Af/xi flL
press
REVIEW
The following articles are reprinted solely as items of interest for the independent evaluation by members of
ATSU. The opinions, statements of fact, and conclusions expressed herein are not those of the Association.
AN EXCERPT FROM ....
All About Time-Sharing and
Remote Computing Services
Surveying the Users
To evaluate the current level of user satisfaction with
specific vendors of remote computing services and with
remote computing techniques in general, Datapro
Research Corporation, in conjunction with the Associa¬
tion of Time-Sharing Users, Inc. (ATSU), designed and
conducted an extensive user survey. Reader survey forms
on Remote Computing Services were mailed to our
subscribers with the February 1976 supplement to
DATAPRO 70, and were also included in ATSU’s
February Newsletter to its members and associates. (For
more details on ATSU, please address all inquiries to the
association’s headquarters address: 210 Fifth Avenue,
New York, NY 10010.)
Qualified responses were received from 475 users of
remote computing services in the United States and
Canada. Many users commented upon their experiences
with two or more vendors and services. The average
number of companies mentioned was 1.56, reflecting a
slight downward trend in the use of multiple vendors’
services relative to last year’s reported average of 1.74.
The individual user ratings earned by the 25 remote
computing companies that were rated by 5 or more users
are summarized in the accompanying Users’ Ratings
tables.
A “Weighted Average of All Ratings” was calculated for
each company by assigning a value of 4 to each user rating
of Excellent, 3 to Good, 2 to Fair, and 1 to Poor. Among
the 15 companies whose services were rated by 10 or
more users, the highest average ratings were earned by:
Company Overall Rating No. of Users
Scientific Time Sharing Corp.
3.33
17
On-Line Systems
3.21
12
The Service Bureau Co.
3.07
80
Cyphernetics
3.05
31
Highly regarded companies with fewer than 10 user
responses included:
Company Overall Rating No. of Users
It should be noted that the DATAPRO 70 subscribers and
ATSU members who responded to our survey may not
necessarily constitute a completely representative sample
of “typical” remote computing users. Furthermore, the
small sample sizes for some of the listed companies may
make it unwise to draw firm conclusions about relative
company performance from the indicated ratings.
However, Datapro believes the survey results that follow
can be of considerable value to users, prospective users,
and vendors of the commercial remote computing
services, provided the preceding caveats are kept in mind.
Responding users were asked to rate each remote
computing service they had used or were using by
assigning a rating of Excellent, Good, Fair, or Poor to
overall satisfaction, cost effectiveness, quality and
availability of technical support, quality of sales personnel,
training effectiveness, ease of use, manuals and documen¬
tation, applications packages, languages and compilers,
reliability, and response time.
Interactive Sciences Corp. 3.36 5
University Computing Co. 3.30 9
Standard Information Systems 3.25 6
First Data Corp. 3.18 6
The ratings assigned by all of the responding users can be
combined to form the following overall picture of user
satisfaction with the current remote computing services:
Excellent Good Fair Poor
Overall satisfaction
26%
56%
17%
3%
Cost-effectiveness
18%
41%
29%
11%
Technical support quality
31%
45%
20%
4%
Technical support availability
26%
41%
25%
7%
Sales personnel quality
26%
48%
19%
7%
Training effectiveness
14%
50%
27%
9%
Ease of use for experienced
45%
46%
8%
1%
DP people
Ease of use for inexperienced
18%
46%
28%
8%
or non-DP people
USERS' RATINGS OF REMOTE COMPUTING SERVICES (Continued)
Company*
APL Series, Inc.
Boeing Computer Services, Inc.
Compu-Serv Network, Inc.
Computer Sciences Corp.
Comshare, Inc.
Control Data Corp.
Cyphernetics
Data Resources, Inc.
First Data
General Electric Co.
Honeywell Information Systems
Interactive Data Corp.
Interactive Sciences Corp.
McDonnell Douglas Automation Co.
National CSS, Inc.
On-Line Systems
Rapidata, Inc.
Remote Computing Corp.
Scientific Time Sharing Corp.
The Service Bureau Co.
Standard Information Systems
Time Sharing Resources, Inc.
Tymshare, Inc.
United Computing Systems, Inc.
University Computing Co.
All others
No. of
User
Replies
7
19
24
40
23
21
31
5
6
134
7
6
5
13
36
12
15
5
17
80
6
5
46
23
9
131
Weighted
Average
of All
Ratings**
2.73
2.75
3.04
3.02
2.85
2.92
3.05
2.55
3.18
3.00
3.13
2.75
3.36
2.88
2.92
3.21
2.70
2.83
3.33
3.07
3.25
3.16
2.96
3.01
3.30
2.85
Users' Ratings*
*
Languages
and
Compilers
Application
Packages
Manuals
and
Documen¬
tation
Availability
of
Technical
Support
Quality
of
Technical
Support
c
P(
iuali
of
Sale
irsor
ity
is
inel
E
G
F
P
E
G
F
P
E
G
F
P
E
G
F
P
E
G
F
P
E
G
F
P
3
3
1
0
1
1
4
0
0
4
2
0
1
2
3
1
3
3
1
0
2
3
2
0
9
5
3
1
4
7
5
1
2
12
5
0
5
9
4
1
5
9
5
0
3
12
4
0
8
14
2
0
4
13
4
0
4
14
5
0
8
11
2
1
10
8
5
0
4
15
3
1
16
21
3
0
8
19
8
1
10
14
15
1
9
18
12
1
13
19
7
1
8
24
7
1
4
15
2
0
3
11
5
1
7
12
3
1
4
10
4
5
4
12
5
2
4
14
4
0
7
11
1
0
6
9
2
1
3
14
4
0
2
6
12
1
3
13
5
0
4
14
1
2
5
19
5
0
8
14
8
0
14
12
5
0
12
11
8
0
11
13
7
0
9
22
0
0
0
2
3
0
0
4
1
0
0
2
3
0
0
3
1
1
2
2
0
1
2
2
0
1
2
3
0
0
2
1
3
0
0
2
3
0
3
2
1
0
2
3
0
1
2
4
0
0
51
65
8
0
20
73
29
2
35
71
21
5
32
53
34
8
27
60
37
5
21
61
33
17
4
2
1
0
2
2
3
0
1
3
2
1
4
2
0
0
4
2
0
0
2
2
0
1
1
3
2
0
1
2
1
1
0
2
4
0
1
3
2
0
3
1
2
0
3
0
2
1
3
2
0
0
1
3
0
0
0
4
1
0
2
3
0
0
4
1
0
0
0
5
0
0
7
3
2
0
4
5
2
1
5
3
4
1
2
5
3
3
2
7
3
1
5
2
4
2
14
16
1
3
12
15
6
1
11
17
7
0
8
13
13
2
12
18
5
1
7
21
4
4
7
3
2
0
6
4
2
0
4
4
3
1
4
5
3
0
6
4
2
0
5
4
2
1
1
10
1
0
3
10
1
0
2
8
5
0
3
7
3
2
4
5
5
1
1
5
7
2
2
1
0
0
0
2
2
0
0
1
1
2
0
2
1
1
2
1
1
0
2
0
2
0
11
5
0
0
1
9
4
1
1
12
3
0
9
6
2
0
12
4
1
0
11
5
1
0
17
45
11
3
19
37
17
2
27
38
9
6
22
36
16
5
24
38
14
4
29
30
15
6
3
3
0
0
3
3
0
0
2
2
2
0
4
2
0
0
3
3
0
0
4
2
0
0
3
1
1
0
0
3
1
0
1
3
1
0
4
0
1
0
5
0
0
0
0
4
0
1
10
29
5
0
14
23
7
2
16
19
11
1
17
18
6
4
14
20
8
3
15
22
7
1
8
10
4
0
6
10
5
0
4
14
3
2
8
8
5
2
10
8
4
1
4
12
5
1
3
2
1
0
5
4
0
0
4
4
0
0
2
6
0
0
5
3
0
0
4
2
1
1
37
65
. 17
1
28
45
|32
16
16
66
32
I 15
22
53
40
15
30
61
|29
122
32
55
30
10
* Only the remote computing companies mentioned by five or more users are listed individually. The 57 companies rated by fewer than five users
are combined in the "All others" entry.
**Users' ratings are expressed in terms of user responses; the legend is E for Excellent, G for Good, F for Fair, and P for Poor. The Weighted
Average of All Ratings" was calculated by assigning a value of 4 to each Excellent rating, 3 to Good, 2 to Fair, and 1 to Poor.
23%
50%
22%
5%
23%
48%
22%
5%
34%
53%
11%
1%
36%
47%
13%
4%
35%
48%
13%
1%
Excellent Good Fair Poor Terminal Type
Manuals and documentation 23% 50% 22% 5% Interactive
Application packages 23% 48% 22% 5% Remote batch
Language and compilers
Reliability
Response time adequacy
As you can see, the users made it quite clear that they
were generally well pleased with the current services in
terms of overall satisfaction, case of use (for experienced
or DP-oriented people), languages and compilers, reli¬
ability, and response time. At the same time, the users’
overall ratings show plenty of room for improvement in
the areas of cost-effectiveness, technical support, training,
documentation, application packages, and case of use for
inexperienced users.
Determining the most popular communications terminals Language
from the survey proved to be rather cumbersome due to
the proliferation of terminals being used arid the frequent FORTRAN
use of more than one commercial service. Consequently, BASIC
avoid bias in the results, Datapro categorized the com- COBOL
municat ions terminal population by I unction only, result APL
ing in the following data:
No. of Terminals % of Total
1137
159
88
12
Thus, the survey results clearly demonstrate that
interactive terminals are still far more widely used than
remote batch terminals for remote computing applica¬
tions. The Teletype Model 33 continues to be the most
popular terminal for remote computing use. However,
other terminal manufacturers such as GE, Texas
Instruments, and Univac are winning a significant market
share.
The programming languages used by the survey
respondents were as follows:
No. of Users
354
216
98
41
% of Total
75
45
21
9
2
USERS' RATINGS OF REMOTE COMPUTING SERVICES
Company*
No. of
User
Replies
Users' Ratings**
Weighted
Average
of All
Ratings**
Overall
Satisfaction
Cost
Effective¬
ness
Response
Time
Adequacy
Reliability
Ease of
Use
(Inexperi¬
enced
Users)
Ease of
Use
(Experi¬
enced
Users)
Effective¬
ness of
T raining
Aids
E
G
F
P
E
G
F
P
E
G
F
P
E
G
F 1
P
E
G
F
P
E
G
F
P
E
G
F
P
APL Services, Inc.
7
2.73
1
2
4
0
1
4
2
0
2
4
1
0
1
2
2
1
0
3
2
0
0
5
1
0
0
2
2
1
Boeing Computer Services, Inc.
19
2.75
3
12
3
1
0
6
10
3
5
11
3
0
4
9
4
2
1
5
11
1
5
11
1
2
1
8
4
3
Compu-Serv Network, Inc.
24
3.04
5
14
5
0
5
11
5
2
7
15
2
0
9
13
2
0
5
12
5
2
12
8
2
1
0
9
10
1
Computer Sciences Corp.
40
3.02
13
22
4
1
7
24
8
1
13
18
7
2
12
19
8
1
9
15
10
4
22
13
3
0
2
24
8
2
Comshare, Inc.
23
2.85
2
16
5
0
2
8
11
1
4
16
3
0
6
13
3
1
1
10
7
4
7
12
2
0
2
10
5
2
Control Data Corp.
21
2.92
3
13
4
0
5
10
3
3
5
12
4
0
5
11
3
0
3
3
8
4
7
11
2
0
2
9
4
1
Cyphernetics
31
3.05
6
19
6
0
2
9
13
7
11
16
3
0
15
15
0
0
3
17
10
1
12
12
4
0
5
15
7
2
Data Resources, Inc.
5
2.60
0
4
1
0
0
1
4
0
0
1
4
0
0
2
2
1
0
3
2
0
1
3
1
0
0
3
0
0
First Data Corp.
6
3.18
3
3
0
0
4
2
1
0
2
4
0
0
2
4
0
0
1
2
2
0
4
1
0
0
0
0
5
0
General Electric Co.
134
3.00
28
78
21
6
13
66
44
8
55
63
11
2
58
57
13
2
22
80
22
5
64
54
8
1
19
71
26
4
Floneywell Information Systems
7
3.13
3
3
1
0
2
4
0
0
4
2
i
0
2
4
1
0
1
3
1
2
4
2
1
0
0
3
1
0
Interactive Data Corp.
6
2.75
0
4
2
0
1
2
3
0
3
2
0
1
2
3
1
0
0
1
3
2
2
3
1
0
1
2
2
1
Interactive Sciences Corp.
5
3.36
3
2
0
0
3
2
0
0
4
1
0
0
3
2
0
0
1
2
2
0
3
2
0
0
0
4
1
0
McDonnell Douglas Automation Co.
13
2.88
3
10
3
0
2
5
4
2
3
6
3
1
6
5
2
0
1
5
6
1
7
6
0
0
1
4
4
2
National CSS, Inc.
36
2.92
10
15
10
1
8
8
15
5
10
15
8
3
9
14
8
4
7
9
13
5
5
17
0
0
4
17
8
2
On-Line Systems
12
3.21
8
1
2
1
3
5
3
1
9
2
1
0
9
3
0
0
1
8
3
0
9
2
1
0
2
5
2
3
Rapidata, Inc.
15
2.70
1
7
6
1
1
6
5
3
3
8
1
2
2
10
1
1
1
8
3
1
1
9
4
0
1
4
5
2
Remote Computing Corp.
5
2.83
1
1
2
0
1
0
2
1
1
2
1
0
2
1
1
0
2
0
1
0
2
2
0
0
0
0
3
0
Scientific Time Sharing Corp.
17
3.33
7
8
1
0
7
6
3
0
11
6
0
0
10
7
0
0
4
6
4
2
8
7
1
0
5
10
0
0
The Service Bureau Co.
80
3.07
16
49
13
1
7
41
25
6
37
38
4
1
36
35
7
2
19
44
15
2
36
34
3
0
22
47
9
2
Standard Information Systems
6
3.25
3
3
0
0
2
3
1
0
0
3
1
1
2
2
2
0
2
2
2
0
3
3
0
0
2
3
1
0
Time Sharing Resources, Inc.
5
3.16
1
4
0
0
1
1
2
2
2
3
0
0
0
5
0
0
1
4
0
0
2
3
0
0
0
3
0
1
Tymshare, Inc.
46
2.96
8
30
6
2
5
18
14
9
8
23
11
4
18
17
7
4
13
13
16
4
18
23
3
2
7
15
16
2
United Computing Systems, Inc.
23
3.01
5
15
3
0
10
7
6
0
5
15
2
1
6
14
3
0
4
13
2
2
9
13
1
0
1
7
8
3
University Computing Co.
9
3.30
5
3
1
0
2
5
1
0
3
5
0
0
3
4
1
0
3
4
0
1
2
4
1
0
2
4
1
0
All others
131
2.85
31
66
23
11
40
45
31
14
48
57
16
11
36
65
22
8
20
48
43
13
55
56
15
1 2
11
40
38
23
* Only the remote computing companies mentioned by five or more users are listed individually. The 57 companies rated by fewer than five users
are combined in the "Ail others" entry.
**Users' ratings are expressed in terms of number of user responses; the legend is E for Excellent, G for Good, F for Fair, and P for Poor. The
"Weighted Average of All Ratings" was calculated by assigning a value of 4 to each Excellent rating, 3 to Good, 2 to Fair, and 1 to Poor.
Language
Assembler
PL/1
MACRO
RPG
No. of Users
17
15
5
1
% of Total
4
3
1
Thus, FORTRAN remains the clear leader in popularity,
with BASIC and COBOL also boasting widespread
acceptance among remote computing users. The survey
respondents were using an average of 2.16 different
programming languages each.
The remote computing applications reported by the
survey respondents spanned virtually the entire spectrum
of business and scientific applications. The leading
applications cited included the following:
Application
No. of Users
% of Total
Accounts payable
48
10
Accounts receivable
48
10
Banking
16
3
Billing
8
2
Communications
4
1
Data base management
60
13
Educational
11
2
Engineering
78
16
Financial
167
35
General ledger
21
4
Hospital administration
3
1
Information retreival
16
3
Insurance
17
4
Inventory control
34
7
Modeling
68
14
Numerical control
2
—
3
Application
No. of Terminals % of Total
Operation research 3
Payroll 21
Personnel 10
Production 20
Project control 7
Sales analysis 4
Scheduling 7
Scientific 27
Simulation 28
Statistical 83
Text editing 6
1
4
2
4
1
1
1
6
6
17
1
In addition to the information previously requested, this
year’s survey was expanded to cover other questions of
possible interest to the industry. This data covers areas
such as amounts spent monthly, changes in 1975-1976
usage, length of use, user categories, and proportional use
of time-sharing/remote batch services. The results are
summarized in the following tables.
Average Monthly Bill for
Remote Computing Services No. of Responses % of Total
Below $500
199
28
$500 to $2,000
239
34
$2,000 to $5,000
123
17
Over $5,000
148
21
Changes in Use of
Services During 1975
No. of Responses % of Total
Increased
430
58
No change
161
22
Decreased
116
16
Stopped
27
4
Expected Changes in Use
of Services during 1976
Increase likely
No change likely
Decrease likely
Will likely stop
Type of Service Usage
100% conversational
time-sharing
Over 50% conversational
time-sharing
Over 50% remote batch
100% remote batch
Length of Use
Under 6 months
Under 2 years
2 to 5 years
Over 5 years
Extent of Communications
Network Required
Local only
500-mile radius
Nationwide
International
Categories of users
Management
Programmers
Other data processing
personnel
All others
No. of Responses % of Total
236
37
218
34
135
21
49
8
No. of Responses % of Total
371
51
221
31
96
13
37
5
No. of Responses % of Total
61
8
215
30
288
39
164
23
No. of Responses % of Total
416 58
66 9
177 24
68 9
%of Total
1,435 15
2,478 25
2,117 22
3,725 38
No. of Users
This excerpt was reprinted with permission from "All About Time-Sharing
and Remote Computing Services," copyright 1976, Datapro Research Corp¬
oration, Delran, NJ 08075.
The entire report is available from ATSU (for members only) as part of
the following package for $15:
The Book:
• "A Manager's Guide to Computer Time-Sharing"
Published by Wiley, 1975 (regular price $9.95), and
The Three Datapro Reports:
• "All About Time-Sharing and Remote Computing Services"
• "All About Alphanumeric Display Terminals"
• "All About Teleprinter Terminals"
(Regular Price $10.00 each)
Payment must accompany all orders. """\
4
THE CASE AGAINST CORPORATE PLANNING MODELS*
Robert J. Allio, Canada Wire and Cable Limited
Management interest in the use of models for business planning and
forecasting has flourished in recent years. Much of this interest
was stimulated by the easy access which time-sharing has provided
to the computer and its ability to process large amounts of data
quickly. However, the development of a variety of simple program¬
ming languages and computational subroutines also has facilitated
experiments with planning models.
The result has been a surge in model building. Seventy-three per¬
cent of those corporations queried in a recent survey** were active
ly developing or applying computer models of their firms. Manage¬
ment scientists and econometricians also have constructed models
of more complex phenomena, such as the national economy.
Corporate management 3 unfortunately, has often been disappointed
and disillusioned by the performance of computer models. Models
have not increased our ability to contend with increasing uncer¬
tainty and some apparently irreversible trends. We desperately
seek better decision-making tools. Why have models failed to meet
our expectations?
The failures of modeling do not derive from any single source. We
must took instead to a variety of factors 3 including the following:
(1) The_ data bas_e i_s flawed
The data base for a model often contains inadequate 3
inaccurate 3 or irrelevant historical information. As
a result, the past may be as obscure as the future.
Output from a model having such a data base will be
unreliable and misleading. More input data may not
improve either the quality or reliability of the output.
Some corporations have been beguiled into the belief
that re liabi titiy of descriptions 3 forecasts, or output
in general increases the size of the data base or the
number of equations in the model increases. In reality 3
of course 3 it is often only the confidence_ of the modeler
which increases as model size or complexity increases.
Simple models may be more reliable than complex models.
( 2 ) Causal rel^a/ti^nshi^^ have^ not been, e_sta^_lished.
Models of business units are predicated upon either
causality or strong correlations among the variables.
The impact of changes in behavior by one function of
the organization (e.g. 3 marketing, manufacturing 3 engi¬
neering) upon financial performance often is hard to
-more-
* Originally presented at NABE's The Use and Abuse of Econometric
Models for Economic Forecasting and Business Planning Conference,
Port Chester, New York, December, 1975.
**Thomas H. Naylor, The Future of Corporate Planning Models, Social
Systems, Inc., 1975.
I
define. The task of defining these impacts is even more
difficult during a period of rapid change. Simple
static systems may he easy to model--hut, the results
have little interest to management. Ironically, it is
the interesting systems (complex and dynamic) which are
hard to model.
Our knowledge of how to huild complex models has greatly
surpassed our understanding of how a complex organization
works. Skill in programming a computer is a poor substi¬
tute for understanding the behavior of the corporation.
(3) Sif_stem behavior^ i_s ignored
Most models tend to treat business units as closed sys¬
tems or make simple assumptions about the key exogenous
variables. The contemporary corporation, however, is
part of an open system. Actions by any of a multitude
of stakeholders, including consumers, environmentalists,
legislators, labor, and. suppliers, can have a profound
effect on the organization. Few attempts have been made
to couple the behavior of the firm to the behavior of
the total system. Even the efforts to predict industry
sector performance based on macromodels of the economy
have been notoriously unsuccessful.
Simulating the effect of business tactics or strategy is
no easier. How will a change in price influence market
demand for a product? How will competition respond to
thi$ price change? Predictions based on the classic
laws of the free economy have less and less validity as
the environment becomes more regulated and monopolistic
or oligopolistic. Political decisions on fiscal and
monetary policy have been responsible for much of the
variance in recent economic forecasts--have any models
incorporated political variables into their structure?
(4) Last i/ear'g motde_l is_ o^Sjolet^
despite Santayana's position on this issue, we are con¬
demned to repeat the past only if we believe the future
to be an extension of the past. Models based on this
premise tend to give unreliable output. Paradoxically,
however, historical simulation is the standard method
for validation of models.
Surprise free scenarios have little value, because the
future is no longer an extension of the past. We live
in an era of rapid change. Many of these changes are
caused by discontinuities and by definition are either
unpredicted or unpredictable. Models must introduce
new assumptions about the future--those models which
ignore the future have not often been helpful to
management.
(5) Model vff 2 cmsf_ to qbancLf if t££. slow_
Models must be adaptable to changes in the environment.
Rates of change, however, are often too rapid for models
to accommodate--the reaction time of the model is too
6
slow. Even in a stable situation, the time required to
collect data, process data, and provide output to the
decision-maker is often incompatible with the needs of
the decision-making process.
The apparently high inertia of many models often is
created by the model builder who falls prey to the
Penelope web syndrome. Penelope, apparently abandoned
by Ulysses, staved off her many suitors by promising to
marry as soon as she completed weaving a tapestry. To
the dismay of her suitors, the tapestry made little .
progress towards completion. Although she wove dili¬
gently during the day, Penelope would arise each night
and undo a portion of her day's work. The corporate
model is always being refined by the model builder, and
seems to approach completion no more quickly than
Penelope's tapestry. To be sure, it is the client for
the model who is the ultimate victim of this syndrome.
The failures of models have contributed to a rapid deterioration
in their credibility and the credibility of model builders. The
isolation of model builders in many firms has not helped. Modelers
talk mostly to other members of their own cult. All too often
they fail to communicate effectively with their ultimate customer,
the management of the firm. As a result, they may not understand
the decision-making process of the organization. Strategic manage -
ment decisions, as one example, are rarely made on the basis of
quantitative data alone (although these data may be an important
input to the decisions). These are some of the specific traps
practitioners fall into:
0 Promote or endorse multiple data bases (e.g.,
the financial data base vs. the planning data
base).
0 Insist that packaged programs are never as
good as the unique program that can be
designed for the organization’s allegedly
unique needs.
0 Build "complete" but high-inertia inflexible
mode Is.
Even if the model be defect-free, top management is often uncom¬
fortable with the process of modeling. As a consequence, they may
be reluctant to accept the output from the model. The final
criterion for the quality of any model must be its acceptability
to the user.
We cannot, however, hold the model builder entirely responsible for
the recalcitrance of management. Most top executives have had
limited exposure to management science. This has created a cultural
gap between management and the model builder. All managers, more¬
over, rely heavily upon their intuition. This reliance creates an
interesting paradox. Managers will believe the output from a simple
model because it comports with their intuition--but, the output will
be dismissed because it is usually trivial. Output from a complex
7
model on the other hand is often counterintuitive. 4s a result,
although a manager's intuition may well he faulty, he or she tends
to reject such output (even though it may he correct).
Effectively designed and utilized models have high potential utility.
What can he done to improve their value and acceptance? Here are
some prescriptions:
0 Give special attention to the accuracy and relevance of
the data base; make sure that valid indicators are
selected*. Examine carefully the assumptions of the
model to assure that they will apply during the time
period of interest.
0 Make the response time of the model fast enough to meet
the needs of the decision-makers and fast enough to
sense any important external change. Consider the use
of a system to provide early warning economic, techno¬
logical, social, or political discontinuities and
turning points.
0 Do not present the model as a tool for making accurate
predictions. Do use the model to prepare alternative
scenarios, i.e., to illustrate the future consequences
of present decisions.
0 Improve communications with management. Emphasize the
benefits of the modeling process, e.g., in displaying
information. Start with simple, short turn-around
models which work (i.e., get management's confidence!).
Models of well-structured decomposable problems provide
the best entree to the later use of strategic models.
Modeling has been abused and oversold. Despite their limitations,
however, models have enormous potential to improve the decision¬
making process. Thus, while it is risky to rely blindly on models,
not to utilize their power is equally risky.
*Failures to properly forecast the 1974-75 decline in the U.S.
economy were the result of using incorrect economic indicators.
* * * *
Reprinted with permission from NABE NEWS,
Association of Business Economists, April
published by The National
1976, No. 8.
8
