The aim quite early on was to make Atlas a home for long-term users. As well as providing accommodation, the decision was that there should be a good library available covering the user disciplines. On a visit to MIT in 1965 by Bob Churchhouse, the Information Retrieval system there so impressed him that the decision was taken to develop a similar system for Atlas but with the information content tailored to the Atlas users' needs.
There was also a lot of interest quite early on in the life of Atlas by people interested in the humanities in various aspects of text processing. This occurred initially at Manchester and also at Edinburgh.
Consequently, Don Russell looked around to see what was available to support concordance activities and decided the best approach was to develop a new system called COCOA. He also implemented Joe Weizenbaum's SLIP system for linguistic list processing.
The Artificial Intelligence community were interested in the use of Atlas due to its power and memory size. Several Atlas Fellows were connected with this discipline. In consequence, early on it was decided to write a LISP compiler, being the only language that the community could agree on as necessary.
The Brooker-Morris Compiler Compiler being available, this was used to produce the LISP compiler. It was probably the earliest use of the Compiler Compiler at ACL.