PROLEGOMENA TO A THEORY OF ORGANIZATION
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP81-01043R001800180002-7
Release Decision:
RIPPUB
Original Classification:
U
Document Page Count:
128
Document Creation Date:
December 23, 2016
Document Release Date:
March 15, 2013
Sequence Number:
2
Case Number:
Publication Date:
December 10, 1951
Content Type:
MEMO
File:
Attachment | Size |
---|---|
![]() | 8.23 MB |
Body:
Declassified in Part -Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
STAT
Declassified in Part -Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Oskar i~i:or;enstern
R~-734
ASTIA Document Number ATI 210734
10 December 1951
Declassified in Part -Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
125
Declassified in Part -Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
U. S. AIR FORCE
PROJECT RAND
RESEARCH MEMORANDUM
This-is-a working paper. It may he expanded, modified, or with-
drawn at any time. The views, conclusions, and recommendations
policies_ of -the United States Air Force.
expressed herein do not necessarily reflect the official views or
~~ R~ n Dom.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
PROLEaO!KENA TO A THEORY OF ORGANIZATION
Oskar Morgenstern
RM-73~+
i-
Summary: This is the first two parts of the proposed
manuscript. The ultimate form of a theory of organiza-
tion will undoubtedly be highly mathematical, but the
ground work must be laid in careful description, which,
since it precedes the theory, is qualitative and
approximate 1n nature.
A framework is discussed within which to make these
initial descriptive analyses of organizations. The
framework deals solely with the centrally directed
organizations. This is analyzed in terms of inner and
outer activities, the delegation and arrangement of
competences (a neutral word introduced to avoid such
terms as "organs"), and the systems of signaling.
Qame--theoretic ideas are drawn upon considerably. There
is also some discussion within the proposed framework
of the concepts of learning, input--output, costs and
size of organizations.
- - .1:
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM~73~+
-ii-
PROLEGOMENA TO A TREORY OF ORGANIZATION
-=
Page
Part I
Introduction. ..
Organization and the Economic Problem
1
1
Part II Prtlimina Description of a Social Organization
w. en ral rec on.
15
1. Introduction .
15
2. Inner and Outer Activities: Aims of
Orga~iizations .
l7
3. The Arrangement of Competences.
2~F
~#. Signaling, Information, and Memory.
~1
5. Mistakes, Errors, and Recovery.
62
6. Conttols within the Organization.
71
7. Degrees of Organization
75
8. Learning. .
78
g. Input-Output and Costs of Operations.
88
10. Size of Organizations
95
11. Expansion and Growth.
? . - ,~ , -
116
Jw .-. ~.
Forthcomingi~
12. Composite_~Organizations of Part V
Part III Conveyor S~retems as Illustrations of Line-flow
rgan za ons
~~Part IV Logistics Theory
Part V Composite Organizations
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~
1-
PROLEC}OMgNA TO A THEORY OFVORaANIZATION
Oskar Morgenstern
PART I -- INTRODUCTION
Organization and the Economic Problem
Lately a great deal of interest has been aroused in prob-
lems of military logistics. This is the science or scienee-
to~e that announces the principles,~and helps to develop the
methods, for determining the needs in personnel and materiel
in order to provide for a campaign, a whole xar, or merely for
the equipment of a single ship or company, For each of these
purposes men and materiel have to be assembled in precise
quantities and places; sequences of operations have to be carried
out in certain order, and there exist various ways in which
the amts. can be accomplished. Among the different possibilities
there may be an optimum procedure. There are also costs
involved. Furthermore, the requirements, once they have been
establAshsd, have to be spelled out ultimately in greatest
detail down to the last bolts and nuts and spare parts that must
fit all the utensils used and are subject to spoilage and wear.
Arrangements must be made for continuous support of the opera-
tions, if necessary.
These tasks are very similar to, if not identical with,
those occurring in business Khere usually continuous production
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~+
-2-
and sales are to be performed. The dissimilarity lies at least
in the vastly greater scope of military operations. This may
indeed produce a qualitative difference and require new methods
for description and analysis. Besides, there exists no theory
whatever describing the infinitely simpler,_problems of business
logistics.
Whether it be a strictly economic or a military logistics
phenomenon, there is for both the common need to organize men
and to assemble material in order to carry out operations with
them in some determined order. There are also objective, tech-
nical factors involved which impose restraints upon these opera-
tions. Among these may be some unalterable rules of a purely
social nature; they shall be counted among the technical fac-
tors when they are not standards of behavior (as understood in
the Theory of names) that arise from the acts of man within a
given framework. The practical infinity of variations of orga-
nizational set-ups which 1s encountered in the social world
requires that particular stre$s be placed upon the descriptive
approach to the problem, before formulating a theory of organi-
sation. Economics, logistics, sociology xould all be greatly
heYped if the simplest principles of organization were well
understood and could be consciously applied to concrete tasks.
The preparatory descriptive work would be facilitated if a firm
conceptual basis could be used. Because of the close_ inter-
~~riotior~.'anct` ?I~t~i~satiq~~ of deso~'1~~on and" the formation at ai
proper conceptual background, these two can progress only gradually
and in halting steps.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
There exists, of course, a field oP common--sense observa-
tion (quite apart from a vast literature and the accounts in
history) which serves as an adequate preliminary basis for first
abstractions. These are exceedingly difficult to make, as must
be clear to anyone who has ever tried merely to state what goes
on in a city as New York, in a university or a large corpora-
tion, not to mention the national capital. Long detailed accounts
are possible in each case, as complex as the situations they
describe (and often done beat by novelists), or sociological des-
criptions can be rendered. These are often given in terms of
complicated, mostly classificatory terminologies that have not
contributed decisively to the evolvement of a theory.
In looking for a theory of a field for which none exists,
or which is to replace an inadequate one, it is necessary to
form some ideas about the properties of the theory tome. The
main feature is that a theory of organization must be quantitative
since any organization (as here understood of Part II) allows
for variations in its set-up and also in its results. These m$y
be, for the moat, conceived of as input-output relations, although
they are _mo_r_e than that. In the former -case,. the similarity to
;so~~. P e~onouio brganiza~ione is apparent. The="auc=!
cess or failure of an organization, its superiority or inferiority
to others of the same type, the identical purpose, or conflict
with those xho attempt something similar, calls for description
of its various states by means of numbers. Yet these numerical
expreeaions need not be of a simple nature, readily accessible
intuitively. Complicated arrangements for the ordering of the
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~F
--~-
different stated of one and the game organization are theore~:~
tically possible. Recalling the difficulties of assigning
numbers for economic set-ups and their different successes and
failures (ef. the discussions about the measurement of utility),
it seems highly improbable that a quantitative theory of orga-
n
nization xhere the difficulties are vastly greater is in easy
reach.
However, even if the precise nature of a quantitative theciry
cannot be imagined, itssqualitative approximation should never-
theless aim at an operational significance of whatever insights
can be gained, and at the expression of success and failure of
an organization, though exact measurements are entirely out of our
reach.
Work in the field of the theory of organization will for
a long time have to be essentially phenomenological. This should
be clearly understood because it will influence greatly the
nature of the empirical xork that has to be done on such a large
scale in this field. But even the phenomenological theory will
ultimately have to be mathematical; but it is not to be foreseen
before the theory actually is established, what its mathematical
character will be like. An idea of this can be formed by look-
ing at the mathematical structure of the theoryy- of games, which
deals with many aspects of organizations and their properties.
The interaction of individual organizations will, as a rule,
correspond to games of strategy. But there is a need to inves-
tigate phenomena within individual, centrally controlled orga-
nizations that are, as yet, not dtalt with explicitly by the
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~+
-~-
theory of games. In other respects that theory has gone fur-
ther; e.g., in showing how and when persons-or equivalent
individual organizations-may sat in cooperation with each other
and to indicate when frid why the bonds tying them together may
sometimes be stronger or weaker. It also shown the extraordinary
delicacy and complication of the set-ape in which society may
arrange its affairs. Examples will clarify this point later
(cf. Part V).
The study of "organization" in economics-apart from the
theory of games-has developed in three directions: (a) the
discussion of economic "systems" or "orders," meaning the eco-
nomic organization of society through "free enterprise," "social-
ism," "corporationiam," etc; (b) the investigation of the opera-
s
tion of individual firms or conglomerations of firms imbedded
in systems arf the type (a); and (c) central authorities and
"institutions," such as Central Banks, Treasuries, etc.
Discussions of organizations of the type (a,)~ have in the
absence of guidance by rigorous theory tended to be historical,
mixed with philosophical interpretation and evaluation. Because
of the great difP~.culty in getting hold of the basic facts in
this field and the lack of methods useful in the descriptions,
the type and amount of information has not greatly changed.
Analysis according to (b) attempt to show: (1~) what happens to
an organization (e.g., a firm) as a whole in various situations
of particular markets under different behavior visa vis other
economic forces and unite, and (2) what the input-output rela-
tiona are for such an organization if variations in inputs or
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~
-.6-
outputs are made, and the efficiency of the set-up is measured
thereby.l But neither given any indications for determining
the inner workings of an economic organization as a firm; e.g.,
in the sense of showing the distribution of authority, the memory
of the institution, the recruiting of the leaders, etc. These
obviously all influence the outcome to some extent. Economics
thus far has therefore only discussed some of the alleged sum-
mart' properties of these organizations and of the factors deter-
mining their success, and there is a decided interest in study-
ing the wider range of questions. Since economics, imperfect as
it is, eonatitutes the most advanced of the social sciences, the
contribution it can make toward formulating the characteristics
of a future theory of organization should not be overlooked even
though it may deal only with a very limited aspect of organization.
Consider, for example, the minimizing of costs as strictl~r
equivalent to the maximizing of profit, or in a more restricted
fields of input and output in physical units. In such a state-
ment at least one hidden assumption is made about organization.
It is that a rigorously iven organization is to be considered
only when it might be the case that the same input is handled
in a variety of organizational forms or set-~zps (operations)
without affecting the real (physical) costs or inputs. Such
variations would be due to or equivalent to, different organi-
zational arrangements. The sequence of work with the same tools
the rhythm of work, the distribution of the flows of raw
? 1 Assuming, for the time being, that (2) is independent of (1),
which is the case only undtr severely limiting conditions not
generally made explicit.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73.4
7-
materials, etc., may all come under this heading. Work gangs
can be put together in many ways, each producing a different
,~-raat~:~:-~-i-#~~t~itny chan~e...~.r},.~npu~.-...It_ is interesting to note
,tithat ~ Adam'r~Smith~! s~~~i7~,1~}st `ration ~.a ? ttie advantages of the
:c: _ s.~ 4tsr_ t? t
division of labor-falls into this category. There pins are pro-
duced in two different manners: first, each worker produces an
entire pin; then each performs only some function which formerly
had to be performed by every one of them. Result is that the ,
output rises aignit'icantly without increasing input of labor or
oosta. For example, the same output can be achieved with less
coat (i.e., in less time) or a larger output with merely more
input of raw material.
The illustration from Adam Smith oan be viewed as a par-
ticular case of the more general problem of the (optimum) assign-
ment of personnel. There the question is asked what the best
arrangement is that can be designed in an organization for per-
sons who have different skills ?a~hd_=tasks, and require different
pay. A purely marginal-cost approach is inadequate and does not
,,
provide the answers.
A theory of the firm that does not allow for these occur-
rences is too xystricted and misses one of the fundamental fea-
tures of the problem it is supposed to explain. It is clear
that there is less determinateness than generally assumed in the
solution of the problem as given by the usual formulae.
1'~ If rearrangementls ~ of this type are considered-and how could
they be neglected=y-the notion of a "fixed factor," of "overhead"
costs, etc., may~ha~e to be revised, even for completely static
conditions. -
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-T3~+
-8-
? That "organization"---x~ot further speclfieally defined-
might be a "factor of production" ?i.e., have something to do
with output, net return, etc.) was recognized by some economists
and stated, e.g. (in his usual, rather vague form) by A. Marshall.
But there is no evidence that thin "factor" has been explicitly
incorporated into current economic static theory. If that had
been attempted, it is safe to say some profound changes would
have occurred:
At present it may seem appropriate for an economist to col-
lect some thoughts and observations about organizations, fully
realizing that e~ren a widening of the economist~~s interest on
the nature of organization is far from leading to a theory in
this field. But there are some essentially economic phenomena
which are primarily of an organizational character and apparently
have not been described. They shed some light on basic problems
and are intimately connected with logistics. The empirical
basis ~;s of a very ordinary kind.
In all centrally controlled organizations-economic and
otherwise with which we shall deal primarily--psychological
factors play a large role. There are the well-known phenomena
of loyalty, the "espirit de corp8," corruption, the fact that
some leaders of organizations seem to be able to obtain a greater
effect than others_ that are set up similarly. In some
I I.e., under the present basic set--up of economics; however, it
is a question whether there is an advantage in making it more
realistic in that sense and direction, when basic ideas such as
the notion that in economics one is dealing with maximum prob-
lems are at fault. The theory of games Ss far better suited
to describe the organizational possibilities; indications will
be given in what follows.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM.-73~+
_g_
organizations the morale is high, in others low, or there are
such fluctuations even within one and the same. If pressed?
how to desoribe this "morale," it is difficult to give anything
but vague and circular anawer8, although there is no denying
that such conditions exist and have to be taken into account in
the operations, however poorly they may be described.
These fields would be fascinating to explore, and any pro-
gress would have important practical consequences. Bttt I
believe that many phenomena now considered to be psychological
can be reduced to Structural properties of the organizations
and are then best discussed in these simpler terms. Others,
where this reduction does not succeed, will have to be left out
of account at the present stage.
There are, furthermore, processes within a single social
organization that have some psychological appearance but are
at the same time of a broader nature. There is, for example,
the rivalry among members, their striving for influence, even
domination, the chicanery with which superiors often treat their
subordinates, the ganging up of some against executives who are
disliked; there is the bargaining among members so that one
favor granted will, in due course, be repaid by another favor,
etc. These are noteonly real, but exceedingly important phe-
nomena, and it would be desirable to understand them well. It
is clear that they are not revealed by organization charts, ao~t
even deeper descriptions of the structures of organizations.
It may, indeed, in contrast to many psychological aspects, be
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
10-
impossible to express them in a formalistic manner. The interes-
ting thing is that they occur even within single organizations;
i.e., such that have some clearly defined central authority. It
is, of course, not surprising; that bargaining appears when an
organization is a composite of several single organizations. In
the latter case we are confronted xith coalitions such ae analyzed
in the theory of games. That such phenomena arise even within
single organizations, thus giving them the appearance of compo-
site ones (cf. Part II, Sect. 72) and obscuring the distinction
itself, is exceedingly interesting but should not guide the first
approach which ought to be limited to the study of the structural
aspects first. It is clear, however, that some formalistic setups--
as e~cpressed in part by organization charts--favor rivalries among
the various groups or offices, while another setup may give rise
to different psychological attitudes. Too little is knoxn about
this at present to allox us to determine whether descriptions in
terms of one of these categories should replace descriptions in
other terms, etc.
In discussing a field for which a theory is desired, it is
necessary, as already mentioned, to form some preliminary ideas
about the properties of the as yet--not-existent theory. This is
not easy to do, but certain rules, abstracted from the experience
elsewhere, can be observed. The main observation is that a
desirable theory xould undoubtedly have to be quantitative in
character and necessarily highly mathematical, and comprise the
great variety of forms in which the object appears in reality.
Before such a goal is even remotely approached, it is necessary
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RP~73~4
11-
? to establish a language, reasonably accurate yet rich enough to
provide the possibility of empirical description that must
preoede the formulation of any theory. Ttie mathematical character
of the future theory may be undiscernible at present-before
very much more information about the structure and the behavior
of organizations is available. It may very xell happen, as stated
in the Theor~r of Games-and Economic Behavior3 (xhich after' all~is
a theory of the interaction of social structures) that fundamen-
tally new mathematical discoveries comparable to the creation of
calculus may have to be made in this field. IIntil such an event
occurs, combinational procedures will probably play a mayor role
as in the Theory of Games.
The axiomatic method xhich has been of such importance over
p er ___
the last fifty years may appear to many a auita.ble tool for
establishment of the first theoretical step$ in the field of
organization. Indeed, some efforts have already been made in this
direetion.2 however, the axiomatic procedure will probably not
be very profitable at this puncture. Axiomatization of an empi-
rical field is only possible after a great deal of descriptive
work has been done. in a form that is rigorous enough to provide
the ground xork for the exact treatment. This has been the
experience in other applied sciences xhich have started with often
?v~gue _ deporptions _t~tfa~re ~~~reeehing tae nwrA exalted position ~o!
John von Newnafin and Oskar?Morgenstern, Theory of Games and`
? Economic Behavior, Princeton, Princeton n vers y real, ~g~~+,
? ~ J. Rrusks,ll and A. Newell
A Model for Or anization Thear
,
. The Rand Corporation Researc emoran um , a e .-'~~}-.SG,
Also RM-blg, For~culating Precise Concepts in Organization Theory,
dated 6-1-51.-
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
x-73 ~+
12-
axiornatization. In fact, the description of the operations of
organizations may in turn have to be preceded by classifications
of types of organizations because there are so many of them that
must be kept apart. It is likely that the sequence will be of
this kind, although surprises are always possible. The main
point is for the reader to contemplate the incredible complexity
of organizational fo~ns, no matter where they are encountered,
whether in the social or biological field.
Bacperimenta with the establishment and operation of social
organizations are possible--and in a rough, uncontrolled form
are going on all the time. But it will not be easy to design
the experimental co::ditions and to choose the most promising ones.
Therefore, the directly available observational evidence should
? be used in the interim period until the experimental state has
been developed. Among that evidence is a great deal of historical
material xhich should be collected and arranged with the aid of
whatever notions about the theory of organization we are able to
form at p-resent. The history of organizations is, of course, the
first task. The suggestion may seem ludicrous, since virtually
all history is one of organizations! But a far more narrox
point of view is, of course, intended. For example, it would be
important to discover the relation between the communication sys-
terra and the size of organizations (Cf. II/10}. The Persian
empire was geographically much larger than previous empires. This
groxth was probably achieved by the establishment of an entirely
reek and vast system of roads'and relay Stations which enabled the
central authority to keep in close enough touch with the Satraps,
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
xM-73~
13-
'and thereby evuld assure that the orders emanating from the
capital would be executed, etc. This system of communications,
information, and control was a novum; an understanding of its
evolution :ahd` ;li~ttat~.ons, depending on technological knoxledge
and progress, is as valuable as ever. We would learn lessons
directly applicable to the conditions under which contemporary
organizations operate. There are numerous other asprets of
organizations, both structural and operational, that could be
dealt with in the same historical manner.
Finally, there is still another largely untapped field:
In the administrative experience of bureaucracy of many nations
and in the evolution of its machinery, an enormous wealth of
information is at our disposs.l. The development of new adminis-
trative practices in the Byzantine empire, their transmis$ion to
Spain and later to Austria, as well as the great reforms in the
Middle Ages in Burgundy, all offer most interesting comparative
examples. Bureaucracy has by necessity produced Administrative
I,aw which in many countries has been codified. These codices
contain numerous rules, regulations, etc., which are, at present,
put in legal terms but could be transfozrmed into adequate non-
normative descriptions. It appears to me very probable that here
is a source-material that, iP reviewed in a modern manner, may
yield a surprising amount of information invaluable for anyone
" i
d i
t
t
b
t
ht
t
i
ti
eres
n
e
n an a
s
rac
approac
u organ
za
on. This is true,
especially if these experiences and formalizations can be compared
with the organizational principles that are becoming available in
the biological and, more specifically, the neurological field.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
From ~hrre it is a direct step to the domain of electronic corn--
puting machines and the even Hexer one of the general theory of
automata.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~
15-
PART II
PRBS,IMINARY DESCRIPTION OF A SOCIAL ORGANIZATION
WITH C$NTRAL DIRECTION
1. Introduction
An organization produces an order out of elements that in
respect to this order are in a lower form of structural rela-
tionship to each other. The organization maintains this order
over a period of time. This is true of the biological as well
as of the social field. It would, therefore, advance the under-
standing of social organization if parallels 'or analogies could
be drawn. This will not be possible unless some concepts are
available that allow a reasonably accurate, though still quali-
? tative, description of social organization. A quantitative des-
cription would obviously be superior but must be preceded by
formation of basic qualitative notions. We shall, nevertheless,
show an occasional particularly striking similarity between the
two fields.
The simplest social organization is a single Individual by
himself. He may be equipped with physical means to carry on his
affairs, but he neither takes orders, delegates authority, passes
on ir~f'brmation, etc., within his oxn organization. This indi-?
vidual has problems of organization to solve: He must determine
-?~ the optimum arrangement of his physical means and find the opts--
mum strategy to use in his contacts with other organizations,
if such contacts exist. The other organizations may be as simple
as his own or of arbitrarily high complexity. If no contacts
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~+
16-
exist, he is only faced by nature and his is then a special case
of the Robinson Crusoe economy.~l He may have any number of auto-
matic devices, no matter how complicated. He may be incompletexy
informed about his own state.
When at least two individuals, with or without additional
physical means, are involved, then there are two principal types
of organization possible: either completely anta~nistic (corres-
ponding to a zero-mum two-person game) or cooperative (corres-
ponding to a nonzero--sum 2~erson game which is equivalent to
a zero-cum three~erson game). The first type shows that even
conflict is to a high extent organized; the existence of organi-
zation is obvious for cooperation, and in those cases where the
zero~um restriction disappears. However, it will be seen that
? this dichotomy is far from simple.
We concern ouhselves primarily with cooperative social orga-
nizations with two or more persons, the organization equipped
with physical means which are required to carry out Sts purposes.
We assume that a central authority is established by whatever
means which directs the organization; i.e., lays out the various
organs, decides about the purpose or aim of the institution, and
initiates whatever operations are believed to be necessary to
achieve the aim. This central authority may be vested in a general
assembly, the owner of a firm, a board of trustees, etc. It may,
therefore, be democratic, oligarchic, etc.; this would be ~imma--
terial to the formalistic side of organizations as here investigated.
~' Special, because the Robinson Crusoe may have in his organiza--
tion individuals whom he has to treat as such, which imposes
new conditions that will become apparent below.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73 ~+
16-
exist, he is only faced by nature and his is then a special case
of the Robinson Crusoe economy. He may have any number of auto-
matic devices, no matter how complicated. He may be incompletexy
informed about his own state.
When at least two individuals, with or without additional
physical means, are involved, then there are two principal types
of.organization possible: either completely antagonistic {corres-
ponding to a zero-mum two-person game) or cooperative (corres-
ponding to a nonzero-eum 2~erson game which is equivalent to
a zero-sum three-person game). The first type shows that even
conflict is to a high extent organized; the existence of organi-
zation is obvious for cooperation, and in those cases where the
zero~um restriction disappears. However, it will be seen that
? this dichotomy is far from simple.
We concern ourselves primarily with cooperative social orga-
nizations with two or more persons, the organization equipped
with physical means which are required to carry out its purposes.
We assume that a central authority is established by whatever
means which directs the organization; i.e., lays out the various
organs, decides about the purpose or aim of the institution, and
initiates whatever operations are believed to be necessary to
achieve the aim. This central authority may be vested in a general
assembly, the owner of a firm, a board of trustees, etc. It may,
therefore, be democratic, oligarchic, ete.; this would be imma-.
terial to the formalistic side of organizations as here investigated.
? 1 Special, because the Robinson Crusoe may have in his organiza-
tion individuals Nhom he has to treat as such, which imposes
new conditions that gill become apparent below.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~
1?-
A "central direction" of the organization is possible only
if the human individuals which belong to it,cooperate to some
extent; i.e., e.ccept the orders that are given in the name of
the organization, cagy them out with a certain degree of pro-
bability, etc. In this sense there is cooperation even in the
most draconic organizational; breaking points exist for every
setup even if we cannot indicate them now. The cooperation
expresses itself in an identification-however rudimentary-of
the individual with the organization.? This characterizes the
"morale." We shall not deal with these aspects, though they are
very important, especially from a psychological standpoint. But
it is better to reduce difficulties to the minimum, and this is
one that can be avoided now.
2. Inner and Outer Activities; Aims of Organizations
In the folloxing we describe certain concepts and phenomena
which together give a possible characterization of the main
features of an organization, but by no means the only one.
?:?: -- ?~+ ~ An , organization engagers in an inr~r. ~d ~.auter anti"pity:
~~ ~ V Either ~a4~~rity cot~tsie~te of the eet of ilk. possible arrarsgisiaents
of individual acts that an organization can undertake, given its
resources, i.e., its physical equipment (e.g., capital) and its
1 Probably there never was one more draconic than the Mongolian
Ax~y organization. There varying numbers of soldiers would be
executed if, in a,--battle, one of their leaders was lost by
enemy action, as a punishment for their failure to prevent his
death. The higher the officer lost, the more of his own men
would have to die. The fact that this practice weakened the
numerical strength of the combat unit was apparently believed
to be compensated for by the great fighting spirit it produced.
This astounding principle of organization worked so well that
the Mongolian armies conquered almost the entire known world
and held it down for decades.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
milieu (e.g., rules of the game, legal frameworlc, etc.). Each
individual arrangement of acts is called an operation. The
separation into two classes of activities is a consequence of
the fact that every organization has an inside and an outside.
In other words, an organization has clear boundaries which set
it apart from its physical and social milieu; it can be defined
in space and time. No matter how scattered in both respects the
activities of an organization are, there must be a concrete
physical substratum.`1
A consequence of this is, for example, that an individual
is either inside or outside of a given organization. It will be
seen below that he can, however, in a particular sense and situa-
tion, be both inside and outside, unsound as this statement may
appear to be at present.
Within the framework set by the (possible) inner and outer
activities, particular choices are made by the organization which
acts through representatives in a binding manner that gill have
to be described.
The sequence of moves by the organization as determined by
the choice of a strategy constitutes the outer operation. The
particular pure or mixed strategy is taken from the domain of
all available strategies, i.e., from the outer activity. Availa-
bility of a strategy means that the state of the organization
~; -. r
~zereiBes~B limiting influence upon the choice of strategies;
x All this agrees, of course, xith the classical notion of the
per8ona iuridica. I believe that a great deal can be learned
or organ za onal theory from a study of administrative law,
especially xhere it is built on Roman Law and is correspond-
ingly codified.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~+
1~9-=
~ e.g., a small corporation may not be able to engage in as exten-
sive an advertising campaign as a richer organization. Therefore,
the technically better strategy is not available.
The inner operation is the sequence of acts (choices) of
the parts of the organization (to be called "competences" of
Sect. 3) involving the use of materials, resources, and machines
taken from all possible operations which make up the inner acti-
vity. The domain of all possible acts is restricted by physical
data, e.g., technological factors, physiological, physical charac-
teristics of humans and animals, legal provisions, etc., which
together are at the disposal of the organization.
Thus we see: An operation consists of acts, which are
choices. An activity is the set of all possible operations from
which a particular operation is chosen. The choices of the inner
and outer operations are not independent of each other.
Just as it is necessary to choose the optimal strategy from
the available strategies=-i.e., to optimalize the outer operation-
it is necessary to ~timalize the inner operation. Their inter-
dependence may assune many varying degrees. The inner operation '
of a preceding period may further restrict the available strate-
gies; for example, a certain crystallization of the physical
equipment of the organization may have taken place which may have
to be carried over into the next time phase. But it is best to
neglect this complication now; besides, it does not come into
play ittth all organizations. An inner operation is optimal in
respect to a given (presumably optimal) strategy or outer opera-
tion, if, under consideration of all constraints, no other inner
operation will make the choice of the outer operation possible.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~+
-~0-
Hors the optimal inner operation is to be identified and how
it is in concreto determined by the optimal outer operation
(which in turn may depend on the preceding inner operation,
? irrespective of whether it was optimal or not) is, of course, a
deep problem. It can be approached only when the description of
an organization has been accomplished.
A weaker requirement than optimality for inner, and outer
operations is their feasibility. For inner operations this means
that they have to be restricted to the technically, physically,
and legally possible arrangements and combinations that the
organization may undertake of all its component parts and the
acts it may provide for these internally. The restriction of ~
the strategies (for the outer activity) was already mentioned.
? A study of organization has, therefore, to consider the
two fields of the inner and the outer activities. We shall be
c oneerned mainly with the first, wYi3ch comprises "organization"
in the narrower sense. Social organizations of a higher type in
which several individual organizations interact establish their
relations to each other, determine payments among themselves, and
cooperate or fight (as the case may be), are best described by
the theory of games of strategy. We are here a oncerned with
questions of organization that mostly precede that wider field
of interaction among individual. organizations. But it would be
impossible to avoid reference to it, because individual organi-
zations are set up precisely in order to deal with others, or at
? least with nature. Also, each organization that requires services
of individuals (e.-g., as employees, stockholders, soldiers, etc.)
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-21-
1 -
'~~
immediately creates situations of interaction of the higher
order, even before the organization as a xhole comes into that
~_._
type of contact for xhich it is formed.
The selection of the optimal inner and optimal outer opera-
tionts can be made only if the aim of the organization has been
formulated. Since the choice of one rather than the other
operation will make a difference in the attainment of the aim,
the latter should be quantitatively stated. In business this is
conventionally accomplished by using monetary terms; 1n other
economic yields by means of formulating a numerical utility.
0
Elaexhtre there exist only the vaguest descriptions of what the
aims of organizations are (e.g., a university) with no attempt
at any sort of quantitatipe formulation. If a quantity is indi-
cated, it becomes sometimes possible to look upon organization
as an arrangement of inputs to be compared with outputs. But
we x111 have to ask ourselves xhether this holds true for all
organizations irrespective of whether or not their aim be quan-
titatively stated (cf. Sect. 9 where more is said about aims).
An inability to indicate a quantity now does, of course, not prc-
0
elude the possibility that a measurement be invented. The
evaluation of operations is exceedingly restricted where no
numbers at all can be attached to the attainment of the aims of
an organization. Discussions of such cases often lack precise
scientific method. Yet often they are not only not impossible
but even precise:? If in war one army wine and the other loses,
the former has reached its aim decisively-hut it is unknown
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~
-~2-
with what deviation from an optimum. And perhaps the aim of
the army should have been to be so strong that no war would have
to be fought! Then the determination of the attainment of the
aim is a different matter altogether.
There is, of course, one (trivial) method of character-
izing the aiml which is useful in the study of biological orga-
nization. This is to point out the fact that any biological
organization is bent on transforming?matter and/or energy. This
is obviously true in a cnost~universal~senae and applies also to
socoal organizations. All types of transformations can be con-
sidered. The formulation of these activities by means of state-
ments about inputs. and outputs is only one particular form of
expressing this.
0
Among the entities an organization is made of are human
beings. As employees, workers, etc., they are placed in certain
positions and are charged with the performance of certain tasks
(under the restriction of feasibility). Their arrangement and
the interplay with the non-human resources of the organization
is entirely a matter for the inner operation. The human indi-
v ideals are in this sense inside the organization. (Yet this
does not make them "organs," or the like, of the organization;
for this more is needed. Cf. Sect. 3.) But different from the
_inlert,other resources of the organization (equipment, supplies,
?_~- r--;etc. ) the- tndividvale ore oleos autsf.~e the ox~gnnization by ;being
~----
:~
As a ru3e, there wi11~ be more then one aim. ' There is atahole
set of aims, whose elements are to be numbers. In the
discussion above we may think of "the" aim at least as a
vector.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-~3--
able to make claims against the organization, e.g., for payment
of wages, salaries, for particular treatment, etc. The inert
resources have no claims; restrictions regarding these are
purely technological and possibly legal, inasmuch as they may
be owned by other organizations. The individuals are-therefore
in a dual position. This is true even of an army which cannot
treat soldiers in an arbitrary fashion, each soldier having some
claim and some status outside the army vis-a--~is the army. This
dual~~nole is one feature that distinguishes a social organiza-
tion from any natural one with r~hich it might otherwise be com-
pared. It creates immediately a higher order of complexity that
is peculiar to social life. However fabulous the organization
of the human body, and of the brain in particular, this duality
alone is a distinguishing feature setting even the small social
organization apart from biological organization.
The consequences are innnediate: The organization can, on
the one hand, order the individual around in his capacity of
forming part of an inner operation, but, on the other hand, the
organization must even include its relation to the same indivi-
dual among the characteristics of the strategy it chooses for
performing its outer operation. The organization is therefore
engaged in at least.-two simultaneous games: one involving the
pursuit of its aim directly, the other in whihh the human
resources are confronted with the organization as a whole.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~
-2 ~--
? 3. The Arrangement of Competences
An organization consists of "oompetences," i.e., elements
that are empowered and able to set acts committing the organi-
zation to its outside world and~or to partake in the inner
operation. There elements are usually individuals or groups of
individuals; but the human ,factor is not necessarily associated
with each competence. In the first case the competence makes a
move in the game of strategy in which the organization is involved
when pursuing its aim; it commits the organization.
In the second case the competence sets acts which transform
or use the inner arrangements of the resources of the organiza-
tion. The competence is assigned variables (or constants) with
a prescribed domain. The competence is given the task to act
under specified circumstances, and the duty to act optimally.
An act by the competence consists of choosing a value for the
variable. This will be done upon receipt of a signal (ef. below).
An act is a choice of a value of a variable (or of a vector)
made by a competence, and it is specified by stating the value
and the variable; an operation is a sequence of such choices or
sets of choices and is described (or specified) by giving the
sequence of values or sets of values. A variable will be called
admissible (with respect to a given competence) when it is within
the physical and legal powers of the competence to choose a value
from the domain as required by the task.
When no choice is left to the competence, then a constant
is assigned to it; upon receipt of an appropriate signal, the
act oorresponding to, and described by, the constant-if admissible
will be set by the competence. We call these the smallest
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
~~
competences of which there may be many in an organization, but
no social organization can consist only of constants, i.e.,
smallest competences. This follows from the fact that an orga-
nization as a whole has to make choices and the representation
of the organization by constants only would_ preclude the exis-
tence of a problem of selection. Every organization is only
probabilistically Formulated, although some to a higher, soae
t o a lower extent.
The smallest competences may easily be compared to those
automata who can respond to an impulse only in a unique manner.
Indeed, many such competences are being replaced by machines.
For example, the man behind the post-office window selling
three-cent stamps can be substituted by an automat. (This is
a not trivial substitution because the latter is entirely inside
the organization, while the employee is also outside.)
Even the smallest competence has a decision to make. For
it, the problem consists of deciding whether the signal is the
proper one that should cause it go to into action.
The representation of a competence as an element of an orga-
n ization setting a value of at least one variable corresponds
precisely to empirical conditions. Theoretically, it might
appear as it the activities of an organization could be com-
p letely described so that it would consist only of a set of
smallest competences in whatever groupings and arrangements that
xight be found convenient or satisfy some architectural or~other
principle. This is, however, impossible (ef. below). There-
fore, moat competences are given discretion to choose what appears
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
xM-734
--~6-
to them to be the optimal reaction to the stimulus that causes
it to move. This freedom of the competence is well Down and
extensively treated in legal writings under the name of "freies
Ermessen?." For law courts it means that the codification has
been carried only to the point t~rhere a further specification of
the duties and the powers of the Court id left undetermined
(within the framework of the containing law). In law the ten-
dency has always been to eliminate as much of this indeterminate-
ness (i.e., unpredictability) as possible.
Competences are ranked (i.e., form a hierarchy) which, how-
ever, need not be the sole hierarchy of the organization. The
highest competence is called the Source. In it is vested the
entirety of all variables of the organization. The notion of
the Source is an undefined concept needed fbr describing orga-
;riizationa. The smallest organization-of no further intereBt-
has a Source with only one single constant. Hence there is no
choice and no further competence. A competence is created by
the assignment and grouping of variables at the disposal of the
Source. The competenoes correspond in some way to the "organs"
of the body; it is preferable, however, to use a neutral term
in view of the dangers of organic analogies at this stage. Com-
petences are always derived either directly from the Source or
indirectly from other Competences. This initiates the compl-l-
Gated process of delegation. Parallel with this process is one
specifying the tasks and duties of the Competences (in military
parlance called Missions). The Competences are therefore char e_d
to carry them out. The admissible variables assigned to them
are to be in a one__to-one correspondence with the impose duties.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-27-
If this principle is carried out for the entire organization
there is a natural way of stating that it is in a particular
form of (hypothetical) equilibrium. It is hypothetical as far
as the construction goes. In its actual operation, events may
occur so that the organization is far removed from "equilibrium."
The variables fall into three classes, inasmuch as some
are purely internal, others purely external, and the rest mixed.
That is, the first variables partake exclusively in internal,
the second exclusively in external ,. operations. The distinction
is easily seen in military organizations where, for example,
in the operation of a ship, numerous operations and entire
sequences of acts are due to the use of variables which are
completely separated from any outer operation. The same holds,
of course, for any organization of some size. The distinction
becomes important, however, especially in connection with the
signaling system (cf. Sect. ~).
It would be desirable to get a clear and complete idea of
what a "variable" is in connection with an organization. To
s ome extent it is an undefined concept, or at least a not--~well-
defined notion. One is, however, comforted in that difficulty by
the observation of H. Weyl: "Nobody cap say what a variable is"l
--and this refers to pure Mathematics! A variable is the form
in which we state that "the organization" can commit resources,
or set acts, or whatever semantically equivalent expressions may
be, in 'orQer to a~Complish some purpose. "Variables" ai~t~ _ __
- - - - .~_..._..._ ..ww~. _.~..._...__~ _ ...... _~ . ~, ~ . _ w a_....r . - - -..._ ~ _ _ .. ......~__..
Philosophy of Mathematics and Sciences, Princeton (199)?
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-28-
available because there is usually a choice between alternative
uses of these resources. The total resources have to be dis-
t ributed or allocated according to some principles connected with
the over-all aim of the organization. The more clearly the aim
is formulated, the more likely will the allocation be success-
ful to the extent that it facilitates the pursuit of the over-all
aim of the organization.
The resources may be the (at first: liquid) c,~pital of a
firm, or the time and willingness to help in a volunteer fire-
fighting company, or the weapons and other equipment of an army,
etc. The form in which the resources are held makes a variety
of operations (i.e., arrangements of their uses) possible. One
way of deciding their use is to specify in complete detail
exhaustively at the begir;ning of each operation what seq~3ences
and combinations of acts are to take place at what moments of
time. Such dispositions are imaginable but not usually found in
reality. One difficulty is that operations tend to be very com-
plex and cannot be specified in advance, especially not when con-
tinuing organizaicions of sufficient size are considered. Another
difficulty- lies in the contact of each organization with Nature,
which produces randomness and requires reactions that are in
principle unpredictable i~n the detail in which the final action
~~ ~~ ~ hay tor~aocur.
1 None of this prevents the use of the notion of a strategy as
in the normalized form of the game. Cf. von Neumann and
~ Morgenstern, oP. cit., Chapter II.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~F
-29--
The domain of the variable expresses alternatives of
action or choice. As n rule, we will identify "power" of an
organization xith the ability to cover a wider rather than a
narrower range of choices. Therefore, if a variable is given
a larger domain, its significance in operations is greater than
if it were a constant. This characterization of variables and
competences--the latter usually. comprising several and identi-
fied by the names given to this~hierarch~o~il arrangement-
significant for the operations and may differ from their heirar=
chieal position. Frequently, however, a higher placed competence
will be in possesaion.of variables with large domains. Superior
competences can restrict or enlarge the domains of lower variables.
In the first case resources may be set free and become available
for others; in the second case the shift may go at the expense
of others if all resources were allocated at the outset. The
shift of resources from one variable to another is not arbitrarily
,possible, sometimes not at a11. When the resources have already
been committed but not used up (e.g., the liquid capital is
transformed in part into instruments and raw materials), such
transfers are eo ipso restricted. All "reorganizations" have to
take these factors into account. The higher competence can ~
restrict the lower one either by a forms.l reassignment or in
the course of operations. The latter means that a choice
?
made by it may make it impossible for the lower competence to
uee the full range of the variables at its disposal. This works
also in converse, as far as operations are concerned; i.e., a
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~+
30-
lower competence may, by using its variables (say, inappro-
priately), impair the full use planned by a higher competence
for its own variables. This situation can, of couree, arise
only if the time sequence of the choices to be made by the com-
petences is not ordered according to their hierarchical sequence.
below).
c ~o what extent a previous inner or outer operation restricts
the possible future inner or outer operations will occupy us
again.
In delegating variables, the Source can never dispose of
all variables lest it lose its own identity. One of the created
competences could assume the role of the Source. This means that
in every social organization of the type discussed here (i.e.,
not yet discussing those described by the solutions of n-person
games) there will exist a basic competence from which all others
emanate and are derived. This Source may, therefore, be a
usurper, the people's assembly, an oligarchy of directors, etc.
Save for the simplest organizations (with only one constant or
even with one variable) there must be a hierarchy through delega-
tion from the Source. The extent of the hierarchies, the number
of variables involved, and the manifold forms of direct and
intermediate delegation with all the attending influences upon
?
It is clear that this is almost never the case. A lower compe-
tence can, however, never formally restrict or enlarge the power
of higher competences; it can do so only operationally (ef.
the moves and strategies of the organization are the main field
of a (future) theory of organization. o
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RIB--734
-~6-
? to them to be the optimal reaction to the stimulus that causes
it to move. This freedom of the competence is well known and
estensively treated in legal writings under the name of "freies
Ermessen?." For law courts it means that the codification has
been carried only to the point where a further specification of
the duties and the powers of the Court is' left undetermined
(within the framework of the containing law). In law the ten-
~d.ency has always been to eliminate as much of this indeterminate-
ness (i.e., unpredictability) as possible.
competences are ranked (i.e., form a hierarchy) which, how-
ever, need not be the sole hierarchy of the organization. The
highest competence is called the Source. In it is vested the
entirety of all variables of the organization. The notion of
? the Source is an undefined concept needed for describing orga--
x~izations. The smallest organization-of no further interest--
has a Source with only one single constant. Hence there is no
choice and no further competence. A competence is created by
the assignment and grouping of variables at the disposal of the
Source. The competences correspond in some way to the "organs"
of the body; it is preferable, however, to use a neutral term
in view of the dangers of organic analogies at this stage. Com-
petences are always derived either directly from the Source or
indirectly from other competences. This initiates the compli-
cated process of delegation. Parallel with this process is one
specifying the tasks and duties of the competences (in miiitary
parlance called Missions). The competences are therefore charged
to carry them out. The admissible variables assigned to them
are to be in a one~o-one correspondence with the impose duties.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-?3~+
31-
Delegation arises out of tKO reasons which sharply distin-
guish the cases to which they apply. In every case the dele-
gating competence imposes the burden of a choice upon the lower
oompetenceo In the first case, the competence delegates, althou !~,
given the technical facilities it could make the proper choice
itself. For example, a higher officer may be able to fire a pan
ticular gun as correctly as a private; but he may not be able to
2'ire it better. To fire it at all would conflict with other
duties, be wasteful from the point of view of optimum allocation,
etc. Hence, it is not done but delegated. In the second case,
the competence delegates because it cannot make the proper choice.
This may be due to technical circumstances (e.g., requirement of
special skill) or because of the position and time of the choice
in the entire sequence of all choices, depending therefore on
particular states of information. In the second ease a true
delegation takes place, inasmuch as without it either no decision
xould be made at all or most certainly not an optimal decision.
We shall now discuss one typical set-up of competences and
discuss the question of the sequence of acts or choices and the
connection with the signaling system (which will be separately
examined in Sect. ~+).
Consider first the Source S and assume it has available the
variables x, y, z, which it can dispose of in any manner it may
choose. That is,~~te assume that there are no technical reasons
tvhy some variables should be associated together, restricting
the freedom the Source has in allocating them to competences it
Kishes to create .- From this~one concludes immediately that in
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~
32-
whatever manner the Source may make its delegations of the
three variables, it must retain at least the power to make and
remake these delegations. If this were not the case, the Source
would have lost its identity and a totally different organiza-
tion would result.
The variables x, y, z can be put into one competence C1
whose task it is to carry out the choices as they become nece s-
sary. Or else the Source may create two, three competences
C1, Cs, C9, distributing the variables accordingly. If only C1
is created, either the Source or, if possible, Ci may make fur-
~ ~~ n ~
ther delegations Cis C1i where C~ is a delegation from CZ.
Example with three vax~lables:
C.1
C1.
~~ ~~lXs ~~ Z)~ ~ ~ ~
or y, or tX; y1 yor x, z or y, z
Figure 1
At least one variable must be retained by Cis therefore, C1
can at most have txo variables. There are six possible dele-
gations. A further delegation to C1 (subordinate to Ci) is
impossible when C1 has only one variable. When it has two
variables, six alternative delegations (i.e., x or y; x or z;
n
y or z) are possible and C1 can consist only of a single vari-
able at which point the possibilities of further delegations
are exhausted as far as this line of delegations is concerned.
This purely schematic illustration shows the following:
Even if only three variables are available, a considerable wealth
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~
33-
of organizational possibilities exists. As far as the possible
a ceomplishments of the organization are expressed by the vari-
ables--and the choices made-there is no difference between
these forms and delegations. But from the point of view of
hierarchies, of competences, the communication among the eom-
petences, the aignaling~system, etc., great differences exist.
Now, as already stated, there will seldom prevail complete free-
dom as to the creation of competenees, since technical, ps~rch-
ological, legal factors may rule out certain combinations or
may make some preferable over others.
It is immediately seen that in spite of the separation of
Ci and C1 and the possible coexistence of a Cz (with other vari-
ables, u , v, .,.) directly set by S together with C1i a~ order
of sequence or simultaneity of the possible acts described by
the variables can be specified oT.may_be allowed to happen upon
receipt of the proper signals. If a specific sequence of acts
is required another organizational device in addition to the
chosen equivalent delegation set-up is needed. This is a par-
ticular signaling system. As already shown, every organization
requires a signaling system without which no actions would occur.
But there exists a certain amount of freedom for the construc-
tion of the signaling system which thus needs to be considered
in conjunction with the system when both are specified together.
Second, we consider the source S and the variables u, v,
w, it can dispose of, now extending the first case so that one
variable, say u, has other variables r, s, t, ae its domain,
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~+
3~-
with r, s, t, each having their proper domains in turn, none
of them being a~constant. This produces a substantial differ-
ence in the effects of delegation and has a bearing upon the
influence of the signaling system. Example with three variables:
C 1: u, v
'yCl?: r-orsort~
,, ..
' (Dom$iri ~of u)
Figure 2
3 creates C1 which chooses between u, v, and CZ with w. If v
and/or K are chosen, they are taken from their domains and this
constitutes the final a~etion. If a signal requires that u be
chosen, one of the variables r, s, t, is selected, each of which
~~
is alternatively assigned to Ci which then selects an appropri-
ate value from the proper domain of the variable chosen by the
superior competence C1.
Now it is cldar that u, v, w, can be set simultaneously or
in any order depending upon the arrival of signals or the cys-
t erratic set-up of the signaling system (obviously, two different
possibilities). But r or a or t cannot be simultaneous with v
or w unless the chance distribution of signals produces such
response, or a specific restriction is placed upon CQ such that
it cannot choose a value for w until after the choice has been
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73 ~F
35-
made for r or 's or t. CZ therefore depends then upon the receipt
of a signal (direct or indirect) from Cl*. This particular sig-
n sling system has therefore depressed CZ from its apparently
e quivalent hierarchical rank with Ci (both directly created by
S) to one operationally below that of Ci. -
This shows than the description of a delegation. system is
incomplete unless the simultaneous signaling system applied to
i t is also explicitly described. A hierarchy of competences
alone will therefore not indicate completely what sequences of
actions have to be expected (besides consideration of the random
distribution of signals, i.e., of events). In the sense that
it is a restriction of the freedom of action if one competence
has to wait upon the prior action of another, the scheme of
delegation is to be amended by making such cases explicit in
order to express the full state of affairs. This is at least
true if the operations of the organization are subject to inves-
tigation and their interdependence, e.g., over time, is to be
studied with regard to the amount of inforri~a.tion within the
organization at each moment of time.
An illustration showing at least a temporary annihilation
of authority of higher competences by lower competences during
operations is shown in the following: Security regulations may
require that each person has to identify himself to the complete
satisfaction of the guard at the gate, even officers far above
the rank of the guard. The guard's competence in this particular
instance, place, and circumstance exceeds that of anyone else.
After this superior has passed, he may dismiss, replace, or
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
~-73~
36-
? otherwise reassign the guard. The pilot of a plane is in
absolute command during the flight, and so it is in many other
situations. This~is partly a temporary superseding of higher
competences, but it is one that is foreseen when the structure
of hierarchies was created and the possibility is built into it.
It would be important to discover whether this._~henomenon arises
necessarily in types of social organizations. At any rate, it
shows that the notions of subordination, higher authority, etc.,
have to be treated with great care and are probably not well
defined unless the domain over which they extend and the specific
situations to which they apply are characterized in considerable
detail.
So we see: Equivalences between apparently hierarchically
quite different organizations can be established by means of
appropriate signaling systems. Thus the intermediate delegation
(Fig. 2) from S to C1~ via C1 can be made equivalent to a direct
delegation from S. Ci is in the position of a general officer
in the army who decides which of the various forces at his dis-
p osal---.artillery, tanks, planes-to use and then, after the
choice, lets the commander of the chosen weapon do the alloca-
tion of the quantity, deployment, location of attack, etc.
?
The question arises next whether the Source has complete
freedom in arranging the competences, forming arbitrary groups
of variables, etc. The answer is in the negative, but the
correct and exhaustive formulation will be difficult. Clearly,
the variables are related to physical circumstances such as
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~+
37-
the working hours of a man and the properties of machines placed
at his disposal. These, in turn, are dependent upon the aims
of the organization and on the strategies it uses, has previously
used, or proposes to use, in its relation to other organizations.
For example, a fire-fighting company has to have engines and
men .to run them; but it may use these in different proportions,
place authority in a variety of ways, etc. Thus, there is
neither complete freedom in arranging the variables in competences
nor is there complete rigidity. The latter is lacking even in
such organizations as line production, set up as conveyor sys-
tems; the question of flexibilityl will be taken up at greater
length in Part III. The freedom is greatest when there is com-
plete homogeneity among the resources. But even then there are
restrictions--e.g., due to the signaling needs, to the degree of
the complication within competences, etc.--and once an organiza-
tion has actually emerged from the initial state, the form it
assumes co-determines together with its experiences the form of
its continuation. Besides, these tLchnical interactions of men
and machines, and among men, which limit the freedom of the Source
in setting up competences, there is a time-sequential factor.
Choices have to be made often in a particular sequence. Now even
if there are no other ob~ectiona against grouping some variables
together into one competence, this may still be impossible if
l The distlnetion between flexibility of an organization regarding
the choice among various hierarchical set-ups and flexibility
in the choices oY' how to pursue its aim, given one of these, ~s
necessary and `not always maintained where the notion of flexibi-
lity is encountered. More about this in Part III.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~F
3$-
the time structure of the choices were such as to make their
execution impossible without producing conflicts in the persons
charged with the tasks of acting---however qualified these per-
sons might be to deal with these variables separately or in
different groupinga.l
The arrangement of competences thus far described makes
one basic assumption that will not often be fulfilled. That is,
the description of the possible operations of the organizatio n
as outlined by the variables and their domain is precise and
exhaustive in terms of the situations in which choices have to
be made. It is justified to make this assumption for a first
outline of the set--up of an organization. But its description
has to be extended. It will be recalled that the extensive des-
cription of a game of strategy can be given in a form that does
not introduce probabilistic elements, but that nevertheless makes
incomplete information of the player, even about himself, possible.2
The characterization of all possible situations with which
the competence has to deal has to be given ex ante; i.e., prior
to their occurrence. The impossibility of accomplishing this
introduces a probabilistic element into the very structure and
desi of almost any organization. In some it will be smaller,
in others larger. In some cases it can be gradually reduced
through the process-of learning (ef. II-7). This probabilistic
factor has to be distinguished from that arising from the choice
1 Thie~is closely related to the idea of the "Maximum Command
Span discussed in Section 11.
2 von Neumann and Morgenstern, off. cit., Chapter ZI.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~+
39-
of statistical strategies to which the organization may have to
take recourse. If no uncertainty at all prevailed, it would not
be necessary to assign to a competence anything but constants;
i.e., upon the receipt and identification of a signal a unique
reaction would occur. However, even in this respect there are
at least two sources of error; i.e., there- is a rudimentary
probabilistic element even in the smallest competence (cf. above).
For example, an automat might respond to the wrong coin or pass
out the wrong commodity, or none, even if the correct button
were pressed or coin inserted. This behavior, however, falls
into the category of errors (cf. Sect. 5) and the role of pro-
bability is different in that respect from the one referred to
above.
The famous and perennial phenomenon of the conflict of
competences arises-if not due to faulty design of the organiza-
tion, e.g., by (sometimes even willfully) assigning the same
duties and/or variables, or overlapping domains, to different
c ompetences-from the impossibility of a complete ex ante des-
cription of all possible situations to be faced by each com-
petence. These conflicts ought to be minimized, if such a mini-
mum exists and can actually be found. This is probably the case
in the first instance, but much more doubtful in the second.
While the phenomenon is of primary practical significance, it
may suffice here to point out its existence and its origin in
the impossibility of a complete description. The conflicts among
competences have a wider significance in that they offer the
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~F
-~0-
possibility of strategic moves by the Source, within the organi-
zation against parts of the organization. But this is a topic
with which we are not dealing in these notes;l if such operations
occur, the organization has to be viewed as a composite one
(cf. Sect. 12).
We can summarize: The Source is in the possession and
undisputed control of all the resources of the organization.
These are allocated by the creation of variables that are combined
in competences which are empowered to select values from the
domain of these variables upon the receipt and recognition of a
signal. The signals are tied to the variables, but they are
received by the competences. The selection of a particular value
of a variable is a choice and constitutes part of an operation of
the organization. competences having only one variable with a
single value as domain, i.e., having only a constant, are called
the smallest competences.
We have, in these discussions, abstracted from the human and
psychological aspects. There are some, however, that must be
mentioned even at this level. The principal consideration is
that a competence is in charge of an individual or a group of
individuals. Now a great deal follows from this fact about the
manner in which variables can be distributed among competences
(depending also upon the nature of the variables, i.e., their
domain). For example, too much concentration may interfere with
1 Illustrations are found in the administration of President
Roosevelt in which agencies were created that were given almost
the same dut~.es as others but different authorities; this was
done in order to get rid of uncomfortable rivals, adherents
who had lost their usefulness, etc.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-?3~
-~1-
the physical ability to make the choices or to make them in
proper time. On the other hand, concentration of variables will
give a competence all the signals and information tied to the
variables in a combined form. This will make some communication
channels unnecessary and the accumulation of information may
actually improve the ability to choose. All this is also related
to the technology of the particular organization.
If there exists a large organization where there is, say,
a president who is formally in charge of operations and res-
ponsible, but where the number of variables assigned to him is
too large for physical attention, this then implies that at least
de facto delegation must have taken place; otherwise a physical
impossibility would have been stipulated.
Signaling, Information, and Memor~r
Every organization relies on at least one signaling system;
but there can be many. One problem would be to find the "optimum"
system, defined, for example, as the one involving the least cost
or producing the greatest speed, the highest accuracy, etc. A
signal serves to excite a competence. An event is recognized by
competence as a signal. This may be a buzz, the ringing of a
bell, a noise made by an enemy, or the reading of a dial, even
by a machine. The recognition means that an interpretation of
t he event is made, e.g., by comparing it with an instruction or
key, a previously established signal, an expectation as to what
the event might be like, by adding it to other signals already
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM--734
--42-
received and treated, etc. We make no distinction between
signals add messages as is usually done in communication theory.
This distinction, necessary there, is not required at present.
At a later stage it will become important when the acts of an
organization are related to the amount of information, the rate
of flow of messages and the efficiency of signaling systems, the
codes used, etc.
There are internal and external signals. The first are
transmitted from source to competences and back, and among com-
petences. Such signals are transmitted, sometimes by means of
a separate inner signaling organization (cf. below....), some-
times without, depending partly on the quantitative extent
of the organization. A signal may be an act of choice by one
competence the results raf which are noticed as an event by another.
Thus, the open fulfillment of a duty may be all that is required
in order to pass on information and a separate signal in addi-
tion may be unnecessary. The event caused by the choice made
by the competence may happen inside or outside the organization
depending on whether the competence is involved in inner and
outer operations. External signals are events coming to the
organization from the outside, even if not intended to be a sig-
n al or a communication (e.g., the noise made by the enemy); this
is then an involuntary signal; only external signals may fall into
this class. It is intentional, for example, if it is a customer
entering a store. Most of these external signals are random
events (Prom the point of view of the organization) with various
probability distributions; they are not random if the other
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-?3~
~y.3_
organization uses a pure strategy composed of several moves.
The internal signals may follow entirely different patterns but
cannot be free from randomness. An internal signal may cause
an act in the inner or outer operation and an external signal
can do likewise.
Any signaling device, especially when it is large and is
t ransmitting a high volume of messages, produces noise, i.e.,
messages. This is empirically very important for those numerous
organizations in which the signaling system is rudimentary and
improvised, and where the question of noise is not particularly
studied. Every organization has to have some, however rudi-
mentary, technique for filtering messages. This will, as a rule,
be achieved by comparing a message with a previous one that had
proved itself to have been correct, or comparing it with an
instruction specifying the duty of the competence, which would
always require the mentioning and description of a signal. The
identification of the signal will cause the competenee~?to,
make one of these decisions: (a) to transmit the signal or a
new one, alone or with other messages, to another competence; or
(b) to set a material act, i.e., to pick a value for one of
its variables. If (a) alone is the ease, then the organization
has a separate information organization included in its own
organization as described by the existence and distribution of
Source, variables, and competences.
We return to this case below. We now consider only (b).
When the competence acts materiall~yZ i.e., decides on, ease
(b), it is acting either internally (in the internal operation
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
of the organization) or externally (making a move, in the game,
on behalf of the organization). If the competence does not act
upon the identification of a signal requiring action (though
this may be an erroneous identification:), then an error has been
made by the competence. If the competence cannot act,a "mistake"
has been made by the organization from the point~~of view of the
competence by confronting it with an "impossible" task. The
latter case arises when the competence has not been given the
necessary admissible variables or domains in order to produce
the response to the signal. E.g., the store may be out of stock
and the stock level is controlled by another authority; or the
military force is underarmed, etc. However, it will be seen
that this need not be a mistake of design of the organization.
On the. contrary, the organization may have been planned by fore-
seeing the occasional occurrence of such circumstances. In
other words, the design may include a probabilistic element, and
the question would then be whether the
mate has been made.
A memory is an indispensable part
has any continuing existence. Even an
ad hoc has more than one single act to
inner operations require more than one
involve more than one move in the game
device is needed to retain information
proper probability esti-
of any organization that
organization created
set; i.e., either the
act or the outer operations
of strategy. Then some
in order to carry out the
next steps. Such a device is a memory organization. First, it
is necessary that the Source know that it has created competences,
what they are, what variables of higher and lower order have
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-734
-1+5-
been allocated where, what duties were specified, etc. This
burden upon the memory is in itself far from trivial, especially
in a large organization. There, however, the memory need not
have every single item that is stored in it in a ready state.
It often suffices that it has the path to the stored piece of
information identified and open. Second, an organization must
know, in some form, its, state of operations.l Few organiza-
tions will know their state precisely (still using an anthropo-
morphic language here); they will often be only statistically
informed and this may be entirely adequate. Precise knowledge
is often required, e.g., in chess; but it is not necessary to
remember the steps that led to the position in order to finish
the play in an optimal manner. It is sometimes even impossible,
ae in Bridge. (It will appear later (cf. ....) that this is a
different information from that which must exist for individual
competences about the functions of other competences in order
that an organization can operate.)
The memory is used in fixing the operations of the organi-
zation. It is not necessary that all items the memory has to
preserve be stored in any particular place identified as the
"memory." The memory of the organization may (in part at least)
express itself in the very form the organization assumes. We
shall call'such a case the protocol memory. This means that
? restrictions are placed upon the variables that are at the dis-
p oral of the competences, or that new variables are introduced
whenever some information accrues to the memory. Phenomena of
l Cf. further elaboration below about the "organization homunculus."
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~+
-~+6-
fir-
,~;,
this nature can be observed within each organization. The pro-
t ocol undergoes a gradual change that epreases a learning
process (cf 7). ?rotocol is, for example, the drill imposed upon
soldiers; it expresses a past experience with the handling of a
particular weapon. It is often found that the best way of coir~-
municating and 'utilizing this experience is in the formalization
of the operations. The fact that this is then a part of the
memory is not alxays clear to those concerned, but the purpose is
achieved. Anothar form in which memory is expressed is routine.
This is an "automatic" selection of the correct value of the
variable, because the frequency with which the same situation is
encountered can be relied upon. There exists, apparently, a
strong parallel in human behavior where, for example, the
experienced pianist need not decide each finger movement cony-
sciously and separately. The remembrance of a piece seems to
lie in the hand itself which almost instinctively strikes the
proper tones.l
There are other forms of storing information. These depend
mostly, it appears, on the size of the organization. When it is
1 arge, special memory organizations have to be created which
function very much like the memories of computing machines.
Memory organizations are entirely internal; i.e., have no
external activities (they never operate on behalf of the whole
organization, oosaitting;it Eo the outdlde.
1 This 1s, of course, exceedingly doubtful and possibly even void
of meaning. Nevertheless, there is some value to the analogy.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~+
--~?-
The staff of an organization (apart from the working
and executive functions) is usually a competence charged with
storage of information, and as such is part of the memory. Its
functions, however, transcend this assignment it also serves to
collect and interpret information which is not derived from a
learning process or other past experience of the organization.
It has the assignment of gathering intelligence also from the
outside of the organization. It is usually assigned to the
Source or the highest competences. But these functions fre-
quently overlap empirically, although there is no structural
necessity.
y _.._. Besides the indispensability of s memory for any continuing
organization, it is required for learning. The functioning and
value of a memory depends upon its accessibility; dead storage
of data does not make them information. The data become informa-
tion only when they are accessible to the proper competence
that can utilize them. Social organizations appear to differ
considerably among each other in regard to the extent and acces-
s ability of their memory. In the human body the memory is
apparently available to all organs on an equal basis. If some
appear to have more easy access than others it is, if true at
all, perhaps due to frequent challenge and stimulus. In social
organizations (always centrally directed:) there is usually,
perhaps even necessarily, this feature: The memory is more
diffiEUlt to get hold of constitutionally for some competences
than for others. This is partly intended this way, partly a
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RA+f 73~
-48-
consequence of the particular signaling and information system.
When planned, it involves intended secrecy; or access to the
memory is reserved as a privilege to higher parts of the bier--
archy. This tendency occurs often together with the opposite
one of distributing (parts of) general information to all com-
petences indiscriminately. ?- --
The memory o2' an operating organization need not and should
not hold its contents indefinitely. If the learning process has
been accomplished and some experience is used--that had been
stored in the memory then the memory will have to be relieved
of these contents to make room for new storage of those exper?--
iences that will aceure after the reorganization. The latter go
on all the time in business, government, and the military forces.
They are in part an expression of the process of learning.l The
information released from active storage most often is wasted;
only in rare cases will it become subject matter for historical
purposes.
Next we observe that a distinction has to be made between to e
system of signalingL;(or network) on the one hand, and the use made
of it (or the actual flow of messages through the system) on the
other hand.
Although every competence must be linked to the network
directly or indirectly, it need not be able to send messages at
all tinres,~, or . Indeed even. tct .re_aeive therii' Srom the Source or oth?e~
- ---- - -- - -- - - - .- . y.
-~__
1 In part they are caused by the need to combat the ever-present
tendencies toward disorganization, i.e., the entropy of any-
thing organized.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
?
4~1
RNi-73~
-4g--
competences (e.g., ship alone at sea in war). Discontinuity
of actual dispatch and receipt of messages, and of the ability
to send or receive them,4ls compatible with the operation of an
organization. Indeed, it is possible that acts about which
information must be sent need not be the subject 'of a separate
message or signal. Thin is, for example, the case when a chain
effect between events exists such that x leads to y leads to z
and z is the signal that all preYiaus acts have been actually
carried out, at least?withia some probability. This may be-the
only message the Source or some high competence requires; it
then receives automatic cumulative information. This is, of
course, a different kind of comprehensive information than that
resulting from a process of digestion of messages carried out
by interpretative suborganizations (e.g., staff}. The accuracy
of the former may be much higher than that of the latter,
because the latter are subject to error of interpretation in
addition to the noise that afflicts each single message separately.
The frequency of messages-in whatever direction--may become
so large that the signaling system is unable to handle them
within time intervals that are short enough so as not to impair
the meaning and value of the message. In that case we speak of
congestion. Although this is a term with immediate Intuitive
meaning, there is no satisfactory mathematical theory available
that would help in determining the point when such systems become
;, , .. ,.,
"congested.?"1_ Yet congestion is undoubtedly a reason why some
Obviously the problem arises for telephone exchanges, railroad
yards, etc. But no theory exists for the latter, and it is
difficult to envisage. There are statistical criteria for con-
gestion of telephone exchanges that are useful for their design
and operation.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-50-
systems--and the organizations they serve--will break down
at least temporarily and perhaps only partly. If the density of
signals within the organization changes, very simple changes in
the structural topological--nature of the system can often take
care of the variation in flows. For a theory of organization,
it would obviously be of prime importance to understand such
crucial phenomena, because it can be surmised that frequently
very minor organizational changes may be sufficient to change
profoundly a given picture beset with difficulties. Adjustments
may, however, not always be continuous but require discrete
alterations in the organization set-up, either of the basic orga-
nization or of the signaling system. These critical points
would seem to be of particular interest for the discovery of the
events assooiated with the size of organizations and for the
understanding of the phenomenon of growth and expansion.
So far this discussion was in terms of special information
intended for a particular, individualized competence or a group
of competences. These were the messages considered for their
excitation. In some organizations there is, as already mentioned,
also a flow of general information which is sometimes even
produced by an organized effort. In these cases all competences,
or at least special groups, are informed of events without any
clear idea that these messages will excibe a particular compe-
tence, or cause a particular act, or any groups. It is merely
anticipated that receipt of these messages may in some way con-
tribute to the_funetioning of the organization as a whole,
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-?3~+
51-
? especially in improving the performance of the recipients, if
only a group of eompetences are affected. '1~aya in xhich this
is carried out are numerous: There are the general alarms,
alerts, etc., that are sounded; there is the publication of
company news items, internal statistics, etc. On a wider scale--
no longer only within a single organization which is the case
here under consideration-there is the role of newspapers, of
scientific publications, meetings, and so on. It is interesting
to note that the distinction between special and general infor-
mation is also made in the human body: The nervous system pro-
viaes the former while the blood stream, reaching every part of
the body, provides the latter. For example, it carries warnings
of danger, etc., by means of changes in its chemistry, pressure,
? temperature.
One Further point: It appears that when there are more
delegations o~ competences (for the same total number of variables
in the organization), there also is need for a larger signaling
system. This is only partly true: The ..signal is related pri-
m arily to the variable and no to the competence. A redistri-
b ution of competences requires a rearrangement of the signaling
system, unless the competence relies exal~sively on outside sig-
nals not controlled by the organization at all. At this occasion
it must be decided what the hierarchy and the sequence of acts
of the newly determined competences shall be. We know (cf.
Sec. 3) that the signaling system has a decisive influence upon
' these. When many variables are in the hand of one and the same
?
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RK-73~+
-~~-
competence there may even be such crowding of messages
that a sub-internal system for handling them may have to be
created. This would at least Involve the setting of priori-
ties for messages, etc., as it indeed often happens. From the
economic point of view, it is interesting to note that different
schemes involve different costs, yet they ms.y be functionally
equivalent. Economic theory has not dealt with these matters.
We return now to the distinction made in Sect. 2 con-
cerning inner, outer, and mixed variables. The possible connec-
tions of the variables to signals is summed up for convenience
in the following table:l
variables:
(Internal:
Signals:
(External:
Inner Outer Mixed
An internal suborganization exists if there is a complete
delegation system within the entire hierarchy of delegations of
cornpetences such that only inner variables are contained therein
and connected with each other, although A and H types may be
intermixed as far as signals art concerned. An internal sub-
organization comprising only type A is a possibility, but is
exceedingly restrained. Any suborganization, being part of the
whole organization, is necessarily open-ended since closed sub-
systems are an obvious impossibility.
1 If the frequencies with which the various combinations occur
were given, an -important measure describing the structure
of the particular organization would be available.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-53-
From these possibilities it follows that the structure
of an apparently even "simple" organization will, as a rule, be
exceedingly complicated. In particular, the organization is
not rsul'ficiently described-as already stated if only the dele-
gation of the variables is indicated. For an understanding of
the practical woFkings of an organization, it is indispensable
to know to what kind of signals the various variables are bound
and what the arrangements within the main organization are.
The same main organization, as described by some hierarchy, can
be of a very different nature from another one, which has the
same hierarchy. The network of delegations of variables makes
a number of different signaling systems compatible. Lattices,
trees, networks, etc., appear to be the proper forms to describe
the connectedness of variables, of eompetences (in which several
variables of different types can be held together) and of sig-
naling systems. The ordinary box like description of organiza-
tional structures, based on the idea of "reporting to" is at least
inadequate. The chains of "reporting" are essentially those of
formulation of duties, responsibilities, etc. They do not
always reveal the power vested in the organs. Much Less does
this description, or even a more detailed one in terms of the
variables of action, give information about the chronological
sequence of acts for any inner or outer operation (this is taken
up below).
Although this last point is of:~~,q~iderable importance, it
`._
suffices here to mention this much: ~We ha.ve distinguished between
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
5~+--
? the activity and the operation (cf. Sect. 2), the former being
the totality of all possible (inner and outer) operations, the
latter consisting of a variety of acts or sets of acts. A1ow it
is necessary to realize that one and the same operation can be
carried out, under certain circumstances and within limitations,
in a variety of forms, such that the chronological sequence of
the choice of values for variables is not rigidly prescribed.
E.g., a salesman can first sell commodity x and then commodity y
or vice versa. ?1.'his is, as we saw, partly a consequence of the
relation between type of signal and type of variable, but also
due to the statistical-probabilistic factor in organizational
set-ups.l
In view of the highly classificatory approach that we must
follox-teeing concerned exclusively with phenomenology-further
distinctions relating to signals and choices (for variables} are
in order. At the end of Sect. 10 we shall discuss the important
problem of the relation between hierarchy and sequences of acts
in operations.
A signal";is an "event" and vice versa. A signal is (after
filtering information for a specific variable or group of vari-
abler. It determines that a choice has to be made for the
variable(s) to which it is related; i? it does not necessarily
specify further, it may carry an implicit order. If the signal
carries an ezplicit order, it annihilates the variable by making
l The phenomenon occurs also in economics, e.g., in the theory
of consumer behavior xhere it matters greatly whether durable
goods are at all involved, etc. It would lead too far to enter
upon this here._- Cf. footnote, page 11~.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
xM-73~+
55-
it a constant. Or at least it restricts the variable's domain.
Signals that are not even implicit orders are only information,
the value of which consists for the competence of the possibility
of making another (i.e., better) choice than without that signal
a t the time when an order-carrying signal arrives. If a com-
petence has all the information that could conceivably influence
the choice of a value for the variable, we say that complete
information exists. There may even be more information available
which is then redundant. Great masses of redundant information
may obscure the relevant information which can then be obtained
only by processes of digestion, compression, etc. (cf. above) and,
in many cases, by inference or other logical operations. (~Phese
latter situations are left out of account here.) Their analysis
would involve the discussion of information theory in relation
to the process of inference.
An "event" becomes known through a signal; in fact, the two
are inseparable. For our purposes this classification has to be
made: (1) events of the organization which are either the out-
come of choices (a) in internal or (b) in exte rnal operations;
(2) other events, which are either (at) physical, i.e., produced
by Nature, or (b~) other organizations' choices. Events of
Nature are determinable if their probability distributions are
kno~rn. Even where-they are completely random in the sense that
their probabilities cannot be estimated, Nature is ntver male-
volent, i.e., bent on impeding the organization in the pursuit
of its aim. The latter may apply random, i.e., mixed, strategies,
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
56-
but that is a different thing Prom the possible randomness of
(at). Here rre have a motive arising from antagonism. If the
environment is not one of pure Nature but also contains ants of
other organizations in a mixture with events of Nature, the
statement that there will be no malevolence in the milieu is
most likely no longer true.
A change in information is thus necessarily equivalent to
a change in static of some organization including Nature; for
example, it is a change in the state of one's own organization
that (generates and) receives the information. t~fhen the signal
does not arise from Nature, it is also a change in the state of
some other social organization, since otherwise there would not
have betn a signal.
Thus, as far as contacts betxeen organizations are concerned,
there is never a one-sided receipt of information since it must
have been emitted elsewhere through a change in the state of
the other organization. There are also changes of the state of
information of an organization about itself; this will be dis-
c ussed separately below.
Next we discuss the operational dependency of choices;
this requires a further classification of variables far discus-
sing the operations of the organization. A competence, in order
to make (a) an admissible and (b) an optimal choice for a variable,
must have information; i.e., must receive or have received, at
various previous moments of time, a number of signals which
involve, in various combinations, the following:
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~+
57-
?
(OC) Technological knowledge: This is fully
incorporated into the domain of the variableT-
the knowledge is deterministic.
(aC' ) physical, legal data
(oC.'~) Random physical events : "Nature"
the knowledge is probabilistic.
(/3) Choices by own organization:
(~~) own inner (probabilistically
choices known)
(~") own outer
(~j) Outer moves of other organizations: Often only
' the fact of such a move may be known, not its
outcome (not even probabilistically); far example,
in Kriegspiel.
?
Choices (i.e., variables) fall into these two further
classes-apart from the above distinction about inner, outer,
and mixed variables:
(a) A choice is separated if it can be made (optimally)
without a specific tie-in with another choice. Such a choice has
no(specific) predecessor operationally. E.g., a customer entering
a store causes a~separated choice to be made by the salesman.
Every variable and/or competence except the Source has, of course,
a h~.erarchical predecessor or superior; but here we are talking
~,~~,
about sequences of choices in operations.
(b) A choice is successive if its execution requires
at least one specific predecessor; i.e., the establishment of a
value for another variable. This occurs most obviously in line
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
xM-z3~+
~8-
production (cf. Part III). We then have connected choices;
the variables concerned may alternatively be inner, outer, or
mixed variables.
(c) A chain of choices is given when in connected
choices each specific predecessor is of the same class. It is
easily seen that there can be only two types of chains of inner
v ariablea, depending either on (~~) or (~~~. There can be no
chains of outer choices. The length of the chains is given by
the number of the specific predecessors of the last member.
The question arises whether in any, even the most primitive,
organization there must be at least one chain. Or whether it
is enough to have at least some successive choices. It is doubt-
ful that any organization at all can be constructed (if it con-
sists of at least the Source and one competence) where all choices
are arparated.
It was shown earlier that an organizational structure is
inadequately described when only the hierarchical setup is
specified. It is necessary to include the corresponding, actually
given signaling system. Nox it becomes clear that even a com-
plete description-in that sense--does not inform about the
sequence of choices in operations. The principal fact is that
the various competences do not necessarily set in a sequence
corresponding to the hierarchy. Often the lowest competence acts
first; e.g., the janitor opens the building before the president
can go to his office. But the highest competence selects the
entire inner and outer operations. That the sequence of acts
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part -Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
.~,x~,,~
RM-73 ~+
59--
? cannot bfa~cording to some hierarchy follows already from the
fact thit*,,~~'
mach organization is confronted with random physical
eventa~:~~f;:The observation of the sequence of acts of an organiza-
~::;, r~ ,
! tioa':~~.th~refore not enough to establish its formal structure
evs~a~`.?i?f one should succeed in separating, in the observation
~'t`~,~o~it' the outside, the inner and outer activities.
F
k~~;.,a, ~ y~.r-~,~~~ The account of the functions of information is incomplete
nization about itself. It is clear that the Source must be
informed about the entire hierarchical structure, but it is not
clear what distribution of one-competence's information about
other competences is required. However, this much can be said:
A competence must know about the competence to which it reports
and about those on which it depends operationally in the sense
of the above remarks. If every competence had full information
about every other St might help but not necessarily; it would
clearly be ti+ras,teful, if not physically impossible, for most
organizations. For that reason we spoke of automatic cumulative
information in operations. In operations it is unnecessary and,
indeed, wasteful, for higher competences to know everything about
leas reference is made to the state of information of the orga-
the lower. It is also useless for lower competences to know all
about those that are very much higher, etc. In no case are the
operations significantly affected by such knowledge; but the same
is true for the general information mentioned earlier. There
also, no direct connection between the amount of information
and the ohoices can be established, although it is reasonable to
? assume that the information will improve the operations. It is
Declassified in Part -Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~+
-60-
noteworthy in that connection that there exists a formidable
barrier for military men, businessmen, or practically anybody
within an organization to describe what they are doing. Simi-
larly, the human being cannot say what is going on in his own
body, although he has plenty of experience in living and knows
that he is functioning. It requires at least one science in
each case:
The distribution of the knowledge of the organization about
itself can be described by suitable mappings. These are not
~~
;'simple; they are everywhere of the one-to-many variety and sha
be described more in extenso at another occasion. We find again
that in a significant way an analogy exists with the mapping of
the human body into the brain. An illustration is offered by
? ,,
the so-called motor-sensory homunculus, as described, for example,
by W. Penfield.l This mapping shows the relative importance given
by the brain to various parts of the body by representing some
more strongly and detailed than others. Similarly it is in orga-
nizations. There, however, the situation may be even more com-
plicated because the various competences---and not only the Source--
have to have their own mappings of the whole organizations. The
Source,, in turn, will have to be informed about their state of
information. - ~~ ~ '~
~.~.
There are, furthermore, several such mappings; the two''"prin-
cipal one s. 'are: (1) the hierarchy mapping and (2) the operational
?K a
mapping. Ad (1): The competences will, as a rule, not have e
? same and complete information about the different hierarchies
W. Penfield, The Cerebral Cortex of Man (1950).
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-734
-bl-
that may prevail in the whole organization. This does not
interfere with their operations. Ad (2): The operations of the
entire organization can be carried out as intended (i.e., even
optimally) when each competence has only information about those
Q ..
factors influencing its own choices and those of the surrounding
competences but not about others.
In this connection-also awaiting elaboration I mention
the further fact that the usual description bf organizations by
means of charts, etc., is at least incomplete. The lacking com-
ponenta are the quantifiers that must be inserted as well as the
clear expression of coordination and subordination which is
?
?
usually blurred. The application of relations logic will be neces-
sary in these cases.l
We saw at the end of 3 that every organization has a pro-
babilistic element incorporated already in its basic structure
of delegation of variables. From this, as well as from the fur-
t her fact that there is noise in the signaling system, it follows
that each inner operation has a (not-intended) probabilistic
nature. This view is strengthened by the observation above that
chains and other dependencies exist among the variables and that
the outcome of choices of predecessors may be known only atatia-
tically, if the outcome is known even to that extent. In many
cases a competence inay be informed only that another one has
chosen a value for a variable, without being told which. Or,
even weaker, the competence may merely be able to assume, with a
1 There is a high similarity here with parentage and itts logical
description.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RIB-73~
-b2-
certain probability, that a choice of a variable and of a value
has been made, inferring from general information and basing the
conclusion upcn its knowledge about the organization as such
and upon its previous behavior. Examples for such occurrences
are easily found in the military field, e.g., from naval opera-
tions conducted under radio silence and scattered over wide ocean
areas.
5. Mistakes,_Errors, and Recovery
In connection with the noise in an information system some
remarks were made about errors and mistakes. This is, however,
a vast and complicated field about which little is known and not
all of this knowledge falls into an acceptable pattern. The
best that can be accomplished is to describe some of the princi-
pal phenomena in the hope that further clarification can be
obtained by extension of such descriptions.
First a definitional d~.fficulty; A mistake would appear to
be anything that involves the misuse of a correctly designed orga-
nization. If such a thing is at all possible, it means that
there cannot be enough self--checks built into an organization
preventing such occurrences 1 Or can a correctly designed organi-
zation pick a non-optimal strategy? This would be possible only
if limited information is assumed. But a correctly designed
organization would have to provide for proper filters for noise.
3o the problem reduces to the obvious impossibility of eliminat3n g
5~ ~ Llle ~I'VW3Ul11~N1.V O.D~/CVVV `va a~ctiv~.aav iaa v.ay ~..~~+a.....~~...?. r ....... ... _
e i:P'..: : s ~ ....~? ~ ~.
_l E.g., is an organization correctly designed in which an incom-
petent person-gets into a top position .vghen the procedures
for appointment and promotion are observed?
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Do all cases of misuse reduce to this? It would not be true
if a correctly designed corporation put an incompetent person
into the position of its president. Can checks be invented which
would prevent such an occurrence? Since we do not know what the
"fundamental organizational" principles are, we know nothing
about their violation. All one can do in this situation, there-
fore, is to describe typical cases in which there is common-
sense~~green~ent that mistakes have been made and that errors
have occurred. The distinction between the two-if it can be
maintained at all-will not always be sharp.
The problem of discovering the mistakes of, and errors within,
an organization is of extraordinary practical significance.
Othezw~tise it would not be possible to tell how the optimum posi-
tion can be rea~ (assuming its characteristics to be knownl)
It is not enough to know that an organization is not at its opti-
mum; it is necessary to be able to indicate where it is and how
one can get to the optimum or at least move towards it. Apure
trial-and-error procedure is an empirical impossibility when many
variables are involved.l Numerical estimation of the distance from
the optimum is also required because otherwise the significane?
of an error or mistake is impossible to determine. To visualize
what is demanded, it suffices to think of a university: What is
its optimum organization? Its aim? How can the aim be formulated
quantitatively? What mistakes does the administration make?
How far do they keep the university removed from its optimum
position? What corrective measures should be introduced? The
1 This is significant for the so-called "theory of the firm" where
all of these things have been neglected.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
-6~+--~
patent impossibility of obtaining anything but the vaguest
answers to these questions~rhich are typical for most admini$-
..
trative set-ups--is an illustration of?c?the~~gra:ve consequences
of the total lack of theory in this field.
It is not better in business. For example, the chairman of
a large department store---~. very able man--$ave this answer to
the question: In xhat perrcentage of cases would a customer who
entered the store and asked for a commodity carried by the store
be told "temporarily out of stock"? "Approximately in 2 - 3 per
cent of the cases." A statistical investigation had put this
figure at 15 - 1? per cent: This enormous divergence of infor-
mation about a vital aspect of the operation of the business shows
? what poor conceptions often prevail about the "optimum." Clearly,
the percentage cannot be zero. But is 15 per cent bad or quite
acceptable? Without any deeper knowledge about mistakes, almost
any answer is as good as any other. A department store sells
many thousands of different items, none of which is decisive for
the outcome of its operations. If the stock is not available,
only a small mistake is made. It may not even be a mistake at
all because it would cost far too much to be better stocked. The
? control system-may become exponentially more expensive for each
reduction of the percentage. Similarly, there is not much of a
disappointment to the customer, because a street block away there
will be another store having the commodity in stock. Or there
may be a close substitute of which even the department store may
carry an ample supply. From the point of view of the economy as
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
R1K-73 ~
-65-
?
a whole, an entirely negligible event has occurred at any rate,
whatever the final outcome for the customer. This ib: clearly
a situation where many small mistakes can be tolerated (if they
are such; ef. the reference to the costs o.f not being in error).
In the terminology used above there is not a mistake here, but
an error. The reason is that "being out of stock" is attribu-
table to the des~n of the organization. The incorporation of
a technical advance lt~ding to a more ridig control system, or
to a system of faster signaling, etc., make it possible to reduce
the percentage, but then there is the technological progress which
we are now not considering. It is also possiblc to treat the
occurrence statistically in the sense that probability distri-
butions, rather than individual events, are considered and that
the correct distribution was chosen, so that no design--mistake was
made.
A very different case is offered by, say, a naval supply
depot that has to provide ammunition, etc., to warships. If
ammunition for i6 inch guns is demanded by a battleship and the
depot does not have them in stock, there is no substitute ammuni-
tion. The same applies to many other items, such as highly
specialized radar gear, electronic tubes, oils, etc. Further-
more, there is no other naval base around the corner which might
be better stocked. The lose for the navy may mean the loss of a
battle or a 'oampa~ign and might be valued at an infinite amount.
It is clear that this contributes to a different inventory policy,
to an "overstocking" in a commercial sense. In fact, there is
no comparison possible, because--as far as these items are
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
1~-73~+
-b6-
concerned the naval base is a different kind of organizat~.on
since it approaches more the type of a digital computer which
operates on an either-or basis and tolerates no mistakes whatever.
We do not say that the base as a whole should be so arranged.
It may be faulty to operate the entire base like a digital com-
puter. Large sectors of the naval base are constructed comparable
to department stores where wide substitutions are possible; e.g.,
when the battleship wishes to get spinach for its provisions,
she might also take broccoli or peas if the base is not supplied
with the desired vegetables. All this becomes of significance
for drawing up a program of logistics where the possibilities of
substitution do not appear to be utilized enough (cf. Part N).
The significance of a mistake in organizations depends also
on the possibility of recovery. As a rule it will be easier to
adjust to many small errors rather than to a large mistake..
especially when the small errors are quickly recognized and sig-
n sled as such is counteraction effective. The noise factor
enters here, too, since false signals of mistakes and signals of
mistakes when none were made may lead to false counteraction.
The human body provides checks in its neural system which make
certain that the body will not be caused to act easily upon the
basis of false signals. Yet when action is called for it comes
as gai;;kly and decisively as if it were an on--off system. The
eye is a prime illustration.l Small errors in social, organiza-
tions can be recognized and reported quickly. This can be done
1 Cf. the work of Selig_ Hecht, about the eye, where it is shown
that 6 - 7 rods have to be excited before a light sensation is
reported to the brain. Smaller numbers are not reported because
it happens frequently that single rods will in error go on and
the brain discounts tku se errors by not reacting.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~+
-~7-
by people who are not, or little, responsible for their occur-
rence; e.g., the salesmen in the department store are not res-
ponsible for inadequacy of the inventory. Therefore, feelings
and pride are not much involved and the remedial counteraction
does not create many difficulties. Anyth~.ng pertaining strictly
to statistical error falls into this class and is, therefore,
o f the more harmless type. Large mistakes, embodied mostly in
the structural features of the organization, are difficult to
discover and still more difficult to remedy. If they are brought
to light they affect many people and probably highly placed ones.
High administrators are notoriously unable to accept criticism;
in that they are quite different from scientists who are con-
stantly exposed to such challenge. Large mistakes occur probably
? mostly with organizations where no real effort has been made to
state the aim, even qualitatively, clearly-let alone numerically.
The mistakes are difficult to determine, cannot be stated uncon-
troversially, and may therefore go unnoticed for a long time.
Common sense is not a reliable guide for finding and remedying
them. Many things in large organizations appear outright foolish
to an outsider, but the disturbance of these practices may have
grave consequences because the practices may not be faulty. The
"obvious" cumbersomness of many government offices and their
procedure invites quick criticism, Pi;equently from businessmen
who, whatever they do, deal with vastly smaller quantities and
cannot fudge from their experience what the best procedure would
be. It appears as if it were simple to find out what other pro-
cedures are at least equivalent, if not patently superior. In
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-68-
? most cases opinions of this kind are of little value because
of the lack of a theory covering such complicated phenomena
(quite apart from the prevalence of purely political prejudices).
The small errors described above do, however, have a
larger aspect that must be well understood. The fact that a
department store cannot and need not be 100 per cent stocked
for all cases makes it necessary to evaluate the statistical
situation and to determine and various probability distributions
involved. The appraisal of this statistical situation can be
correct or it can be fundamentally wrong: Here is the field for
the large mistake visa vis-~a. statistically determined set-~zp.
-rue ~mais a.naividual errors ar`e aCOeptable it they occur wi~hin~
r----- _ _. _ - - -- - -_ _ ..~ - - - - --- - ~ - ? _ - -- --~ - - - - --- - . .
the framework of a correct appraisal of their probability, and
the estimation of their influence upon the operation of the entire
organization. I believe that the distinction between "errors"
and "mistakes" is real and of importance, even though at present
not yet followed up into all ramifications.
Recovery, closely related to the notions of error and mis-
take, is a profoundly important notion which plays a great role
in the whole range from biology to military strategy. For a
centrally directed social organization there are two basic eon-
d itions when the phenomenon may arise: (1) Recovery from the
eftects~of own error and (2) recovery Prom outside shock and
pressure. {
(1) is incorporated into the set-up of the organization,and
recovery is possible if the distribution of errors is properly
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
estimated; i.e., an estimate has been made of the feasibility
of avoiding error and the impossibility oP achieving the aim of
the organization more closely (to the extent that this is depen-
dent on the organization's own acts). For example, one loses
by `being out of stock but can absorb such losses to a certain
~-
extent. `~P instead of errors, mistakes are involved, the situa-
tion is more complex. It is again a quantitative problem: If
a wrong probability distribution, in the above case, is~?chogen;
recovery may very well be possible; if a mistake of the nature of
placing an incompetent man in charge occurs, recovery may never
be possible.
(2) Here is a problem about which nothing appears to be
known, except in the moat general sense, for individual social
organizations. This is undoubtedly also a quantitative problem-
it would require measuring the limiting extent or amount of
damage done from which recovery can still be expected. But
ef. Sect. 9 about the restricted value of the notion of costs2
Even the qualitative description is quite obscure and will have
to await many studies, the nature of which cannot be easily
specified. It is likely that highly specialized organizations,
dust as such biological organisms, are most vulnerable. Speciali-
zation can refer either to the detailed structure or to the lack
of flexibility in operations, which we know not to be necessarily
the same. A good illustration for the second kind of specializa-
tion is a certain tree-living bear which can eat only leaves of
one of two kinds of Eucalyptus trees. This animal would be highly
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~
-70-
? vulnerable to changes in its environment. The swine and the
rat, on the other hand, have a most diversified food intake. So
has the human being, thanks to his techno~:ogical knowledge. He
is also little specialized---as was often pointed out-being
bettered in virtually every field, by some animal. But in the
combination of all of his abilities he is more adaptable than any
other living being. Similar conditions may be found in indivi-
dual social organizations, some of which perish after they have
become petrified, i.e., unable to adapt themselves to changing
environment. Others survive even mutilationss Transfer of vari-
ables and even creation of a new Source are not infrequent but
perhaps more often encountered not in centrally directed, but
;i'f~ the more complex composite organizations (ef. Sect. II/10) with
? which we are not dealing now.
Machines differ usually from living organisms in their
inability to recover from shock. This is true of a motor, for
example, that cannot adapt itself to many detrimental changes of
its environment. However, the difference must not be carried tao
far since it is possible to build a high degree of adaptability
into a machine. A gyroscope, controlling the altitude and direc-
tion of flight, will bring an object back to the required posi-
tion if it has been dislocated but not "too much"may adverse
air currents. But an automobile motor, once beginning to run
down, has no recuperative powers whatsoever. It is simpler to
have a mechanic make repairs than to build such devices into the
? machine itself, even if devices can be imagined.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
xM 73~+
71-
?
The phenomenon of recovery is so obviously beyond the
scope of these notes that it has been sufficient to point
out that an understanding of this field would be immensely sig-
nificant. But it appears to be inaccessible to anything but
common-lace remarks. (About the relation to learning, cf.
Sect. 8).
6. Controls within the Organization
Any organization embodies controlling devices, however
rudimentary and hidden they may be. Passive control consists
of the comparison, for each competence,?of duty and the choice
actually made for the variables upon receipt of sign8.ls. The
comparison proceeds as a rule at least on two levels: (a) the
control sees whether the signal has been received, identified,
and responded to (within a time limit) by the competence that is
in charge of the variable. It. compares whether the right vari-
a tiles have been used considering the state of information at that
time; (b) the control may also compare the value of the variable
actually selected with the hypothetical response that "should
have" been made. The correct response could be contained in a
catalogue listing all possible situations. It could, indeed,
be looked up by a machine, which might perform the whole control
function and might only be a machine within a machine, a servo--
.
mechanism. The control organ, functioning at a later time than
when the action took place, will often be in possession of more
information than the competence at the time of action. In that
respect control is conceptually always separated from action. It
offers no guarantee that it would have done better than the
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM 73~+
_72_
controlled competence at the lower level of information. There-
fore, the control organ is not a superior substitute for the
controlled.
When the aim of the organization is not numerically expressed,
i.e., when there exists only a vaguely described payoff, the pro-
cess of control of type (b) becomes vague, if not entirely impos-
sible. It is then when "judgment," "experience," and "wisdom"
enter; all are euphemisms for the fact that one does not have
scientific knowledge how to cope with these situations.
Active control consists of the interference with a choice
due to be made by a competence by specifying a new domain for
the variable or even making it a constant. It is based on the
ability of the Source to reassume and redelegate any delegated
variable. Such an occurrence may be based upon the results from
passive control of past acts and/or upon operational information
that would not become available in time to the competence under
the existing signaling system. Therefore, active control may
involve the reassignment of signals, while leaving the hierarchy
unaffected. This will, in particular, be of importance in orga-
nizations where separate signaling systems, i.e., a suborganiza-
Lion, exists. Active control is part of the ~ neral executive,
or officer, as distinguished from the special who is connected
with separate variables---~ience, the institution of a general
officer who can override a~ competence$ under him, while a
(special) officer-can exercise authority only within the delegation
set~zp where the authority is transitive. For a general officer
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
xM--73~+
73-
?
the authority is also lateral (this is further discussed in
Sect. 10).
The control may be set up as a separate ays tem of compe-
tences. The previous observations about hierarchy, delegation,
signaling, etc., apply again. It is an organization within an
organization. The control system, however, does not need a
memory separate from that of the basic organization, nor any
separate signaling system. These latter two are then found to
be universal in that they are not necessarily related to separate,
distinguishable sub-organizations within the main organization.
This ought to be of considerable interest when fudging the appro-
priateness of organizational structures. We shall, however, not
discuss this point any further.
The control ~.s primarily concerned with the internal func-
tioning of the organization and has there its easier task. It
also checks on the outer operations which are more difficult to
check because the records are, as a rule, one--sided; i.e., inner
operations consist frequently of transactions involving eompeten-
ces in succession, or have other clear internal interdependen-
cies, so that the behavior~of a given competence can be checked
not only on the basis of its own records, but even by that of
another competence. For outside contacts this is generally
impossible. The internal records are available for such pur-
poses, but, even if complete, tell only part of the story. Fur-
thermore, a control of the outer operations would require a very
complete knowledge of the strategic situation of the organization
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-?3~
-? ~-
as a whole and this evaluation is hardly a routine operation.
It requires far more and is a function to be fulfilled by the
Source. Routing control of the moves made by competences is
possible when these oan be compared with expected, routine moves
which are repetitious in recurrent situations. All~eontrol
functions aim at holding the organization together and improving
it. Since a control is indispensable in any organization, even
where "efficiency" is not an overt aim, this indicates that in
each existing organization there is also always a tendency towards
disintegration. This is similar to the process of entropy in
thermod
ynamics. Organization in itself, wherever it occurs,
whether in social or natural life, is something exceptional and
? extraordinary. The natural state is one of lower order, if not
of disorder. Consequently, an organization will not tend to stay
in its state of order but rather tend to fall to pieces. It is
therefore necessary for each perpetuating organization to provide
for tendencies to oppose such trends. The essentially internal
control of an organization is to accomplish precisely this. What
the causes for this trend towards deterioration of social organi-
zations are can only be surmised in broadest terms. A future
theory of organization would have to make the explanation of this
basic phenomenon one of its primary objectives. The entropy of
organizations differs widely but it can overtake even the appar-
ently best-established ones. Some eudden and surprising collapses
of formerly mighty armies tell important stories. Some systems
? of organizations appear to come under particular stress of this
kind.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
? 7. Degrees of Organization
RM-73~}
-7 ~
In~several preceding passages an implicit use was made of
concept such as "over-organization," "higher" and "lower" forms
of organization, etc. It would be desirable to make these terms
preoise. However, it seems impossible to -achieve this at the
present state of our knowledge, although intuit~.vely these notions
seem quite accessible. However, they involve some idea of an
"optimum" or "equilibrium" which would_bei~hard to formulate
except in a completely tautological manner. It is quite likely
that the notion of the equivibrium of an organization has to be
changed and broadened as in the theory of games which deals with
composite social organizations. There the characterization of
an equilibrium by a single number or set of numbers had to be
_~ replaced by new concepts.
? Intuitively, one would speak of "over-or -T"-_~.__~, ,__~,
`~ 8snization" when
"unnecessary" formal steps are provided, when the domain of vari-
ables is restricted, etc., while "it is clear" that the choices
could be left to lower competences without their referring back to
higher ones, when "too many" competences are created (i.e., too
-~- - ~ ~ few? v~ria-b-l-ea--concentrated in one) , or when internal controls --
'?~~ dare provided 'while the aeleotive ?'~?~'-''~~ ~ ~ - ~~
., ,~ ~ w :J poxer of comPettncea shou_ 1Q not
~! .eV ~
be interfered with, etc. When there is a tendency towards more
constants in an organization, it would appear to coincide with a
"higher" form. But~on the other hand, an organization will be
better able to cope with its problems when it preserves more
? rather than less freedom to choose. This requires that its
resources be held iri a form that does not commit it to only one
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~
-76-
single course of action. But such a view contemplates alterna-
tive actions spread over a lengthy part of time--s further com-
plication. For a sharply defined aim, to be achieved in the
immediate future, the proper organization is the one that fits
precisely the needs of the situation; e.g., a fully automatic
conveyor system is of this kind (for the inner operation only:-
but let us assume it to be optimal for the outer, as it should
be). This complete specification can be useless if a substantial
change occurs in the purpose (e.g., in this case even vrhen con-
siderably fester units of the same things are to be produced--e.
?
very different set-up might then be optimal
A low state of organization is one in which the resources
are relatively amorphous for the purpose, i.e., the strategy.
It is thus impossible to have an acceptable measure of the degree
of organizationYor disorganization `f'or a strictly static easel)
without consideration of optimality of operation for a concrete
situation involving only the immediate future. A low form may
allow the transformation of the resources into a more specific
set-up; disorganization, on the other hand, may well consist of
the rupture of important functions of an otherwise highly diversi-
fied set-up equivalent to its inability to transform further
homogeneous resources into the required specifications for moves
in the play of the game.
It is noteworthy that in this language the " over" of an
organization-~rhich is the freedom to choose from a wide range
of possible strategies and corresponding inner operations-requires
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
xM-73~+
77-
?
this interpretation: (a) Por a fixed situation, it manifests
itself in having made the optimal choice, from the homogeneous
form in which the resources were originally given by committing
them to a particular form; (b) for a situation which is still
flexible, it manifests itself in availability of a greater amount
of resources that can still be committed, where at least part
of these resources are in a homogeneous (i.e., not specified)
form and application. Thus, for example, an organization can be
powerful, but at a low state of organization, etc.
To interpret: (a) means that the organization, in antici-
pation of a concrete situation, has used its resources in an
optimum form. Comparison with other organizations--}possibly hypo-
thetical--that commit different amounts of resources gives a
measure as L-o how wide the range of strategies is for them. In
such a scale, the place of relative "power" of the organization in~~
question could be found. Such a scale would be exceedingly dif-
ficult to construct because of the formidable task of computing
the optimum strategy. (b) means that the resources are still
held in an uncommitted form, as, for example, the capital of a
firm being available in cash before any purchases of machines
and raw material have been-Made. In this condition it is possible
t o compare strength or prn~er in a simple manner (assuming that a
p articular situation is to be faced}. In the economic illustra-
tion, the liquid form of the capital resources is a sound notion,
because we know approximately from the prevailing prices and
our technical knowledge how this can be transformed into specific'
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~+
-78-
goods and services.l But for non-economic organizations it is
not a priori clear whether resources not yet brought into specific
form have quite the same tangible existence, let alone can be
given numerical expression as in the ease of cash which can be
compared with prices. Thus the civilians available for an army-
even if they have been trained before-are something different
from the soldiers they are xhen fitted out and arranged in com-
bat form. This observation applies, of course, particularly to
the organic field where the materials from which a living body
is made give no indication of the strength and power of the
organism into which they will be built unless the organizing prin-
ciple is also formulated and applied. In the economic case such
a principle is given in the technological knowledge being com-
bined with the purpose of obtaining a maximum (of profit) by
choosing a strategy in a fairly well-known milieu within which
action can take place. However, the question how a lesser degree
of organization is transformed into a higher one~far transcends
the possibilities of this discussion.
8. Learning
The understanding of the phenomenon of learning will be of
primary importance for any ulterior theory of organization. Here
only a brief sketch can be provided, in view of the formidable
difficulties in principle and especially at the present stage.
Learning has been extensively studied in psychology, biology,
1 If the firm attempts to do this it enters upon a game, so that
the statement in the text is not absolutely correct. But it
is not necessary to go itiirthe r into these complications.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-79-
? physiology, etc., but it is fair to say that even there, where
so much ingenuity has been applied, the understanding is still
highly incomplete. The problem has appeared also in the dis-
cussions about the construction of machines, such as computing
machines or those that play games with human opponents, where it
xould be important to understand whether these-machines could be
s o designed that they learn from the opponents' mistakes. Learning
involves complicated feedbacks as well as the assistance of, and
access to, a memory.
It is a riori clear that learning occurs also in social
organizations, whether individual or composite. Some are knoKn
to adapt themselves to changing conditions, to emerge from adverse
influences, and to survive under the greatest odds even over
centuries. The universities are a classical illustration: TheYe
the same basic things have been going on while the outer world
changed-=--transmission of knowledge, training of new scholars,
and new contributions to knowledge. The learning may have con-
sisted of the ability to adapt to variations in the milieu in
order to do essentially the same thing over time. In other
cases, learning may appear to consist of doing something diPfer-
ent (i.e., have changing aims) in an essentially stable milieu.
Sometimes both efforts are intermixed. It remains to be seen
in what sense one is justified in speaking of "learning" in all
these cases.
It is perhaps--if this may be said in parenthesis-not very
surprising that @f a,ll organization, universities ought to be
able to survive well since they--ire are told-a collection of
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM~-73~
-80-
the finest brains of each age: Even if universities have not
developed a theory of organization and survival that they oan
apply consciously, they should easily find the proper adapta-
tion processes. So one should really not be too much surprised
that they have survived over hundreds of years, but rather
wonder why they have not done far better for themaelves.~
In those fields where learning haae been studied it tip:, inevi-
tably found to be a matter of feedbaelca. Certain activities are
carried on by the sub~ects,and t;~e results are observed by them
and their superiors in the light oP the situation as they se~'it.
The feedback.--~.f successful, i.e., accepted-operates auah that the
new operation constitutes an improvement over the older in the
same, or nearly same, circumstances. Each operation produces new
information which, by means of the memory, is compared with infor-
mation already in possession of the subject. It appears that
wherever learning has been studied, the aim or function has always
been quite clear to the subject and the investigation. If neither
were the case, the observation of the learning process and of its
success would-=undoubtedly, be far more difficult, if possible at all.
The subject evaluating the experience and arranging for the
feedback to_become effective in a learning sense (i.e., as we
shall see, by--a change in the structure or in the code of the
organization) may err and instead of a positive, we get a negative
effect. We recall that the aims of social organizations, even
when centrally directed, are often obscure or only formulated
1 It is doubtful that universities in general are a good illustra-
tion because the true institutions are not centrally directed
but fall into the category of composite, cooperative organiza-
tions. This-does, however, not apply to those American univer-
sities which=archaically-a.re still run (organizationally) like
business corporations.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
? in the vaguest generalities. This leaves, therefore, consider-
able leeway for the possibility of "negative" learning.l
Assuming the aim to be clearly understood, we now discuss
briefly.a distinction that may serve to give a first outline:
_ _ _ _ _ _. ._ ~:, _ . _ j
(e-.) Learning for tYie Qatar aperatiaas TLearning arises only
if there is incomplete information and/or a deficiency {e.g.,
physical) in attaining the aim, and if the physical possibility
is actually given to adapt the organization. Under these cir-
cumstances an operation produces information that is used in~
determining the next operation when faced with the same or nearly
same situation in which action is required. Now in our discus-
sions so far we have, unless specifically stated otherwise, always
assumed complete information, as this term is understood in the
Theory of Games, where naturally the problem of learning, i.e.,
discovering the opponent's intentions and style, also ariaes.2
The assumption of complete information precludes any learning by
an organization for which a clearly defined aim is described. It
will always be at an optimum, although successively over time
there may be very different outer operations (i.e., strategies)
_~ ~w ., +`
._
1
al conditions wi
i
h
_
~ r.
c
ys
~- because of txo factss (1 ). The eoromon p
~~ change, bringing about as the moat notable consequence a~change
in the other players' strategies. (2) The strategies available
1 "Negative" learning is a different phenomenon from the "entropy"
of organizations, i.e., their tendency to go over to lower forms
of organizational setup or efficiency.
2 Cf. loc. cit. There it is shown that the problem really does
not ea-ist~' or the theory because of the equivalence of the des-
cription of a game in its extensive and its normalized form.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RK--73~
-82-
?
in the second period will depend on the changes in the resources
of the organization due to the success or failure of the strategy
used in the first period. For both these reasons there is a
repeated "adaptation," but in this case it does not represent
a learning process in the above sense, because there were no
deficiencies and there is not the need to meet requirements posed
by the same situation. If the same conditions existed and a
non-optimal strategy had been used, followed in such manner by
others that finally the optimal one is reached, then one could
speak of learning in the correct sense. But if optimal strategies
sucoeed,each otrer,`no learning takes place.
'Th#e~~ob8ervation extends not only to strategies but to
the moves of which they are composed. In other words, no rear-
rangement of moves is possible without thereby altering the
strategy; this follows from the equivalence mentioned in footnote 1,
page 81. Since each optimal strategy guarantees only the maximum
protection in non strictly determined games, which are the rule,
it may by remote chance be that in successive situations dif-
ferent optimal statistical strategies yield successively increas-
ing payoffs. This might easily give the impression of a suc-
cessful "learning" when in fact it has nothing to do with it.
An entirely different situation arises when incomplete
information exists which is, of course, the empirically more
realistic case. Then learning will occur even if the situations
encountered by the organization should be strictly identical with
each other, provided the first strategy was only an approxima-
Lion to -the optimal one. It~ris4Y,~ for exaaaple, bi.. toEo d"ilfi,au~~$ ~ ~~`
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
AK-73~
-83-
? to compute the optimal strategy. Thus, the first one used gives
a vast amount of information that can be profitable for finding,
or approaching, the optimal one. The information would become
available to the organization by comparison of the outer opera-
tion by means of the memory with the theoretical characteristics
of the optimum strategy and lead to the incorporation of the
lesson into the next outer operation.l This involves feedbacks
which may even have to be provided for specifically. This is
learning by mistake. In order to function, the mistake must be
recognized as such, which shows the importance of that notion
as discussed earlier. The feedbacks will not be automatic by-
products of an organizational set-up; they have to be specifically
provided for. It is here where interesting and important empiri-
cal studies will have to be made. The practical arrangements
vary from the anonymous "suggestion boxes" to the calling in of
management experts. These techniques will all become gradually
super$eded by the development of Qperation~s analysis which, so Par,
is applied mostly to the military field in which it has arisen.
Industry and business are lagging behind the military; for that
there are deep psychological and social rbasone.2 Hor-~ever,
1 We neglect to mention, however, that a similar process must be
going on in the other organizations against which the strategy
is to be used, so that there is no such stability. But we shall
not complicate matters any more at this ~unett~re.
2 The lag is interesting, because management control was developed
by business. But operations analysis goes far deeper and by
virtue of its wider scope endangers some of the prerogatives
of business executives. There is only a small and scattered
literature on Operations Analysis. Cf. P. M. Morse and
J. Kimball, Operations Research (1950) and a recent pamphlet
by the Nations esearc ounc
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM--73 ~
-8~--
? operations analysis applies more frequently to the inner opera-
tions, while we are at present still discussing the outer. Even
before the rise of this new applied science, the observation
and planned study of onus own operations are frequent, especially
in the military organizations. There historical offices exist
charged with writing the histories of campaigns, wars, etc.
These histories do not serve in the first order some vainglorious
desires, but serve to understand the past so that lessons can be
learned. -This is a systematically arranged feedback which works
via the military academies, laboratories, and proving grounds.
Zt is most interesting to note that in that respect too business
is lacking in interest and execution. It may be due to the lack
of resources (human and monetary), to the short 11fe of the aver-
age firm, but it must also be related to a lack of understanding
of the issue involved. It suffices to compare the type of stan-
dard company-anade business"histories" (mostly more expensive
forms of advertisements of no analytical value) with the often
highly scientific and critical character of military history as
written by the military themselves. Business is also suffering
from the desire to preserve secrecy. An example is offered by
the Bank of England about whose operations 50 or even 100 years
ago the full truth cannot be established because the records are
nod completely released, although they exist, and all concerned
are dead .
3o we see: Learning in regard to the outer operations con-
,
sists of approximating the optimum strategy for a given situation
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~
-85-
? and to achieve the same for successively varying situations.
This is accompanied by the endeavor to discover the intentions of
those organizations with which cooperative or antagonistic con-
tact is made; she process by which this learning is accomplished
must involve the inner operation which is to be optimalized in
respect to an outer operation.
(b) Learning for the inner operation: When the inner
operation is optimal with regard to the outer irrespective of
whether the outer operation is optimal or not--x~o learning pro-
cess will be found. If, as a consequence of the variations in
(optimal) strategies, there are equally optimal variations in the
inner operations, no learning or adaptation is involved. This
follows from the remarks under (a). So the interesting ease is
? again that of incomplete information, this time concerning inner
operations. We recall that the structure of the organization
embodies probabilistic elements and that the organization cannot
be informed other than probabilistically about its inner opera-
tion and its whole status. Learning does not necessarily take
place but these are conditions under which it can occur. Whether
it actually does happen will depend first on the existence of
the necessary feedbacks and second on their utilization.l Neither
.~~? let. ~ta~ Vie. e~pe~ted a *priori~? _ ~ _ ~ ~~ j
'~ ~ - _-._~ , ~ ' -- ~ _ 31noe~ we are discussing `oal the lortetal -e`ids o!'~he ei~tion ~ ~- T-~-
(b) shall not be considered in detail. Learning for inner
1 We may, of course, asaumc that the organization has the desire
to learn. This would be implied by the customary assumption
that a maximum of achievement is to be attained. About some
empirical obstacles to learning, cf. the last paragraphs of
this section. -
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73 ~+
-86-
? operations will consist of those changes in arrangement and
utilization of the resources of the organization that are charac-
terized by time and motion studies, eliminating unnecessary
steps, determining sequences of acts, speeds of conveyors,
routing of papers, etc. zn all this there is a wide field where
"experience" is important, which is only another word for the
feedbacks at work. It would require a very detailed knowledge
of particular cases before any generalizations are possible
about the origin and direction of the feedbacks that are built
into an organization whose inner operations are not at optimum,
It is clearly impossible to push f~,irther without making concrete
assumptions about the form, complexity, and material content
and purpose of the organization.l
Because of the combination of the two facts that learning
is, to a high extent, absorbed through reorganization and that
each organization has its own individual historical experiences
causing learning processes, a,peculiar problem arises: How is the
identity of the organization preserved? Is not, after a long
chain of such transformations, the organization as actually
observed something widely different from the original one? The
name may survive or even not; at any event this would not be con-
sidered as material by anyone. The question of identity over
time will arise only for long periods or when the learning is
l This is precisely what operations analysis tries to accomplish.
There is no general theory there, as yet. There are only rules
of procedure, so far, and the attempt to go about the task in a
scientific spirit as well as equipped with the moat modern tools
that statistics and scientific analysis can provide, It is pro-
bable that they-e will not be any science in this field without
a (simultaneous) far superior knowledge about organizations
than we have nox.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
814-73~+
--87-
? exceedingly rapid, involving many mayor changes; this can be
considered as dynamical ands therefore, beyond the scope of these
notes. However, 1t is interesting to observe that the same
occurrence is true for the human body= Cells are being replaced
almost continuously, functions disappear, and new ones are
assumed, etc., so that the question of the identity of the indi-
vidual is a problem that arises sooner or later even in onu s
own consciousness. Some of the changes not only are due to the
natural causes of growth and age, but reflect our experiences
and many conscious efforts to absorb these.
Leaving the formal questions aside, we observe that the
actual learning processes within an organization cause many forces
to appear that make learning easy or difficult, as the case may
? be. The new experience to be incorporated into the structure of
the organization, or into its protocol, or into the operations
themselves, will frequently encounter great opposition. There
are, for example, always competences-i.e., represented by indi-
viduals: that will feel their posit~:on threatened through the
creation of new variables, competences, abolition of others, etc.,
forms in which learning would materialize. There are persons
who are on principle opposed to new things and methods, the old
procedure has become routine, experience does not become known
widely enough, etc. Thedt'?are all significant forces, and their
study will reveal much of sociological value. The results should
then be incorporated into the study of the formal aspects of the
process of learning. Some organizations, being highly adaptable,
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM-73~
-88-
succeed in learning rapidly and often; "adaptability" is,?then,
only another term for learning.
9. Input--0utput and Costa of Operations
In economic organizations such as firms and households
the evaluation of the result of an operation is usually in
terms of costa and profits, obtained through $ome comparisons of
inputs and outputs. Current economic theory uses this pattern,
too. The comparisons would have to be made on two levels: First,
various operations are to be considered which take place within
the same given organization as described by the hierarchical del,e -
gation of variables and the information system; second, for dif-
ferent organization patterns.l If the economic organization is
not a firm, but a cooperative association such as a cartel,
labor union, etc., the cost-profit principle is immediately seen
to be in grave jeopardy. We shall return to this below. But firs t
this observation: The current notions of costs, profits, inputs
and outputs assume that they can be sharply, unambiguously
defined and measured objectively with sufficient accuracy. This
is, however, not the case, except in a conventional sense-"profits"
are often fixed entirely without regard to "costs";2 the,3atber
depend, among other things, on the rate of depreciation of fixed
capital assets which can be varied arbitrarily within wide limits,
?
1 This separation is achieved to some extent (i.e. subject to
the restrictions_discuased at the end of Sect. 1~ in the theory
of the so-walled planning costs and represented by the envelope
curve of variable costs.
2 Cf. 0. Morgenstern, On the Accuracy of Economic Observations
{195O)s pP. 27 32. ?
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
~_ __
RM-734
-89-
etc. The gross receipts of a firm depend not only upon the
physical-financial facts involved, but upon the policy or strategy
used in its dealings with its competitors, customers, and supplje ra.
To the extent to which this is the case, any cost--profit charac-
terization of the operations of a given organization, even in
the economic field, is at least incomplete, if possible at all.
So we see: The economic quantities upon which the notions
of costs and profits are based are not sharply definable. Thus,
the determination of "coats of individual operations" and of
entire organizations is likely to turn oat to be of a highly com-
plicated nature, if it is possible at any level, exo~pt in the
above-mentioned conventional manner which need not-and in fact
does not-satisfy scientific requirements.
Assuming now, however, that cost-profit comparisons as made
in business are a reasonable procedure, this has to be observed:
Bvery social, centrally directed organization has by necessity
an economic aspect. Every one of them consumes or uses-;eEa]cce .
goods, however rudimentary the quantities, and involves some
expenditure. This is true of a church, a hospital, a university,
etc., which in this sense incur "costs." None of these organiza-
tions has a primary busineas~purpoae. Therefore, this input-
output relationship is, if not meaningless, at least subordinate.
So, for example, the_success or failure of a university is not
expressed by a financial Surplus or deficit as a consequence of
a year's operation. ~It would be absurd to use either as a
criterion.
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
RM--73 ~
90-
Thus, if an input-output characterization is nevertheless
desired, it has to be in terms of the aims of these organizations,
and has to use whatever is being processed from a less-organized
state to that of the 'accomplishment the organization strives to
aehieve.l No general treatment is possible,-because there are
as many different kinds of inputs and outputs as there are types
of organizations, the latter distinguished by their aims.
Descriptions of this kind are not easily made or even
envisaged, because of the frequent difficulty of formulating the
aims of organizations, let alone their numerical expression (or
lesser orderings). Thus, it becomes almost impossible to des-
cribe organizations in terms of proper inputs and outputs. For
example, is the input of a church sinners and the output sa~:nta?
Or, at the universities, is the input ignorants and the output
scholars? What is the input in the army? What the output? Cer-
tainly soldiers or weapons are not being manufactured there, but
something far more complicated is to be accomplished. What is
the output of a prisonY Is the input criminals, the output
punishment? Or reform? Etc., etc. Along list could be made.
The fact that all these organizations have economic and
financial aspects and. in many circumstances must behave as if
they were economic institutions (e.g., trying to buy as oheaply
as possible) imposes a cost aspect upon their description and
obliterates the appraisal of their--outer:-operations in those
r
~1 A university, for example, might make absurd expenditures--in
? terms of its aim--arid stabilize its budget at the same time:
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
Declassified in Part - Sanitized Copy Approved for Release 2013/06/11 :CIA-RDP81-010438001800180002-7
... i.`r~'S41`