Document 11044713

advertisement
££2N»^}
Center for Information Systems Research
Massachusetts
Institute of
Alfred P. Sloan School of
Technology
Management
50 Memorial Drive
Cambridge, Massachusetts, 02139
617 253-1000
DECISION SUPPORT SYSTEMS
AND
PERSONAL COMPUTING
Peter G. W. Keen
Richard D. Hackathorn
October 1979
CISR No. 47
Sloan WP No. 1088-79
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Ha cka thorn
—
Page 2
ABSTRACT
This paper explores the relationship between Decision Support
The aim is to clarify the key issues
Systems and Personal Computing.
relevant to helping managers make effective use of computer technology
in their own jobs. The paper suggests that there exists a useful
distinction between Personal Support (PS) and Organizational Support
This distinction is critical for designing a system for effective
(OS).
decision support. The emphasis is placed on Personal Support by
Finally, it is suggested
discussing development and research issues.
that Personal Computing lacks a conceptual base for exploiting the
potential of small-scale computer technology and that Personal Support
can provide such a framework.
OUTLINE
1.0
MOTIVATION FOR DECISION SUPPORT
DECISION SUPPORT SYSTEMS
1.1
1.2 PERSONAL COMPUTING
1.3 CONCLUSIONS FROM PREVIOUS RESEARCH
2.0
DEFINITION OF SUPPORT SYSTEMS
2.1
UTILIZING COMPUTER TECHNOLOGY
2.2 LEVELS OF SUPPORT SYSTEMS
3.0
FOCUS ON PERSONAL SUPPORT
COMPONENTS OF A SYSTEM FOR PERSONAL SUPPORT
3.1
3.2 BUILDING, USING, AND EVOLVING
3.3 THE DYNAMICS OF INTERACTION
4.0
RESEARCH ISSUES RELATED TO PERSONAL SUPPORT
4.1
IDENTIFYING IMPORTANT VERBS
4.2 LACK OF GOOD RESEARCH METHODOLOGY
5.0
CONCLUSIONS
ACKNOWLEGEMENT
The authors wish to acknowledge the valuable discussions with faculty
and students of the BA508A graduate seminar on "Personal Support
Systems" during the spring of 1979 at the Wharton School. University of
Pennsylvania.
n^n^H^
SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hacka thorn -- Draft #5 (10/10/79)
»,i,CISION
3
MOTIVATION FOR DECISION SUPPORT
1.0
Support
Decision
Page
Systems
designed to assist managers
(DSS)
interactive
are
in complex tasks
computer
aids
requiring human judgment.
The aim of such systems is to support and possibly to improve a decision
process.
the
Personal Computing (PC) is
direct
use
umall-scale
of
information processing task.
aspects of the technology:
a more recent
systems
The
user
individual
complete
has
access, usage,
an
by
term that describes
for
any
control over all
program development and data
management.
This paper explores the relationship between DSS and PC.
The aim is
to clarify key issues relevant to helping managers make effective use of
computer technology in their
of DSS and suggests
a
own jobs.
distinction
Organizational Support (OS) that
apolication
of
the
discusses design
concludes with
objective is:
concept
assessment
of
of
Decision
Personal
Support (PS) and
critical to effective
Support.
of systems
key
provides an overview
The
for personal
research
issues.
paper
then
support and
The
overall
(1) to define where Personal Computing fits into Decision
Support; and (2)
to exploit
provides
1.1
between
in itself seems
and development
an
The paper
DECISION SUPPORT SYSTEMS
the opportunities
that Personal Computing
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hackathorn
Page 4
—
The
movement
DSS
about
began
years
ten
ago.
was
It
largely
stimulated by MIT's Project MAC, which both provided access to computing
power to the individual
vast increases in
human-machine symbiosis that promised
handle complex
studies of
Morton,
problems.
DSS usage
[Licklider,
1965]
in organizations.
A coherent
1978]
There
extending the
?
our ability to
been several
have
1979; Keen
[Alter,
conceptual framework for
emerging, and practitioners are
pointed towards
first time and
user for the
and Scott
Decision Support is
academic work
of DSS by
building innovative systems to support a range of managerial tasks.
Decision Support requires a detailed
Unlike OR/MS
habits, needs, and concepts.
Decision Support is
rather than
based on
prescriptive
and much of traditional MIS,
descriptive paradigms
and
rationalistic
theme in Decision Support is that
not understand.
understanding of the manager's
of decision making
perspectives.
central
A
one cannot improve something one does
The act of "supporting" a
manager implies a meshing of
analytic tools into his or her existing activities.
The term "Decision Support Systems" is partly a rallying cry.
is no formal theory of Decision
from practice in this
built and
a number
Support as yet;
applied field.
of conceptual
key
theory largely emerges
that many
studies completed,
generalize from experience and thus
becoming apparent that the
Now
word
systems have been
we can
begin to
adequately define the field.
is
There
"support"
and
It is
that the term
"decision" may be context-free, misleading, or even inaccurate.
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hackathorn
Draft #5 (10/10/79)
Page 5
—
Most
of
conceptual
the
Its
manager.
individual
psychology;
decision
organizational
practitioners are
base
human
on
is
making.
has
comes
emphasized
from
problem-solving,
It
clear,
is
concerned with
far more
DSS
on
theoretical
focuS
the
literature
the
cognitive
rather
than
however,
that
organizational issues.
The
systems that they need are built around activities that involve planning
and coordination.
While the
system to support
a
priesthood of
manager's
decision
budget
faith prescribes a
the DSS
the lay missionaries
,
build a system to support the organization's budget process
1.2
.
PERSONAL COMPUTING
a new rallying cry.
The term "Personal Computing" is also
of the hectic enthusiasm of computer
many segments of
along by new
[Isaacson
society.
brings new
insights
to
technology
and
individual.
Computing as an information
has direct personal
control
definition implies
a
substantive meaning and
issue
Hackathorn
over
all
Support side
linking computing
of
defines
[1978]
stages
relationship
broad impacts on
Admittedly pushed
1978]
processing activity
personal
complements the Personal
et.al..
fundamental
the
to having
taking on more
technology, PC is
the
hobbyists, the PC area has emerged
of microprocessors
from minor applications
Born out
in which
of
of DSS.
PC
the end user
the activity.
This
technology
that
the
to
Personal
shares many
of the
same aims as DSS, while emphasizing small-scale technology and localized
systems.
The experience and ambitions
banner have much
to
offer
true
software tools developed by PC can
Dewoy
of those who
believers
in
DSS.
march under the PC
The hardware and
similarly extend the applications of
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hackathorn -- Draft #5 (10/10/79)
DSS, and
decision
support
provides
Page 6
experience
an
base
design
and
criteria for PC.
1.3
CONCLUSIONS FROM PREVIOUS RESEARCH
The Personal Computing field is very new and lacks research to build
upon
[Niles,
there
However,
1978].
have
several
been
detailed
experience has been gained to
descriptions of DSS usage, and sufficient
permit at least the following conclusions:
1)
'Easy' applications of DSS
success is high) are
find aspects of the problem- solving
finance, where it is simple to
process that can
manager.
better
be
functional areas as
involve such
those which
the likelihood of
in which
(i.e., ones
handled
the
by
machine
than by the
it is most practicable to build a DSS around a
In general,
micro decision (e.g., the selection of a product price or setting an
advertising budget)
or
convenient
retrieval
of
those
where
data
(e.g.,
an
automated file drawer).
2)
'Hard' applications by
interdependencies. as in
contrast
are
strategic planning, and
problem solving or decision is only
Macro decisions,
such
interdepartmental
activities
as
are many
where a manager's
part of a more complex process.
creating
and
there
a
multicriteria
substantial difficulties in implementation.
plan
with
problems,
pose
divisional
It should be noted that
easy applications generally involve PS and the hard ones involve OS.
Practitioners want to tackle the hard problems and may not gain much
insight from experience with PS.
1572S58
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hacka thorn
Draft #5 (10/10/79)
Paee 7
—
3)
DSS evolve.
one.
The
final' system is
New tools shape new uses
a DSS and
often
evolve
very different from the original
and vice versa.
imaginative
Managers learn from
new applications.
This means
that the initial design is partly an experiment;
a good DSS contains
mechanisms for its own obsolescence (or evolution
depending on one's
perspective).
4)
The evolutionary nature
of DSS
has led
emphasizes getting
started through
easily and quickly
changed
emerged: a DSS
and
consists of
extended.
user-system dialog and a set of
verbs
or
operators
"find", or
(e.g..
"smooth").
[Insert Figure
5)
DSS are hard
1
A
system that
can be
common structure has
interface that
manages the
modular routines that correspond to
"display",
Evolving
operators, as shown in Figure
design strategy that
a flexible
software
a
to a
"compare",
DSS
a
largely
"extrapolate",
means
adding new
1.
here]
to
evaluate
benefits and take
a
long
'improved' decision
decision outcomes.
organization but
since
time
making is
A
DSS
requires
can
a
they
provide
mainly qualitative
to institutionalize.
not necessarily
not
be
complex
the same
simply
process
Furthermore,
plugged
of
as better
into
the
adjustment
and
adaptation.
6)
Support involves
a
very
detailed
understanding
process, task, user, and organizational
is to be embedded.
of
the decision
setting in which the system
Figure
1
DSS Structure
User
~n
-t
r
I
SOFTWARE INTERFACE (manage r
user- system
DATA BASE
Operator
based on
verbs
(e.g.
'
I
New
Operator
(e.g.
LIST)
"SMOOTH"
INITIAL SYSTEM
EVOLVED SYSTEM
(
SOFTWARE INTERFACE:
OPERATORS
Dimensions of Quality
Ease of use
Flexibility
Helpfulness
Analytic power
Compatibility with User's
problem-solving process
(correspondence with user's
verbs)
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hackathorn
Page 8
—
It seems clear that a standardized design structure has emerged from
independent efforts to build
and its corollaries
—
DSS and that
evolution
—
learning
and
notion of support
it is the
that
makes a DSS
different from interactive models or online information systems.
DEFINITION OF SUPPORT SYSTEMS
2.0
Systems is that this area has
One major problem in defining Support
been principally
characterized
concept of Decision Support does
system.
then to evade
the issue by
available and suitable
criteria
for
defining
however,
emphasize Decision
DSS should be
stating that a
Decision
technology.
'suitable'.
technology has meant time-sharing
techology;
the
not necessarily imply a computer-based
meaningful to
be more
It may
computer
by
the
On
Support
other
Support and
based on any
involves finding
hand,
'available'
or minicomputers until recently.
The
technology is now changing very rapidly to include computer networks and
microcomputers.
The basic
objective
of
applying
the
technology
to
support decision processes remains the same.
2.1
The
concept
of
'replacement' in
the
UTILIZING COMPUTER TECHNOLOGY
support
DSS
Models, graphics, analytic
potential too.1
that
need
building a system in no way
been
has
literature
[Keen
contrasted
&
techniques, software,
to
be
matched
to
guarantees its use.
Scott
with
that
Morton,
and hardware
of
1978].
are all
their context.
Simply
For the manager, there
are way3 of applying quantitative methods and computer technology:
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hackathorn -- Draft #5 (10/10/79)
1)
Ignore them and
have been
There
do nothing.
Page 9
examples of
so many
insensitive, unrealistic, and incorrect application of technology to
management tasks that
managers'
own
experience
judgment
and
This is the
Replace decision making.
and data
Research, Management Science,
situations the
preferable
is
effective than any computer system [Hoos,
2)
in many
arguable that
it is
and
more
1972].
traditional aim of Operations
The technology
processing.
is used to rationalize, automate, and establish rules and procedures
Thus, a credit
for particular tasks.
scoring model can replace the
and judgment (although
loan officer's analysis
usually the officer
can override the model).
3)
Support
the
decision
extremes of doing nothing
means augmentation of
and provision of
This
process.
clearly
falls
between
and replacing human involvement.
Support
processes,
improvement, education,
that can be
integrated into problem
existing
ad hoc tools
the
solving, communication, and analysis.
A key
difference between
replacement and
support, that
in itself
implies a difference design approach, is that the former aims at solving
a problem or getting
while the
an answer
person.
For replacement the area
itself.
A complete system
of emphasis
is needed that
performed through the system.
adaptation, since the design is
There
and facilitate
is the
task or decision
allows the whole
can be no
on helping a
task to be
concept of learning and
centered around the task structure.
contrast, support suggests ongoing
to begin
latter focuses
development so that
the process
of augmentation
By
the main aim is
and improvement.
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hackathorn -- Draft #5 (10/10/79)
Page 10
The 'best' system for replacement is one that improves efficiency and/or
generates a better
solution.
procedures and processes.
to
For
It may be hard
of this,
Because
'hard' benefits.
support,
system improves
'good'
a
to link these to outcomes and
there can
rarely be
prediction of system uses and hence of costs and benefits.
a reliable
Obviously, a
proposal to support a task must be justified in much the same way as one
to replace decision making but in this situation, the estimates of costs
and benefits and outlines
Development activity.
a predictable and
and assumes that
based on
final venture.
that
In general,
prototypes.
evolutionary.
Instead,
preliminary
some
experiment
direct
initial
phase
replacement
complex and
The
one uses
provides
this
The
products; these may be
and
essentially
are
a Research and
It is foolish to plan this activity as if it were
carefully evaluated and
development.
design
for
R&D
needed
that can be
guidelines
for further
is
will
strategy
of
be small-scale and
one
is
innovative but
nature
a phased strategy
creating
of
are not exploratory
Decision
Support
implies
distinctive techniques for system building:
1)
the main
problem is
to get
started, to
build a
complete 'first'
system that can then be evolved
2)
There must be continual
area of study is
learning, in
the users'
a literal
sense.
For PS the
problem-solving processes;
and for OS
information flows and needs for coordination.
2.2
LEVEL 3 OF SUPPORT SYSTEMS
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hacka thorn
Page
—
A frequently cited framework
Scott
Morton's
which
[1971],
structuredness (i.e.,
classification of
combines
structured
versus
and strategic
a third dimension
be added, as shown in Figure
dichotomy
of
task
with Anthony's
operational
control,
This framework
seems to
(i.e.,
PS and OS.
—
has been Gorry and
unstructured)
planning).
apply equally well to both extremes of
this paper is that
Simon's
activities
managerial
management control,
for discussing DSS
11
The main proposal of
task interdependency
—
should
2.
[Insert Figure 2 here]
The most obvious characteristic of situations
in which DSS has been
successful implemented to assist managers (i.e., the 'easy' applications
discussed in Section 1.3) is that the
with the aid of
There are
the system.
actors or units.
user can make a complete decision
By contrast,
no interdependencies with other
several DSS recently developed to support
planning activities in large organizations involve tasks that are highly
dependent on other
organizational units.
is called "sequential"
by Thompson
decisions, each of which can not be
passes on some output.
This
[1967] and
type of interdependency
involves a
sequence of
carried out until the preceding one
Between sequential and independent lies "pooled"
interdependency, which is
inputs from Activity B but
more complex and
fluid.
also passes data back
Activity A requires
to A.
A process with
pooled dependency is highly interactive among the activities.
Examples of sequential, pooled, and independent tasks in relation to
the Anthony framework are shown in Figure
3.
Figure 2
MANAGEMENT ACTIVITY
OPERATIONAL
CONTROL
MANAGEMENT
CONTROL
STRATEGIC
PLANNING
INDEPENDENT TASK
POOLED INTERDEPENDENCY
SEQUENTIAL INTERDEPENDENCY
Page 12
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 <1 0/1 0/79)
Keen/Hacka thorn
—
[Insert Figure
here]
3
decision through
can reach a
others.
private analysis, though
her own
his or
been made,
choice has
when the
interdependencies. the manager
not involve any
Where the task does
it must
communicated to
obviously be
The point is that the problem-solving activities do not involve
interdependencies.
of the
value and the quality
may be of great
Personal Support
In this situation,
by the
decision improved
provision of a
system designed for direct and individual use by the manager.
Where
coordination
as
for
For
need
provide
to
for communication and
Organizational
example.
problem-solving.
will
budgeting
Support for
as a vehicle
be as much
computer-based aid will
Any
others'.
with
closely
mesh
must
activities
individual's
each
interdependencies.
sequential
involves
task
the
for storing
facilities
department level, allow these to
disaggregated, detailed budgets at the
be integrated first by division and then for the company as a whole.
Pooled
face-to-face
requires
interdependency
discussions.
No single unit can carry
with others.
This suggests
and
out its work without interaction
additional
an
negotiations
type
of
support
—
Group
Support (GS).
This allows users to develop joint plans through a system
that provides
fast
response
that
and
is
designed
to
permit
easy
exploration of alternatives and explanation of analysis.
The three
of
types
Organizational (OS)
—
support
—
Personal
all share a common aim.
to distinguish among these
types of
support as
(PS),
Group
(GS), and
However, it is essential
distinct, but related,
FIGURE
3.
Expanded Gorry-Scott Morton Framework
Independent
Pooled
Sequential
Structured
Order filling
Scheduling room
assignments
order entry
Unstructured
Notice writing
Scheduling management meetings
Shipping and
receiving
Structured
Task scheduling
Task assignment
Project Costing
Unstructured
Promotion
decision
Personnel matters
Budget formulatio
Structured
Office
Arrangement
Factory location
Factory building
Unstructured
Time Scheduling
Top Management
hiring
Merger
perational
ontrol
tnagement
introl
rategic
inning
—
UH.II I
M
—
—^^—— —
m
il
|
||
I
LUI.
I
1
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hackathorn
Page 13
—
components of a Support System:
1)
PS focuses on a
decision
(e.g..
U3er
setting
price,
a
in
a
discrete task or
selecting
a
stock)
users
of
class
or
that
is
relatively independent of other tasks.
2)
of individuals
GS focuses on a group
.
each
of whom are engaged in
separate, but highly interrelated, tasks (e.g., office activities).
3)
OS focuses
an
on
operations
sequence of
actors
and
or
activity
involving
(e.g.,
building
a divisional
task
organizational
a
marketing plan, capital budgeting).
A PS may be used within an
small-scale system
to
set
help
For example, each manager may use a
OS.
own
his
In both cases the
decision making;
aim
the key
of
budget,
and the
integrates these budgets using an
divisional staff then coordinates and
OS.
marketing
system
the
difference is
is
to
support the user's
focuses on
that PS
a person,
while OS is on an organizational process.
DSS emphasizes Personal Support,
While the conceptual literature on
several of the most successful
applications of Decision Support involve
Group Support and, more
recently, Organizational Support.
the DSS built
Morton [1971]
by Scott
production planning
provides Group
For example,
which facilitated
marketing and
It allowed
managers from
Support.
two departments to come together and use the system as a means of making
tradeoffs, sharing ideas, and reaching a consensus.
There are at
lea3t a few
DSS, designed for
have been difficult to institutionalize
involved sequential interdependency.
Personal Support, that
because the broader task really
For example,
BRANDAID is a system
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hackathorn
Page 14
—
that
in
Their
individual brand might
result in sales
company's products)
another of the
responsible
the managers
BRANDAID.
addition
Organizational Support in
to
of
plan
the
(i.e,
the
plans
the
In
for an
the expense of
be addressed.
had to
In addition,
required
brand
was not directly compatible
in a format that
from
outputs
level
next
process of analysis.
'
of
Scott Morton,
&
being gained at
integrating
for
decisions to be presented
the
'cannibalization
of
particular, the problem
[Keen
the
at
very different
This required a
organization.
with
integrated
were
plans
managers
development
the
facilitated
brand
individual
marketing plans by
1978].
substantially
company
one
This
or
whole
in
needed
process
place
of this Personal
Support.
This concept of
some differentiated technical
literature
text-editing, document
assist
tools that
on
to
use
OA
to
OA
to
electronic
and
professional
Support,
mail.
'knowledge
such
These
and
The
order
improve
entry
the
literature
operations,
efficiency
on
OA
does
The
as
are
workers.
concerned with Organizational Support.
streamline
functions.
much like DSS.
Personal
emphasizes
appeal
and
decentralized activities,
administrative
not only in
important,
building blocks very
preparation,
Practitioners are far more
wish
seems
For instance, Office Automation (0A) is a packaging of
relation to DSS.
conceptual
support
of
levels
They
coordinate
of large-scale
not
seem
to
acknowledge the differences in priority and focus; hence, differences in
the design cri eria for Personal and Organizational Support are also not
acknowledged
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hackathorn
Draft #5 (10/10/79)
Page 15
—
FOCUS ON PERSONAL SUPPORT
3.0
The discussion above
support systems and clarifies where PC
of this
focusses
paper
on
peripherally dealt with
published so far
provide
systems described by Keen
was
really
used
little
into
very
insight
and Scott Morton[ 1978]
Support.
and willing to
of implementing experimental
systems.
small-scale microcomputers, which
shape
the
recent.
concepts
We now
have a
necessary
whole,
the
the
Of
the six
large
only
them.
emergence of
Personal Computing
technology for
building
for
only
undertake the risky effort
created
has
PS.
addition,
In
been
only one, PROJECTOR,
,
the
On
has
The rest
empirical studies
The
organizations have been able
field, is very
which
Support,
DSS literature.
Personal
for
and DSS can be meshed.
Personal
in the
definition of
more differentiated
provides a
PS but
need to
Decision
Support
provides some guidelines but few examples.
3.1
COMPONENTS OF A SYSTEM FOR PERSONAL SUPPORT
A system for Personal Support can be
viewed as analogous to a staff
—
assistant to whom a manager issues commands
A major research issue is how to
interface that provides
the
communication expected with
questions are
more
complex,
conceptual work to build
upon.
quality,
staff
and
flexibility,
assistant.
there
For
been
has
How can a range
OS,
less
the
1976]
organizational
context
and
ease of
the research
empirical or
of users be supported?
Should the interface be more structured and standardized?
be meshed into
[Keen,
identify the user's verbs and build an
same
a
"Do this!".
How can a DSS
of communications, control
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hackathorn -- Draft #5 (10/10/79)
Page 16
systems, and demand for coordination?
From the
perspective
of
staff
a
assistant,
components for
the
Personal Support divide into three main parts:
1)
Is the manager
The Interface:
able to talk
with the PS
as with a
How much does he or she have to learn to structure
staff assistant?
the dialogue?
2)
The Operators:
this".
Do they
correspond
Obviously, the comparison
criterial
provides some useful
is
for
interface does not facilitate a simple
managers' command: "do
the
to
not
design.
and activities; and
(2)
Organizational Support, the
coordinator.
In both cases,
is flexibility and
is
if the
command) or is
Similarly, if the operator cannot
verb, it is unlikely
it
example,
it cannot support him or
easily be integrated
since: (1) it cannot
but it
a "help"
not built around the manager's vocabulary,
be directly related to a
For
one,
dialogue in which the PS can
explain its responses if necessary (through
her in the way an assistant does.
literal
a
not
that it will be used
into the user's concepts
clear
what it supports.
corresponding analogy
is with
For
a staff
the distinguishing feature of the human
adaptivity.
The quality
and usefulness
of the
support system similarly depends on these attributes.
3)
The Database:
cabinet or a
i3 equivalent to
the assistant's file
Clearly, the contents,
indexing, and access
The data base
library.
method determines the range, flexibility,
uses.
cost, and adaptability of
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hacka thorn
Page 17
—
3.2
BUILDING, USING, AND EVOLVING
Personal support implies
or predefined system.
no "final"
steps and modes of use for an
one may be able to identify the necessary
clear that the
it seems
building OS),
organizational process (i.e. in
Whereas
necessary focus is on the person when developing a PS.
This implies the
need to:
1)
design a dialog first;
2)
provide a preliminary set of operators;
3)
permit the manager to learn from the system; and
M)
to evolve the PS by extending its operators.
Learning and flexibility
They
suggest
"middle-out".
a
design
seem
sequence
On the basis of
which aim at identifying
be
to
central
similar
aspects of support.
Ness's
to
definition
of
observation of the
uses and discussions
the designer
builds an interface,
key verbs,
which is complete enough to permit a meaningful user-system dialogue but
which will certainly need
set of operators
is implemented.
with which to react, and
extension. A preliminary
later alteration or
The
user now
the designer has a
has a
concrete system
vehicle by which to verify
his design.
The natural sequence of development of a system for PS is:
1)
Using existing operators:
a prototype.
The initial system is a complete one, not
The operators reflect
on prior to the development
the verbs that
of the PS or ones
but that also provide new capabilities.
the user relied
that are easy to use
Page 18
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hacka thorn
—
soon develop sequences of analysis, will
it is likely the user will
"TREND". "SUMMARIZE", "PLOT").
results: use
when reviewing monthly
analyzing a
"PROFIT"; or
then "REPORT",
use "STATUS"
first
customer account:
in
(e.g.:
routines
higher-level
up
in effect build
of any value,
the tool is
If
Constructing sequences of operators:
2)
These sequences will tend to be user-specific.
Developing new operators:
3)
gains experience and
ready for (or
key assumption here is that as the user
A
the system,
confidence in
Learning
new operators.
even specify)
he or
she will be
leads to new
verbs
4)
operators, or occur
as
verb
adverbs qualify a
of
result
a
user's
the
learning (e.g.,
differentiated subfunctions
introduce
and
mean modifying
PS may
Fine-tuning a
operators:
Changing existing
within existing operators.
5)
nouns.
versus sales expenses" translates
The phrase "graph profits
into an operator "GRAPH"
nouns
two
the
with
the database are
general, items in
In
Adding data about objects:
"PROFIT" and "SALES
EXPENSES" being its arguments.
6)
Changing data structures
relating
retrievable set of objects which
or set
of
definitions
the
for
to
objects:
The
database is a
may need a particular organization
interface
to
identify
them and
them.
associate them with the operators that act on
These six aspects
behavioral concept of
concept, a support
nouns.
of use and
support
system.
Learning, and hence
evolution provide a
and
Support
the
technical
begins
evolution of the
by
link between the
realization of that
identifying
verbs and
Support System, similarly
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hacka thorn -- Draft U5 (10/10/79)
involves new verbs.
Page 19
An operator must be defined
in terms of a verb and
vice versa for the system to evolve.
It needs to be acknowledged
list are easy
to handle
with existing
the operators;
different to the
though
even
An interface
software.
redesigned without affecting
user
items on the above
that the first four
the system
performs
it
can be
seems entirely
same functions.
the
Operators can be added and altered, especially if a flexible development
language, like APL,
manage and change
is
used.
However,
data structures.
DSS literature on database design
far
more difficult to
is little
discussion in the
it
There
is
and the tools for
data storage in PC
both
use
are at present very limited.
3.3
THE DYNAMICS OF INTERACTION
The discussion above
interactive
implies
development.
The
Clearly, the key first step is
whose discussion of
"levels of
that in section 3.2 above,
design
interactive
process
relies
on
to identify "good" verbs.
language" in
the use
gives the example of
involves forecasting commodity
of
prices.
PS and
evolution.
Berry [1977],
of APL parallels
an economist whose job
His explanation of
elaborate, and he stresses its complexity, and difficulty.
out that carefully listening reduces the
a
his work is
Berry points
task to "smooth prices," which
requires an APL function for exponential smoothing.
Some verbs are obvious;
routines such as "PLOT."
verb reflects
a
complex
most PS will
-LIST."
"FIND"
analytic
require retrieval and display
or "TABLE."
routine,
such
In some cases,
as "PROJECT."
the
This
Page 20
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hacka thorn
—
programming
dynamic
a
routine invokes
relationships between managerial learning
between
interface must be built
so that new
here
complex
the direct
is
and extension of
evolutionary
and
use
evolutionary
made
point
key
The
assumptions and techniques.
involves
that
model
the PS, an?
Obviously,
design.
the
added easily and
operators can be
quickly.
approches to the design of
A central difference between traditional
assume completeness of specification and
a "final" system.
issue is getting started, building a system
and that
can
is that the former
needed for support systems
computer tools and those
and
learning
encourage
For PS, the
that the user can relate to
evolution.
most effective
The
the development of
approach for this seems to be to put most effort into
operators.
a "humanized" interface and a few key
M.O
RESEARCH ISSUES RELATED TO PERSONAL SUPPORT
validity of
One test of the
research
can
issues
formulated
be
understanding of the issues.
two categories:
( 1 )
proceeding further
the PS
The
those issues
with
PS;
and
whether a set of
framework is
that
produce
will
(2)
we
gaps
need
that
to
know about before
become
existing DSS research when viewed from a PS perspective.
U.1
IDE!.
IFYING IMPORTANT VERBS
deeper
be presented in
research issues will
that
a
apparent in
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Ha cka thorn -- Draft #5 (10/10/79)
deal about the user and must develop
For PS we need to know a great
processes (Stabell terms
studying decision
effective methodologies for
this "Decision Research").
designers
particular,
In
be able to
process must be couched
outputs from the
In both cases the
must
criteria for designing the software
identify important verbs and define
interface.
Page 21
in a form that allows direct translation into software (and occasionally
hardware) design.
[Berry,
Contreras.
1977;
key research
Some
1978]
questions in this respect are:
1)
How many important verbs
makers use?
do most decision
Clearly, if
easier to get started with a PS
the average is five, it will be far
than if it is 20 or 200.
2)
What verbs do
decision makers
in different
evolved into a useful
3)
What is
a
interface"?
"good
usability and
1
Support System for a
are some general verbs used for
manager? It seems likely that there
search and display of suta ("FIND,
in common?
verbs that can form the initial
Does there exist a set of primitive
Personal Support and be
tasks have
"PLOT," "LIST," etc.).
'
What
quality? [Stirling,
are
the
measures
users'
1975; Little,
1975]
of
At present
"humanizing" or "customizing" the interface is a haphazard task with
few validated rules.
These questions all focus on better ways of applying existing skills
and technology for
nature.
PS
development.
There are other more elusive
Most obviously, if the distinction
to revise
They
the
cognitive focus
existing
of
most
loose
DSS
are
primarily
empirical in
conceptual issues to be resolved.
between PS and OS
definitions
research
Decision Support.
of
[Keen
is valid, we need
&
Scott
The
Morton, 1978;
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hacka thorn -- Draft #5 (10/10/79)
Stabell, 1971],
seems fully applicable for
little work relevant to OS.
about organizational
The
Page 22
but there is relatively
PS,
propositions of Galbraith [1973; 1977]
information
processing
of
seem
practical value
here.
4.2
LACK OF GOOD RESEARCH METHODOLOGY
Many of the
empirical.
There is a
effective use of case
absence
complete
studies to
of
systems
the
are
research and
illustrate key concepts.
some detailed descriptions of DSS.
the lack of distinction
These suffer, however, from
methodological and
comparative
of
validate or
Keen and Scott Morton [1978] provide
several
knowledge are
our current
gaps in
discussed
but
PS
as
between PS and OS;
are
clearly
OS
(especially the Portfolio Management System which is the main example in
the book).
One conclusion to be drawn from the arguments presented here
is that we need more descriptions of
started
and
then
evolving
understanding of how managers
the
DSS usage; if the issue is getting
system,
we
learn the general
must
get
a
detailed
patterns of evolution.
One key question is the extent and speed of change:
1)
Do managers tend to
verbs and not
operators that
rely on
accept ones
that require
support their existing
new concepts
and modes of
analysis?
2)
What types
of
operators
anciytic, conceptual, etc.
—
—
data
seem
based,
model-based, heuristic,
most beneficial or acceptable to
the ustr? Are they based on individual differences, cognitive style,
etc.?
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Ha oka thorn
Page 23
—
Many of the
questionnaires are the
structured interviews and
least validated) approaches.
than
flows
Organizational
OS.
can
[Hackathorn, 1977a], but cognitive processes
observable.
tools for
most widely-used (and
problem is probably
The
The
Protocol analysis,
primitive, at best.
studying decision processes are
PS
measurement.
involve
above
questions
more complex for
identified
reliably
be
are elliptic and often not
Stabell's argument, that the key issue for Decision Support
is the development
of
methodologies
processes, seems of
central relevance for
technologies,
rests
this
on
studying individual decision
for
PS.
comprehensive
a
Regardless of available
understanding
of
the
individual user.
5.0
CONCLUSIONS
Most applications of Personal Computing
of programmable calculators,
The PC
journals provide
At
"technology
these
push",
simple data retrieval.
uses that
business systems with
typical clerical functions.
developments
have
being
than
go much
time the microprocessor-based
towards small
has pushed
rather
and
of practical
the same
canned software packages for the
and direction of
games,
few examples
beyond checkbook balancing.
hardware underlying PC
video
have been simply extensions
been
pulled
The scope
determined largely by
by
the
information
processing needs of individuals.
PC requires
a
conceptual
base
small-scale computer technology.
Support provides such a framework.
for
Our
By
exploiting
argument
here
the
is
potential of
that Personal
definition, it involves discrete
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Draft #5 (10/10/79)
Keen/Hackathorn
Page 24
—
tasks that do
procedures.
complex
require
not
linkages
with
is essentially that
The aim of PC
provide individuals with tools under their
systems and
of Decision Support: to
own control so that they can
to become extensions of
adopt and adapt them
other
their own problem solving
processes.
Current concepts of
very narrowly
PC focus
with Decision Support, the technology should be
an end.
Unless PC defines
that
of PC, software as
much
as
hardware,
for managers.
The
reducing risk.
That is, it is far
evolving larger systems from
that is tailored
R
and
to the
small
The meshing
to provide cheap
used
is important
mainly in
easier to justify a Personal Support
$1,000 than
one involving a
Since
usage cost.
prototypes,
D process
needs and
be
cost factor
relatively high
larger investment and
highly suited to the
can
costs under
System whose initial hardware
systematic base for
no
It seems clear that the technology
base.
personal tools
As
seen only as a means to
design strategies.
identifying applications and developing
of PS and PC constitutes that
has
it
end.
on the technology.
tools
the
of building
PC relies on
of PC seem
the initial version
decision processes
of a particular
individual.
Personal Computing extends
to carry out the decision
new technology.
Just as
Decision Support.
research that can clarily
Decision
vague rallying cry to an emergent
so too can Personal
Support
has
challenge now is
The
where to apply the
moved
from a somewhat
discipline for research and practice,
Computing build its
distinctive technical base and
provide decision makers with effective tools that they now lack.
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hackathorn -- Draft #5 (10/10/79)
Page 25
REFERENCES
Making in
Decision
Computer-aided
of
Study
A.
"A
Alter, S.
Organizations." Ph.D. Dissertation, Sloan School, M.I.T., 1975.
"How Effective Managers Use Information Systems." Harvard
Alter, S.A.
Business Review, 1976.
Decision Support Systems: Current Practices and Future
Alter, S.A.
In Press.
1979.
Challenges. Addison-Wesley
Paper presented at
"The Democratization of Computing."
P.
Berry,
Eleventh Symposium Nacional de Systeraas Computacionales, Monterrey.
Mexico, March 15-18, 1977.
Ph.D.
Cognitive Factors in Interactive Decision System.
Carlisle, J.
Dissertation, Yale University, 1974.
Proceedings of A Conference on Decision Support
Carlson, E.D. (ed.)
Database, vol. 8, no. 3. Winter 1977.
Systems.
Corporate Planning.
System and
Decision Support
L.E.
Contreras,
Informal Communication, 1978.
Addison-Wesley, 1973Galbraith, J.R. -Designing Complex Organizations.
Addison-Wesley, 1977.
Organizational Design.
Galbraith, J.R.
Decision Systems: An
Man-Machine
of
Design
"The
Gerrity, T.P.
Application to Portfolio Management." Sloan Management Review, vol.
12, no. 2, pp. 59-75. Winter 1971.
for Management
"A Framework
Scott Morton.
G.A.,
and M.S.
Gorry,
13. no. 1, pp.
Information Systems." Sloan Management Review, vol.
55-77, Fall 1971.
Proceedings of
Hackathorn, R.D. Modeling Unstructured Decision Making.
Data Base, vol. 8, no.
the Conference on Decision Support Systems.
3, pp. 41-42, Winter 1977 (a).
Decision Support Systems."
R.D.
"Design Criteria for
Hackathorn,
Proceedings of the 9th Annual Conference on the American Institute
of Decision Science, Chicago, pp. 150-152, October 1977 (b).
'Research Issues in Personal Computing." Proceedings
Hackathorn, R.D.
December
Washington,
1978. pp.
of the National ACM Conference,
547-551.
Berkeley, University
Hoos, I.R.
"Systems Analysis in Public Policy".
of California, 1972.
Personal Computing." IEEE
Isaacson, P. et.
"The Oregon Report:
al.
Computer, vol. 11. no. 9. PP. 86-97, September 1978.
Keen, P.G.W.
"Computer Systems for Top Managers: A Modest Proposal".
Sloan Management Review, Vol. 18. No. 1, pp. 1-17, 1976.
Keen, P.G.W.
"The Intelligence Cycle: A Differentiated Perspective on
Information
of
Federation
Information
Processing." American
Processing Societies Proceedings, 1977 National Computer Conference,
vol. 46, pp. 317-320, 1977.
Keen, P.G.W. and E.M. Gerson.
"The Politics of Software Engineering."
Datamation, pp. 80-86, Nov. 1977.
,
DECISION SUPPORT SYSTEMS AND PERSONAL COMPUTING
Keen/Hacka thorn -- Draft #5 (10/10/79)
Page 26
"Information Access For The Policy Analyst: A Portable
Keen, P.G.W.
the
American
Technology." Presented To
Educational
Finance
To Appear in DS Working Paper Series.
Association, January 1979.
"Decision Support Systems and the Marginal Economics of
Keen, P.G.W.
No.
79-03-16.
Department
Effort." Working Paper
of Decision
Sciences, The Wharton School, 1979.
"Studies of Decision Support Systems:
Keen, P.G.W. and Hackathorn, R.D.
Working Paper No. 79-03-17.
A Review and Critique".
Department of
Decision Sciences. The Wharton School, 1979.
"The Organizational Context of User-centered Computer Program
Kling, R.
Department of Information and Computer
Designs." Working Paper,
Science, University of California, Irvine.
Libraries of the Future.
Cambridge:
M.I.T. Press,
Licklider, J.C.R.
1965.
"Brandaid." Operations Research, vol.
Little, J.D.C.
23, no.
4, pp.
628-673, May 1975.
Management Decision Systems: Computer Based Support
Morton, M.S. Scott.
Cambridge, MA: Division of Research, 1971.
for Decision Making.
"Decision Support Systems: Theories of Design".
Paper
Ness,
D.N.
presented at Office of Naval Research Conference on Decision Support
Systems, University of Pennsylvania, November 4-7 1975.
"Type P and Type S Systems." Working Paper
G.E.
Ness, D.N. and Hurst,
78-03-01.
Department of Decision Science, The Wharton School.
University of Pennsylvania, 1978.
F.R,
Gray,
P.,
Carlson,
and Hayes,
J.
The Personal
Niles, J.M.,
Computer and Society: A Technology Assessment.
Presented at the
National Computer Conference, Anaheim California. June 1978.
Stabell, C.B. On the Development of the Decision Support Systems as a
Marketing Problem. Paper presented at the International Federation
for Information Processing Congress, Stockholm, Sweden, August 197^.
Research:
Description
and
Diagnosis of
Stabell,
C.B.
"Decision
D.
In
Heradstreit and 0.
Decision Making in Organizations."
Narvesen (eds.), Decision Making Research: Some Developments. Oslo,
Norway: Norsk Utenriks politisck Institute, 1977.
Sterling, T.D.
"Humanizing Computerized Information Systems." Science,
vol. 190, pp. 1168-1172, 19 December 1975.
Thompson, J.D.
Organizations in Action. McGraw-Hill, 1967.
SEP
1 AS86
ML HIT
MAR
o 2 1389
^J6.!.4W)
MAY
ACME
B00K3IN0ING
SEP 6
CO., INC.
1983
100 CAMBRIDGE STREET
CHARLESTOWW, MASS.
HD28.M414
no.
1088-
79
Keen, Peter G./Decision support system
737947
D*BKS
.QQ 1.36576.
.
3
005 0M2
TOflO
3
flTD
TOAO DDM M^3
Mr2r2
\tfn
<rlO*1
3
TOAD DOM MT3 MMA
-7q
3
TOAD DOM 52M MOA
HD28.M414 no.1090- 79
Bullen,
73795.1
Christ/Distributed processinq
.0* BK S
Q Q.1 3 6 5 J.
.
\Q>
.
3
TOflO
3
OOr?
0Mr2
TOAD DDM 52M
30=]
T2M
(0^1-7^
m 'c-yo
3
TOAO DDM
MEM
5r2M
3
HD28.M414 no.1092- 79
Stewart. Rosem/The nature and variety
73794.1
D»BKS
00131990
3
<iOflO
001
Tflfl
M73
tO^' 7 *
3
TOAD DDM 52M MMD
.-7?
3
10A0 DDM SAM
3
TOAO DDM
lfc.5
*5
M
c
13
3flD
^y 7<A
3
TDAO DDM MT3 MOb
TOAD DDM 5A1 7Tb
c c
i
l-'&D
Download