Anda di halaman 1dari 5

Case 1-2 Ackoff’s Management

Misinformation Systems
This case is adapted from a classic article entitled “Management Misinformation Systems.” It was
written by Russell L. Ackoff and appeared in Management Sciences. In the article, Ackoff identified five
common assumptions about information systems and then explained why he disagreed with them.

REQUIRED: information problem primarily, but not exclu-

sively, as one that arises out of an overabun-
Read the five assumptions, contentions, and
dance of irrelevant information, most of which
Ackoff’s explanation. For each of the five,
was not asked for, then the two most important
decide if you agree or disagree with Ackoff’s
functions of an information system become fil-
contentions. Defend your stand by preparing a
tration (or evaluation) and condensation. The
report to explain your beliefs. Be prepared to
literature on the MIS seldom refers to these
defend your beliefs in class.
functions, let alone considers how to carry
them out.
My experience indicates that most man-
agers receive much more data (if not informa-
Assumption 1. Most management information tion) than they can possibly absorb even if they
systems (MISs) are designed based on the as- spend all of their time trying to do so. Hence
sumption that the critical deficiency under they already suffer from an information over-
which most managers operate is the lack of rele- load. They must spend a great deal of time sepa-
vant information. rating the relevant documents. For example, I
Contention 1. I do not deny that most man- have found that I receive an average of 43 hours
agers lack a good deal of information that they of unsolicited reading material each week. The
should have, but I do deny that this is the solicited material is usually half again this
most important informational deficiency from amount.
which they suffer. It seems to me that they suf- I have seen a daily stock status report that
fer more from an overabundance of irrelevant consists of approximately 600 pages of com-
information. puter printout. The report is circulated daily
This is not a play on words. The conse- across managers’ desks. I’ve also seen requests
quences of changing the emphasis of an MIS for major capital expenditures that come in
from supplying relevant information to elimi- book size, several of which are distributed to
nating irrelevant information is considerable. If managers each week. It is not uncommon for
one is preoccupied with supplying relevant many managers to receive an average of one
information, attention is almost exclusively journal a day or more. One could go on and on.
given to the generation, storage, and retrieval Unless the information overload to which
of information; hence, emphasis is placed on managers are subjected is reduced, any addi-
constructing data banks, coding, indexing, tional information made available by an MIS
updating files, using access languages, and so cannot be expected to be used effectively.
on. The ideal that has emerged from this orien- Even relevant documents have too much
tation is an infinite pool of data into which redundancy. Most documents can be consider-
managers can reach to pull out any information ably condensed without loss of content. My
they want. If, however, one sees the manager’s point here is best made, perhaps, by describing
2 CASE 1-2

briefly an experiment that a few of my col- wise, should be an essential part of an MIS, and
leagues and I conducted on the operations that such a system should be capable of han-
research (OR) literature several years ago. By dling much, if not all, of the unsolicited as well
using a panel of well-known experts, we identi- as solicited information that a manager receives.
fied four OR articles that all members of the
panel considered to be “above average” and
four articles that were considered to be “below
average.” The authors of the eight articles were
asked to prepare “objective” examinations Assumption 2. Most MIS designers “deter-
(duration 30 minutes) plus answers for gradu- mine” what information is needed by asking
ate students who were to be assigned the articles managers what information they would like to
for reading. (The authors were not informed have. This is based on the assumption that
about the experiment.) Then several experi- managers know what information they need
enced writers were asked to reduce each article and want.
to two-thirds and one-third of its original length Contention 2. For a manager to know what in-
only by eliminating words. They also prepared a formation he needs, he must be aware of each
brief abstract of each article. Those who did the type of decision he should (as well as does) make
condensing did not see the examinations to be and he must have an adequate model of each.
given to the students. These conditions are seldom satisfied.
A group of graduate students who had not Most managers have some conception of at
previously read the articles was then selected. least some of the types of decisions they must
Each one was given four articles randomly make. Their conceptions, however, are likely to
selected, each of which was in one of its four ver- be deficient in a very critical way, a way that fol-
sions: 100 percent, 67 percent, 33 percent, or lows from an important principle of scientific
abstract. Each version of each article was read by economy: The less we understand a phenome-
two students. All were given the same examina- non, the more variables we require to explain it.
tions. The average scores on the examinations Hence managers who do not understand the
were compared. phenomena they control play it “safe” and, with
For the above-average articles there was no respect to information, want “everything.” The
significant difference between average test MIS designer, who has even less understanding
scores for the 100 percent, 67 percent, and 33 of the relevant phenomena than the manager,
percent versions, but there was a significant tries to provide even more than everything. She
decrease in average test scores for those who had thereby increases what is already an overload of
read only the abstract. For the below-average irrelevant information.
articles there was no difference in average test For example, market researchers in a major
scores among those who had read the 100 per- oil company once asked their marketing man-
cent, 67 percent, and 33 percent versions, but agers what variables they thought were relevant
there was a significant increase in average test in estimating the sales volume of future service
scores of those who had read only the abstract. stations. Almost 70 variables were identified.
The sample used was obviously too small for The market researchers then added about half
general conclusions, but the results strongly again this many variables and performed a large
indicate the extent to which even good writing multiple linear regression analysis of sales of
can be condensed without loss of information. I existing stations against these variables and
refrain from drawing the obvious conclusions found about 35 to be statistically significant. A
about bad writing. forecasting equation was based on this analysis.
It seems clear that condensation as well as An OR team subsequently constructed a model
filtration, performed mechanically or other- based on only one of these variables, traffic flow,
CASE 1-2 3

which predicted sales better than the 35-variable intuition usually does very badly (e.g., What are
regression equation. The team went on to the correct odds that 2 of 25 people selected at
explain sales at service stations in terms of the random will have their birthdays on the same
customers’ perception of the amount of time day of the year?). For example, very few of the
lost by stopping for service. The relevance of all results obtained by queuing theory, when
but a few of the variables used by the market arrivals and service are probabilistic, are obvious
researchers could be explained by their effect to managers; nor are the results of risk analysis
on such a perception. where the managers’ own subjective estimates of
The moral is simple: One cannot specify probabilities are used.
what information is required for decision mak- The moral: It is necessary to determine how
ing until an explanatory model of the decision well managers can use needed information.
process and the system involved has been con- When, because of the complexity of the deci-
structed and tested. Information systems are sion process, they cannot use it well, they should
subsystems of control systems. They cannot be be provided with either decision rules or perfor-
designed adequately without taking control into mance feedback so that they can identify and
account. Furthermore, whatever else regression learn from their mistakes.
analyses can yield, they cannot yield under-
standing and explanation of phenomena. They
describe and, at best, predict. ASSUMPTION 4: MORE
ASSUMPTION 3: GIVING MANAGERS Assumption 4. The characteristic of most MISs
is that they provide managers with better cur-
rent information about what other managers
Assumption 3. It is frequently assumed that if and their departments are doing. Underlying
managers are provided with the information this provision is the belief that better interde-
they need, they will then have no problem in us- partmental communication enables managers
ing it effectively. to coordinate their decisions more effectively
Contention 3. Operations research (an aca- and hence improves the organization’s overall
demic subject area dealing with the application performance.
of mathematical models and techniques to busi- Contention 4. Not only is this not necessarily
ness decisions) stands to the contrary. so, but it seldom is so. One would hardly expect
Give most managers an initial tableau of a two competing companies to become more co-
typical “real” mathematical programming, operative because the information each acquires
sequencing, or network problem and see how about the other is improved.
close they come to an optimal solution. If their For example, consider the following very
experience and judgment have any value, they much simplified version of a situation I once ran
may not do badly, but they will seldom do very into. The simplification of the case does not af-
well. In most management problems there are fect any of its essential characteristics. A depart-
too many possibilities to expect experience, ment store has two “line” operations: buying and
judgment, or intuition to provide good guesses, selling. Each function is performed by a separate
even with perfect information. department. The Purchasing Department pri-
Furthermore, when several probabilities are marily controls one variable: how much of each
involved in a problem, the unguided mind of item is bought. The Merchandising Department
even a manager has difficulty in aggregating controls the price at which it is sold. Typically,
them in a valid way. We all know many simple the measure of performance applied to the Pur-
problems in probability in which untutored chasing Department was the turnover rate of
4 CASE 1-2

always ordered optimistically. Therefore, using


the same curve, she read over from Q1 to the

upper limit and down to the expected value,
from which she obtained Q2, the quantity she
Q1 actually intended to make available. She did
not intend to pay for the merchandising man-
Q2 ager’s optimism. If merchandising ran out of
stock, it was not her worry. Now the merchan-
Q3 Op
dising manager was informed about what the
timis purchasing manager had done, so he adjusted
his price to P2. The purchasing manager in
c ted turn was told that the merchandising manager
had made this readjustment, so she planned to
istic make only Q3 available. If this process (made
P1 P2 P3 Price
possible only by perfect communication
between departments) had been allowed to
Figure 1 continue, nothing would have been bought
and nothing would have been sold. This out-
come was avoided by prohibiting communica-
inventory. The measure applied to the Mer- tion between the two departments and forcing
chandising Department was gross sales; this de- each to guess what the other was doing.
partment sought to maximize the number of I have obviously caricatured the situation in
items sold times their price. order to make the point clear: When organiza-
Now by examining a single item, let us con- tional units have inappropriate measures of per-
sider what happens in this system. The mer- formance that put them in conflict with each
chandising manager, using his knowledge of other, as is often the case, communication
competition and consumption, set a price that between them may hurt organizational perfor-
he judged would maximize gross sales. In doing mance, not help it. Organizational structure
so, he utilized price-demand curves for each and performance measurement must be taken
type of item. For each price the curves show the into account before opening the floodgates and
expected sales and values on an upper and permitting the free flow of information between
lower confidence band as well (see Figure 1). parts of the organization.
When instructing the Purchasing Department
about how many items to make available, the
merchandising manager quite naturally used ASSUMPTION 5: MANAGERS NEED ONLY
the value on the upper confidence curve. This
minimized the chances of his running short,
which, if it occurred, would hurt his perfor- Assumption 5. A manager does not have to un-
mance. It also maximized the chances of being derstand how an information system works, only
overstocked, but this was not his concern, only how to use it.
the purchasing manager’s. Say, therefore, that Contention 5. Managers must understand
the merchandising manager initially selected their MIS or they are handicapped and cannot
price P1 and requested that amount Q1 be made properly operate and control their company.
available by the Purchasing Department. Most MIS designers seek to make their sys-
In this company the purchasing manager tems as innocuous and unobtrusive as possible
also had access to the price-demand curves. to managers, lest they become frightened. The
She knew that the merchandising manager designers try to provide managers with very easy
CASE 1-2 5

access to the system and assure them that they provided. With the help of the system designer
need to know nothing more about it. The and volumes of old daily reports I began to plot
designers usually succeed in keeping managers the stock level of the first listed item over time.
ignorant in this regard. This leaves managers When this item reached the maximum “allow-
unable to evaluate the MIS as a whole. It often able” stock level, it had been reordered. The sys-
makes them afraid to even try to do so, lest they tem designer was surprised and said that by
display their ignorance publicly. In failing to sheer “luck” I had found one of the few errors
evaluate their MIS, managers delegate much of made by the system. Continued plotting showed
the control of the organization to the system’s that because of repeated premature reordering
designers and operators—who may have many the item had never gone much below the maxi-
virtues, but managerial competence is seldom mum stock level. Clearly, the program was con-
among them. fusing the maximum allowable stock level and
Let me cite a case in point. A chairman of the reorder point. This turned out to be the
the board of a midsize company asked for help case in more than half of the items on the list.
on the following problem. One of his larger Next I asked if they had many paired parts,
(decentralized) divisions had installed a com- ones that were only used with each other, for
puterized production inventory control and example, matched nuts and bolts. They had
manufacturing manager information system many. A list was produced and we began check-
about a year earlier. It had acquired about $2 ing the previous day’s withdrawals. For more
million worth of equipment to do so. The board than half of the pairs the differences in the
chairman had just received a request from the numbers recorded as withdrawn were very large.
division for permission to replace the original No explanation was provided.
equipment with newly announced equipment Before the day was out it was possible to
that would cost several times the original show by some quick and dirty calculations that
amount. An extensive “justification” for so the new computerized system was costing the
doing was provided with the request. The chair- company almost $150,000 per month more than
man wanted to know whether the request was the hand system that it had replaced, most of
justified. He admitted to complete incompe- this in excess inventories.
tence in this connection. The recommendation was that the system
A meeting was arranged at the division, at be redesigned as quickly as possible and that the
which I was subjected to an extended and new equipment not be authorized for the time
detailed briefing. The system was large but rela- being.
tively simple. At the heart of it was a reorder The questions asked of the system had been
point for each item and a maximum allowable obvious and simple ones. Managers should have
stock level. Reorder quantities took lead time as been able to ask them, but—and this is the
well as the allowable maximum into account. point—they felt themselves incompetent to do
The computer kept track of stock, ordered so. They would not have allowed a hand-operated
items when required, and generated numerous system to get so far out of their control.
reports on both the state of the system it con- No MIS should ever be installed unless the
trolled and its own “actions.” managers for whom it is intended are trained to
When the briefing was over, I was asked if I evaluate and hence control it rather than be
had any questions. I did. First I asked if, when controlled by it.
the system had been installed, there had been
Source: Reprinted by permission of Russell L. Ackoff,
many parts whose stock level exceeded the max- “Management Misinformation Systems,” Management
imum amount possible under the new system. I Sciences 14, no. 4 (December 1967). The Institute of
was told there were many. I asked for a list of Management Sciences, 290 Westminster Street,
about 30 and for some graph paper. Both were Providence, R.I. 02903.