On the whole, with regard to accuracy and completeness, the base
performance of individual interpreters in the groupof experienced PI's
was lower than anticipated. However, without the usual aids provided
interpreters, and considering performance across all subject-matter
specialties, average performance of 75% accuracy and L6% completeness
for component scoring and 50% accuracy and 39% completeness for component-
modifier scoring provides & substantial starting point for improving in-
terpreter output. Certainly, under more normal operating conditions
(e.g.: availability of reference aids, specialized target knowledge),
; quantity and quality of output could be expected to be appreciably
higher.
i Ihe findings of this study indicate that both inter- and intra- UN
| individual differences are & significant factor in imagery interpreta- I |
tion output. The concept of a good 'generalist' interpreter is most |
likely a fallacy. Output of personnel assigned on the basis of best
abilities was found to be better both in quality and quantity than the
output of randomly assigned interpreters.
Accuracy and completeness of information are markedly improved by
the pooling of independent reports. Accuracy approaches 100% with an
accompaniment of some increase in completeness by the simple expedient
of pooling reports of a small number of highly qualified individuals.
PERFORMANCE OF PI TRAINEES USING TACTICAL PHOTOGRAPHSSL
Widely separated targets may be even more difficult to interpret
than photographs in which the various target-images tend to be func-
tionally and positionally related to one another. The task of examining
a large number of photographs in a short period of time in order to
locate and identify small tactical targets may be one of the most de- |
manding tasks that can be imposed on a photointerpreter. For tactical | EI |
photographs, the average completeness of information extraction is Hl
likely to be even lower than that for strategic photographs. It is re-
latively easier to miss an isolated target concealed in the surrounding
countryside than it is to miss a known component of an integrated stra-
tegic complex. The average accuracy of pnotointerpreters in interpre-
ting tactical photographs is also likely to be lower than for strategic
photographs, since in tactical photographs there are many more natural i |
and man-made positions and objects which look as if they could have LE |
military significance.
Vi Uu
£/ Appreciation is expressed to Lt Col. Eugene H. Manley, Chief, Air m
Reconnaissance Branch, U. S. Army Intelligence School, for assistance | |
in the formulation of the problems and development of test materials. '
= 17 +
Archives 6
