product, suitable for cartographers and analusts, can be 
created. Again these products are in the form of CCT or 
hardcopy. Systems supporting this processing must be more 
flexible than Pi systems and typically meet end-user 
expectations less often than P1 systems do. The  trade-off 
betueen  flexibilitu and speed (general purpose computers vs 
array processors) may be misunderstood or ignored. in favor 
of the tendency to select hardware without regard to 
software. 
The user interface starts to become important in P2 systems 
since some computer—literate analysts may like to have a 
hands-on relationship with the data. Unfortunately, this 
typically takes the form of peering over an operator's 
shoulder. 
Jj: First Level Application Systems (Al) 
Usually billed as ‘turnkey’, end-user stations, first level 
application systems provide more specialized enhancement, 
arithmetic (ratio, add, difference: principal components: 
etc) and classification capabilities. User controllable 
image raster display hardware is a sustem prerequisite. 
System output tends to consist mainly of specialized 
hardcopy and statistical tables, charts, plots and matrices. 
In Al systems, flexibility is crucial. Various combinations 
of operations mau have to be constructed for each end-user's 
needs. Manual, spatial input plays as important a role as 
does data quality and system throughput. Manipulating the 
processing programs can become more important, and more 
difficult. than manipulating the. data -— from both the 
programmer'/s and the user/analyst’s points-of-vieu. 
The user interface is absolutely critical at the Al level. 
Many computer literate analysts, eager to get hands-on 
control of data processing, are discouraged by  intimidating 
and humiliating machine rebuffs. They return to "over the 
operator's shoulders" mode. Many Al systems that are 
available either fall into disuse or require far too many 
hours of operator coaxing to achieve desired results. I 
have been told on several occasions that: "the results were 
not exactiy what we wanted, but it was too much trouble to 
Te-do the work.” In fact. I am familiar with a sustem whose 
documentation was so troublesome to read that one of the 
Chief analysis using the system was unaware that the system 
incorporated classification programs, even though these 
algorithms were of central importance to the sustem. He 
hired a programmer to write the apparently ‘missing’ 
programs. 
212 
  
TL +I AD CN 
40 ~0w.a Tm 
u I