born  85 years ago I am now in deep retirement the project will need to continued by a younger generation of scientists. If you are interested in knowing more, or have any comments (favourable or unfavourable) about the future of the research and project archives, I will be delighted to hear from you.