- 1333 - 
least three phases. Each phase contains a development-evaluation cycle. 
The earliest systems (GCARS I and II) were strictly research systems. 
In Phase II, roughly 1970-73, the earlier versions were modified to work 
on IBM 360 systems and an extensive series of evaluations were made in 
Canada, the United States, and Europe. A totally new system of programs 
called GMAPS-GCARS was developed beginning in 1973 and incorporated 
many changes shown to be desirable by the previous work. In these new 
systems the steps of model building were separated from those corridor 
selection; the GMAPS programs (General Map Analysis Planning System) 
being responsible for the former, while a new GCARS program suite was 
responsible for the latter. 
The GMAPS-GCARS programs are designed to utilize interactive terminal 
dialogs in time sharing environments. The earlier GCARS II programs 
also operated this way, on one of the earliest interactive time-sharing 
systems (the Purdue PROCSY System), and had shown interactive computing 
to be very desirable. Lack of availability of suitable computer systems 
delayed the implementation of new interactive systems until a DEC-10 
System became available in 1972. 
GCARS System Design Goals 
The GCARS System development was guided by six design goals: 
1) The system should be machine independent; that is, it should be 
easily implemented on a variety of computers built by different 
manufacturers. 
2) The system should be economical to use. This goal was interpreted 
as modest computer core-storage requirements, and short calculation 
times. 
3) The system should provide effective and convenient methods of man- 
machine information interchanges. This goal appeared necessary in 
order to allow the engineer to apply his decision-making capabil- 
ities. 
4) The system should have sufficient flexibility to allow: 
a) suitable quantitative measures of all pertinent factors 
b) the analysis of pertinent factors alone or in varying com- 
binations. 
5) The system should have sensitivity to the factors being analyzed 
and include techniques of ranking and discriminating between the 
alternatives generated. 
6) The system should have general compatibility with existing planning 
methodology and available, more detailed, design systems in terms 
of resolution and data requirements. 
  
  
  
  
  
Obviously these design goals represent the ideal case; it was recognized 
that conflicts within and among these goals might prevent their complete 
achievement. Nevertheless they did represent, and continue to represent, 
an ultimate yardstick against which all computer-aided planning systems 
should be measured.