/EY
HD28
.M414
43
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
The Challenge: To Be
Part of the Solution
Instead of Being the Problem
February 1993
WP# 3541-93
CISL
Stuart E.
WP# 92-08
Madnick
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
The Challenge: To Be
Part of the Solution
Instead of Being the Problem
WP # 354
February 1 993
CISL
Stuart E.
Stuart
1
-93
WP# 92-08
Madnick
Madnick E53-321
Sloan School of Management
Massachusetts Institute of Technology
01239
Cambridge,
MA
MAR
2 5 1993
Reprinted from the Proceedings of
the
Workshop on Information Technology
December
&
Systems (WITS'92),
1992, Dallas, Texas.
Part of the Solution Instead of Being the
The Challenge: To Be
Problem
Stuart E. Madnick
John Norris Maguire Professor of Information Technology and
Leaders for Manufacturing Professor of Management Science
Sloan School of Management
Massachusetts Institute of Technology
Cambridge, Massachusetts 02139
(617) 253-6671
smadnick@eagle.mit.edu
Abstract:
Computer Science and Information Systems (CS/IS) fields in
Depending upon the directions that we take, CS/IS can
problems - or be a major contributor to the problems. The WITS
Society at large, and the
particular, are facing considerable challenges.
either lead to solutions for these
community is uniquely positioned to address and solve problems that are outside of the traditional
domains of either science and engineering or management. Examples are given of applications-driven
technology-intensive research needed by business to support globalization, increased productivity, and
rapid adaptation.
The Challenge: Symptoms
We
can
all
take delight in the latest advances in information technologies: the newest high-
speed chip, the increasingly exotic multi-media workstations, the incredibly sophisticated software,
etc.
Given
there are
Some
all
these great advances,
some very
of the
it
Nationwide percentage enrollment
86 and has been dropping ever since.
specific
numbers may be
Some have claimed
subject to debate, the
Chief Executive Officer attitudes.
feel that
that they are
it is
way
lies
ahead.
Instead,
to get rid of
Student perceptions.
study, three points were
that
it
in
CS/IS peaked around 1985-
has dropped by as
drop seems
much
as 75%.
clear.
Various reports have indicated that the majority of
they are getting sufficient value for the
making. Although there
primarily a
an enthusiastic future
are illustrated below.
Declining student enrollment.
do not
that
serious challenges facing the computer science/information systems (CS/IS) fields.
symptoms
Although the
would seem
may
amount
of information technology investment
be good reasons for outsourcing IT
something that
it
CEOs
activities, for
many CEOs
unpleasant.
is
There have been various studies made of career plans of students. In one
made regarding
careers in CS/IS.
First,
having good career opportunities. Second, CS/IS were viewed as
1
-
the CS/IS fields were viewed as
fields in
which one could make good
money.
T.V.
But, third,
shows
like
CS/IS were viewed as
fields that
were not
"LA Law" but not "Chicago Information
fun!
may
(That
there
is
why
there are
Technologist").
Economic analysis of the relationship between CS/IS and productivity.
indicated that, at the macroscopic level,
explain
Several studies have
no correlation between the increasing amount of
investment in information technology and a corresponding increase in productivity (some studies have
Although we can argue about flaws
indicated a negative correlation.)
studies, the fact remains that they
Taken separately or
do place
CS/IS
the
fields
collectively, these factors
attracting students or effectively impacting industry.
in the simplicity of these
on the defensive.
do not indicate
a
healthy situation for
Rather than leading to despair, these symptoms
should trigger a reexamination of our existing assumptions and actions. The CS/IS fields have for years
been major forces
for
reexamination and change in
manufacturing, transportation,
resisted but
do usually
Although
I
lead to
will
many
These "paradigm
fields
etc.
new and more
be presenting
industries, such as financial services,
almost always difficult and often
shifts" are
productive forms.
my own
and
It is
now our
time!
personal views on these issues, these concerns and the
desire to seek effective responses has attracted the attention of others in recent years.
and most comprehensive studies was reported
recent
Agenda
for
makes use
in the
One
book, Computing the Future:
Computer Science and Engineering by the National Research Council [HL92].
of
some
of the
A
most
Broader
This paper
material from that study and the author's participation in related panel discussions.
The Challenge: Research Process
There are various ways
many
engineering fields the following process
1.
certain
An
application-specific
number
2.
(e.g.,
which research areas are identified and pursued. For example,
in
is
followed:
identified (e.g., aircraft
wings develop cracks
after a
of hours of flight).
The key aspects
a vibration machine
3.
problem
is
in
of the
is
used
problem are abstracted and brought back
to
simulate the affects of hours of
to the laboratory for
study
flight).
Theories are developed and experiments are conducted to measure factors, explain reasons for
material failures, and propose alternative materials or structures.
4.
The
once validated,
A
results of this research are transferred back to industry to validate the conclusions and,
to
make use
of these
new
solutions.
disconnect from reality can take place
disconnect can take place
if
step
1
is
if
step 4
is
omitted.
Furthermore, over time a further
omitted and the input for the abstractions of step 2 comes primarily
2
-
(or only)
from the already abstracted output
results attained
have
less
would not be providing
and
less to
do with
benefits to industry
CS/IT by students and executives -
of step 3. Eventually, the research being conducted
the problems facing society at large.
and
exactly the
this ultimately
would lead
symptoms noted
Thus, CS/IT efforts
to a decrease in interest in
beginning of
at the
and the
this paper.
This
concern also has been noticed by others. For example, the Association of Computing Machinery report on
"The Scope and Directions of Computer Science: Computing, Applications, and Computational Science"
[A91J warned:
"A close interaction between computer researchers and others is essential so that the
questions under investigation remain connected to real concerns. Otherwise computing
research can drift into irrelevance and cease to earn public support. For this reason it is
in the best interests of the computing profession for computer researchers to engage with
applications."
Before going on,
human
is
it
should be noted that such abstracted research can be an important facet of our
search for knowledge and
is
what we often
a key element of
refer to as "basic science."
The
issue
not whether such approaches are "bad" but, rather, whether there are other approaches that can
lead to
more rapid
Such
businesses.
a process is depicted in Figure
Many
that consultants
problems
solutions to important problems facing our society today.
may
be
1.
It
of these problems can be resolved
perform
in
new enough
despite our enthusiasm, not
starts
by looking
at the
major problems facing
by applying existing solutions
helping companies to apply reported research results).
or difficult
all
enough
so that solutions
do not currently
Business Needs
Hard Problems (Requiring Research)
\
CS/IT challenges
1.
exist.
But, certain
Furthermore,
problems can be best solved through CS/IS research. From the
hard problems, some important CS/IS research challenges can be found.
Figure
(this is often the role
Process to Identify CS/IS Research Challenges
list
of
As one colleague
global transportation
noted, developing CS/IS-based information-intensive solutions to complex
management problems, such
as intransit visibility, can be just as challenging
research as developing query optimization algorithms - but can have impacts that can be measured in
billions of dollars of savings.
WITS Core
There
is
ambiguity and disagreements regarding the definitions and boundaries of
fields
such as
computer science and information systems. Furthermore, given the rapid changes, the definitions are
The CS/IS research agendas overlap with each other as well as
likely to continue to evolve over time.
with other fields in science and engineering as well as with various disciplines of management.
some
other hand,
research issues have been largely neglected because
management
researchers
On
felt
the
that
they were solely technical problems and thereby avoided them; at the same time, the science and
engineering researchers
felt
nobody focused on them!
core of the
the
that they
In
were
many ways,
WITS agenda. Research
management problems and
solely
this
management problems and
also avoided them; thus,
segment, depicted in the center of Figure
in this area requires a
the technologies.
Science and Engineering
Management
2.
The WITS Core
s
-
4
represents the
deep and enlightened understanding of both
w
Figure
2,
Some
researchers in the field of Information Systems
Harvard, have argued that although IS-related research
moved
(IS),
may
such as Prof. Warren McFarlan of
continue to expand,
will (or has)
it
into the research agendas of the related functional areas, such as strategy (IT strategy),
organizational studies (IT-based organizational change), and marketing (creation of
bases), leaving the IS field empty.
will diffuse into existing
"siezed"
Without debating the correctness of
management
disciplines, the topics within the
by any other other management
enormous data
this position that IS research
WITS
core are unlikely to be
discipline since they require significant technical research
expertise.
In the remainder of this paper,
driving
WITS
we
will identify
some
of the important business needs that are
core research.
The Challenge: Addressing Key Business Problems
Much
"Management
of the content of this section
emerged from
in the 1990s" research project at
MIT.
this author's
participation in the
This research project culminated with the
publication of the book, The Corporation of the 1990s: Information Technology and Organizational
Transformation (Edited by Michael Scott Morton) [S90].
Business today
is
characterized by dramatically increased geographic scope due to
globalization and world-wide competition, severe time and productivity pressures, and a rapidly
changing environment that often requires the restructuring of a company's operations in a very short
time [M92]. For example, top management
which each
performs
site
all
may need
consolidation
(e.g.,
decentralization
much more
in
which each
office
is
company in
the disintegration of
IBM
at the
into multiple
summary
responsible for only specific
information exchange from other offices.
through bank mergers) are occuring
(e.g.,
organization from a
of the functions relevant to the business, requiring only
information from other offices, to a company
functions, requiring
to restructure the
Movements toward
same time as movements towards
"Baby Blues").
Often the movement toward consolidation of previously independent companies or divisions
into a single unit
is
motivated by the goal of gaining economy of
scale.
Thus, a rapid integration of
separately maintained databases, processes, and networks would be necessary.
conversion, interoperable network protocols, and transportable software systems are
technological features necessary in such an environment.
Automated data
some
of the major
Over and over
comes up. Popular phrases include
the issue of time
"just-in-time", "continuous
flow of information", "time-based functionality", and "time-to-market".
compress the time from product concept
ship,
and the time
respond
to
financial services industry
abandonment, of
major obstacle
less
production, the time from product order to product
to a competitor's action.
have a
life
As an extreme
many
case,
span, from product concept to
first
innovations in the
deployment
to final
than a month. In the business community, computer systems are often viewed as a
to time
"Once upon
reads:
to start of
Businesses are trying to
compression rather than a
a time
I
facilitator
(one executive has a sign in his office that
thought that computers were the solution,
now
I
realize that they are
my
major problem.")
One example
being sent from the
the ships
that illustrates these needs
USA
to
were underway,
and problems involves several ships
full of
help people in a foreign country suffering from a political disruption. While
the political situation
improved
in that country but a natural disaster struck
another nearby country. The question posed was: Could one or more of the ships be diverted
with the
new
disaster?
goods
This involved knowing what goods were supplied (they had
multiple sources, each with their
own computer
stored,
which containers were on which
though
all
ships,
help deal
come from
systems), in which containers had those goods been
and where was each
of the necessary information existed in
to
some computer system,
systems involved that a manual inspection of the contents of
Even
of the ships currently located.
all
there
were so many disparate
the containers on each ship
was
ordered. Thus, a delay of several days occurred before the appropriate ships could be redeployed.
Although the above example may seem
more
the
norm than
the exception.
to
be an interesting but isolated case,
it
represents
much
For example, during the Gulf War, of the 40,000 containers of
material shipped to the Gulf, about 28,000 had to be manually unloaded and inspected in order to
determine
their contents.
In general, the physical
of the supporting information.
movement
of material
was
faster than the
movement
Transportation and logistic systems, in general, represent major
challenges to the effective utilization of information technology [Q91].
The Challenge: Key Research
Issues
Effectively integrating information from multiple sources both within organizations
between organizations represent both an important solution
challenge for CS/IS research [MSW90,
rich"
and "information poor"
the data.
if
MW91, SM89a]. An
they do not
Three particular challenges
know how
are:
to
many
critical
and
business needs and a key
organization can be simultaneously "data
to identify, categorize,
summarize, and organize
data semantics acquisition, data quality, and data
semantics evolution.
6
-
Data
semantics
As business operations become increasingly dispersed
acquisition.
geographically and functionally, differences in work processes at each
for
each
become more
site will
these differing sites
must
critical,
and
depends on which
what purposes the
for
would need
is
we would need
environment,
this is
to
A
being used.
when
primary measure of
be able
useful integration of this
17 different
sales.
Which
using the term
is
company's databases
when
necessary. Before these differences could
to represent the
semantics of the data as used in each
terms
sometimes called the "context" of the data [SM91b]. Further research on metadata
and context representation would provide
meanings and
their
is
company has
and functional group within the company
to reconcile conflicting definitions of
be reconciled,
definition
premium" which
office
definition
performed by people trained
leading to data incompatibilities and inconsistencies
For example, a certain insurance
interact.
definitions of the term "net written
definition is used
site
knowledge about data
the basis for increasing the
automation of the data reconciliation and integration process [SM89b,
facilitate the
SM91a).
Data quality. Organizations have become very concerned about quality in areas ranging from
manufacturing quality
relatively
little
information
is
to
attention.
software program quality.
Issues relating to data quality are
moved through
To
the organization.
feasible as increasing
and informally took
numbers
knew
account
this into
becoming increasingly important as
a large extent, data quality considerations in the
past were handled through personal familiarity; the user
his or her organization
Data quality, in comparison, has received
the characteristics of the data used in
when using
of information sources are used,
many
the data. This approach
not well
known
are increasingly exposed to data with various levels of quality for which they
familiarity.
Furthermore,
many
to the user.
do not have
its
For example, the source of a given piece of information
quality.
element in judgements about
quality attributes that
may
not
They
first-hand
current automated processes for converting, merging, and manipulating
the data renders inaccessible information about the original data that might have
information about
is
its
credibility
and quality [WM90].
There are
many
is
conveyed
often a key
additional data
be important, such as accuracy, completeness, timeliness, and
stability.
Defining and measuring the important data quality attributes and properly maintaining this qualityrelated information as data
challenge.
With
moves through and between systems represents
makers
this quality information, decision
will
be better able
to
a significant research
make
effective use of
the data.
Evolving semantics.
in
It
must be
realized that
autonomous databases are independently evolving
semantics as well as in context. For example, consider the situation of stock exchanges around the
world. Not only are the stock prices changing continuously, but the definition of the stock price also can
-
7
-
change. At some time in the future, the Paris stock exchange will probably change from being measured
in
French francs
ECUs (European Currency
to
explicitly report the currency,
changes from reporting
it is
to "last closing price" or
actual prices, as currently happening at the
database of stock prices,
when doing
it
"ticker tape" data feeds
implicit in the context of the source.
nominal price"
"last
The normal
Units).
More
that the
subtle examples include
from a percentage based pricing
Madrid stock exchange.
must be recognized
do not
to
Furthermore, in a historical
meanings had changed over time especially
a longitudinal analysis.
What
is
needed
to represent the desired (or
it
meaning
to capture the
assumed) meaning of the
Then
or another database.
way
not only a
is
would be possible
receiver,
of each of these sources but also a
which may be
a
human, an
way
application,
development context mediator algorithms
to
formally compare the semantics of the source and the receiver to determine
if
to
they are compatible,
Research on the source/receiver model represents a
partially compatible, convertible, or incomparable.
direction towards solving this problem [SM91b].
Conclusions
It
has been observed that the evolution of society goes through periodic discontinuities.
these transition points,
uniquely positioned
to critical
problem
to
new paradigms and
be a leader
areas.
fields of
make
helping to
in
we
But, before
endeavor emerge.
this
happen by
The WITS community
At
is
effectively applying technology
can help transform society,
we have
to
be prepared to
transform ourselves.
Acknowledgements
Work
reported herein has been supported, in part, by the National Science Foundation, MIT's
International Financial Services Research Center (IFSRC), MIT's Leaders For Manufacturing
(LFM)
Program, MIT's PROductivity From Information Technology (PROFIT) Program, and MIT's Total Data
Quality
Management (TDQM) Program.
REFERENCES
A91
Association for Computing Machinery, "The Scope and Directions of Computer
Computing, Applications, and Computational Science", Communications of
Science:
the
HL92
M92
ACM, Volume 34(10
p. 131.
Hartmanis, J. and Lin, H, (Editors), Computing the Future: A Broader Agenda for
Computer Science and Engineering, National Academy Press, Washington, D.C., 1992.
Madnick,
S.,
"Chapter
Information Technology
Wang,
MSW90
October 1991,
16:
in
Putting IT All Together Before
it
Falls Apart", in
Action: Trends and Perspectives, Edited
by Richard
Y.
Prentice-Hall, 1992.
Madnick,
S.,
Siegel,
M. and Wang,
Laboratory (CISL) Project
at
Y.R., "The Composite Information Systems
MIT, IEEE Data Engineering, June 1990, pp. 10-15.
-
8
MW91
S. and Wang, R., "Logical Connectivity: Applications, Requirements, and an
Architecture", Proceedings of the 1991 Hawaii International Conference on Systems
Madnick,
Sciences.
MIT Management
Q91
Quiddington, P., "Cruising Along the Information Highway",
Magazine, Fall 1991.
S90
Scott-Morton, M. S., (Editor), The Corporation of the 1990s: Information Technology
and Organizational Transformation, Oxford University Press, 1990.
SM89a
Madnick,
S. et al,
Evanston,
SM89b
Siegel,
the
SM91a
IL.
NSF
"Schema Integration Using Metadata", Proceedings of
S.
Workshop on Heterogeneous Databases, December 1989, Evanston, IL.
M. and Madnick,
Proceedings of the
SM91b
Siegel,
NSF
M. and Madnick,
1989
Siegel,
Composing Answers from Disparate Information Systems,
Workshop on Heterogeneous Databases, December 1989,
"CISL:
Proceedings of the 1989
VLDB
S.,
,
"A Metadata Approach
to
Resolving Semantic Conflicts",
Conference (Barcelona, Spain), September 1991.
M. and Madnick, S., "Context Interchange: Sharing the Meaning of Data",
Record, December 1991, pp. 77-79.
SIGMOD
VVM90
Wang, R. and Madnick, S., "A Polygen Model for Heterogeneous Database Systems:
The Source Tagging Perspective", Proceedings of the VLDB Conference (Brisbane,
Australia),
August
1990, pp. 519-538.
/
Date Due
Lib-26-67
MIT LIBRARIES
3
DUPl
TOAD ODflMbSbl
b