"DEVELOPMENT OF A CASUAL: USER INTERFACE FOR SIMULATION MODELS"
Gregory Van Buren
Research and Simulation Manager
David. H. Goodstein
Director
Inter/Consult
Technology Center
21 Notre Dame Ave.
Cambridge, MA 02140
ABSTRACT
One of the traditional obstacles to effective utilization
of simulation models has been the great deal of time spent
learning languages in which models are written and keeping track
of the specific variable names and equations within models, To
remove this excessive psychological burden from busy executives
and to refocus attention towards the actual behavior being
replicated, Inter/Consult has been researching development of
highly supportive user interfaces to models. These interfaces
prompt ‘users bv stating the nature of the model's assumptions
then asking what changes they would like to make. Through this
on-line question-and-answer dialog users can build and compare
scenarios without prior knowledge of computer languages and
mathematical formulas or specific model components.
Qur paper presents reactions to the interface by members of
the graphic arts industry who have used it. We discuss further
improvements which are being made to the interface to make our
models more accessible to non~expert users, Finally we explain
why we feel that tightlv-focused, easy-to-use, dynamic simulation
models are of invaluable benefit to any industry such as graphic
arts where craft-oriented skills are being replaced by rapidly
evolving new technolgies.
652
INTRODUCTION
We feel it is vitally important that executives from varied
backgrounds take advantage of the insights which can be gained
through use of simulation models -- comprehensive tools which
integrate different assumptions and perspectives on industry
trends. A simulation model can act as a nexus which combines
different members of a strategic planning committee's views into
a single scenario.
The interface which has been developed for "TECHNOLOGY
DYNAMICS" models allows users to easily work with our models
while spending less time studying the model's operations and more
time analyzing behavior end establishing policy. A clear
understanding of the workings of models at the conceptual level
is absolutely essential for their effective use, yet detailed
knowledge of model structures or internals is not. We recognize
that major efforts must be made to determine the nature of
relationships within simulation models, yet we strongly feel that
as more people become involved in that process results will be
improved. Our overriding goal in the development of the interface
has been to increase the number of people manipulating and
examining our models.
Our first interface was built around Inter/Consult's model
“The World of Electronic Color", adapted from the Image Assembly
System model. The research that produced the Image Assembly
System simulation model was funded by a grant from Inter/Consult,
through Professor John Morecroft's Corporate Research Program at
the Sloan School of Management of the Massachusetts Institute of
Technology. The modeling was done by David Kreutzer and Robert
Lucadello, with important design advice from Dr. Alan Graham, and
reported in "Strategies for Investing in Electronic Color".[1]
This first application of TECHNOLOGY DYNAMICS deals
specifically with color image assembly or page makeup systems.
These million dollar plus systems are now revolutionizing the
process of preparation of printing plates for color magazines,
books, and catalogs. The TECHNOLOGY DYNAMICS model is helping
potential purchasers to evaluate alternatives about when and how
to automate their color prepress production. It also serves the
needs of systems and material suppliers who must understand how
technology maturity, price, industry structure, system
capabilities .and user field experience will interact to affect
the evolution of interactive color assembly systems marketed in
this decade.
WHY DYNAMIC MODELS
Selection of electronic color systems is currently based on
some mix of data, expert opinion and intuition, yet the potential
653
4
consequences of such a decision suggest that better methods for
strategic planning are needed. Rarely are any but the largest
companies able to consistently and comprehensively monitor
industry developments in an organized fashion so that their
decisions are based on the most recent technological data.
Unfortunately, the current state-of-the-art in market and
technological forecasting for the color separation segment of the
graphic arts industry is comparable to the agricultural industry
when its principal forecasting tool was the Farmer's Almanac. In
response to this critical situation, Inter/Consult has designed
and developed TECHNOLOGY DYNAMICS to assist industry decision
makers in the critical process of planning for the changes of the
next decade.
TECHNOLOGY DYNAMICS emerges from Inter/Consult's work over
the last three years on resolution of the problem of integrating
ail we know about trends in technologies with our market research
on the size of potential markets for new products. Its purpose is
to serve as a tool for achieving a better understanding of the
industry and for evaluating alternative corporate policies. much
of the groundwork on Technology Dynamics methodology was done in
our study, conducted with the Institute for Graphic
Communications, entitled, “The Future For Electronic
Pagination,"[2] which was published a year and a half ago. The
study's results and methodology were highly effective, but lack
of time and our inability to find suitable computer-based
modeling tools limited us to use of ordinary statistical analysis
packages and hand calculators for actual forecasts made in that
study.
TECHNOLOGY DYNAMICS addresses what we at Inter/Consult have
_come to call the failure of forecasting. Traditional market
research techniques in the iid 1970's stiowed themselves unable to
cope with the rapid change in products and market shape induced
by the emergence of the microprocessor. Econometric models are
excellent forecasting tools in situations where historical market
Gata provides a good indication of the pattern of future market
growth. In situations where product structures and market
conditions are changing drastically within short periods of time,
however, new and more intelligent forecasting methods are
demanded. System Dynamics models are uniquely designed to
replicate behavior of systems which are undergoing change.
Certainly no one would argue with our contention that graphic
arts and other skilled, labor-intensive markets being automated
by computer technologies will see tremendous changes within the
next few years. These will affect the competitive enviroment,
products produced, the way that value is added and the industry
demographics. 01d competitors will die and new ones will emerge.
654
When studying emerging markets, forcasters must not only
take advantage of traditional economic theory and practices, but
also draw upon survey techniques developed in the field of
sociology and cybernetic theories from engineering disciplines.
According to Forrester, "In modeling economic behavior all kinds
of information should be used, not merely numerical data. Rich
stores of information about economic structure and governing
policies are available from mental data bases built up from
experience and observation."{3] | System Dynamic models provide a
computational environment where nonstatistical information has a
meaningful context; the TD user interface allows it to directly
tap the vast mental databases of experts and decision makers
throughout the graphic arts industry.
Another attractive aspect of the System Dynamic approach is
that it attempts to deal directly with the interconnected
feedback loops which combine to form systems and impact market
behavior. “It takes the philosophical position that feedback
structures are responsible for the changes we experience over
time," wrote George Richardson and Alexander Pugh.[4] We at
Inter/Consult have spent a great deal of time pondering the
eventual outcome of related events in the evolution of markets
for computer graphics equipment. We have witnessed the learning
7
curves which vendors, service businesses and, eventually, end-
users go through when major breakthroughs in computer based
production are introduced in printing and publishing
environments. We are convinced that a systems approach which
anticipates the eventual maturing of revolutionary products is
necessary for strategic planning in situations such as this.
‘THE PROBLEM
The new electronic color technology has become the driving
force behind a dramatic evolution now visible in the color
separation industry. Traditional color printing preparatory shops
and the lithographers they historically served are now able to
purchase digitizing scanners and electronic color page makeup
systems with direct computer-to-plate output to automate their
operations, The resulting displacement of traditional skills and
of the labor-intensive craft orientation creates a potential for
rapid change in the structure of the industry. These systems also
necessitate a previously unheard of consideration for industrial
participants, the assessment of long-term strategic implications
of the decision to purchase computerized production equipment.
Executives responsible for evaluation of electronic color
systems must not only assess the current state-of-the-art but
also determine whether future products will be faster, more
capable and/or less costly. Strategic planners must determine
655
whether the timing of competitors’ investments in new production
technology will force them to move faster than they might wish
to. A. E. Gardner writes, "Contemporary decision making must now
be done within the context of a complex, rapidly changing
confluence of technological and unpredictable economic
considerations."[5]
Innovative electronic systems have already altered the
rules of the competitive game in the color separation industry.
As the capital cost of the highly productive new systems falls
over time, acceptance by existing companies will inevitably lead
to overcapacity in the sector. Moreover, low entry cost in
capital and operator skills will attract new corporate
competitors. Finally, as opportunity and operating costs fall
even lower, those businesses which once purchased services in the
open market will instead acquire their own in-house production
capability. This destabilizes the existing supply and demand
economics of & service-based industry. It forces dramatic changes
in the pricing, profit margins and customer profiles of the
service-oriented sectors. Experience in markets which previously
survived the complete cycle of automation like commercial
typesetting and newspapers indicates that there is a complete
restructuring of the in-house, service and captive sectors of a
market which sustains this type of transition, To quote Ephraim
9
Arazi, president of Scitex Corporation, the new electronic
systems “open up entire new markets and change the traditional
boundaries between sectors in the printing and publishing field.”
THE MODEL
The "World of Electronic Color Model” incorporates six
major sectors. These address technological trends, vendor issues,
and market conditions, as well as the concerns of customers,
individual firms within the industry and new entrants.
The Separation Industry Electronic Capacity Secter emulates
the displacement of craft operations by digitized systems. As the
price and performance of equipment improve, the number of units
sold ‘and installed increases.
The Electronic Color System Vendor Sector models the
behavior of system vendors. Changes in price and performance of
systems directly affect acceptance of systems by users in both
service and in-house situations.
The Color Separation Market Sector emulates the mechanisms
of supply and demand which determine the open market pricing and
profit margins for jobs performed on a service basis.
The Customer Sector traces the movement of work into and
656
10
out of the market, It examines the relative costs of alternatives
to color printing and considers the long term demand for color
pages.
The Individual Firm Sector takes real-world data from
industry sources or from the actual users of the model and
emulates the operational and financial life of the individual
company competing for market share on a service basis. The firm's
operating cost and utilization of its capacity, debt load,
investment rate and pricing of service all translate into a
determination of its market share and profit.
The Competitors ‘and Inplant Operations Sector models the
purchase of systems by in-house operators as described above. In
conjunction with the Color Separation Market Sector, it provides
model users a way of emulating the changes in customer base,
service demand, pricing, profitability and industry capacity
which occur over the maturation of a technology displacement
cycle.
Although this model specifically focuses on color page
makeup, other industries which are being impacted by new
technologies have similar structures and exhibit strikingly
similar behavior. We are presently developing a model which deals
with electronic technical publishing systems as part of a three
lL
volume study sponsored with Printing Industries of America. These
systems involve the integration of word processing, data
processing, Cad/Cam, and computer graphics with traditional in-
plant printing, composition, typesetting, and electronic or laser
printing systems. Integrated Electronic’ Technical Publishing
systems are being designed to handle the majority of corporate
information requirements,
~ THE INTERFACE
The concept of TECHNOLOGY DYNAMICS is that users feed in
their own views on industry trends which may or may not be the
same as the base run assumption provided by us. The goal is
development of internally consistent scenarios on how the color
systems ratio of performance/cost will evolve and how the markets
for them will develop. The connection between the dynamic model
and the user is in real time on a black and white video terminal,
with a dot matrix printer for on-demand hard copy. Output can
also be sent to a remote 4-color pen plotter or laser printer.
In designing the TD interface , several specific interests
were considered, The primary driving force in the implementation
approach was that all of the assumptions that comprise the model
should be spelled out in plain English. Second, users ought to be
able to change those assumptions through a dialog with the model
ih English. Finally, output from the model should be available
657
immediately and should be easily understandable without knowledge
of Dynamo or host system conventions.
We felt that top decision makers would be more likely to
use the model if they could concentrate on its content without
learning about its programming structure. Thus we translated the
majority of the model's equations into sentences. The user can
examine the various assumptions by going through a tutorial
online questionnaire. We ‘segmented the questionnaire into
sections which roughly correspond to the sectors of the model's
flow diagram, This allows people to address their specific
concerns without’ having to wade through a great deal of
extraneous information.
Each statement of the model's assumptions is followed by a
question as to whether or not the user agrees with it. In
instances where the user disagrees with our base run he or she is
prompted with questions about what the "correct" assumption is.
The interface then converts responses into Dynamo rerun
statements.
This approach limits the casual user to making changes which
can be replicated in rerun statements, yet we have found that 90%
of the types of scenarios our clients want can be generated
13
through the interface questionnaire dialog. Where structural
changes to the model are desired, someone with a detailed
understanding of both the model and Dynamo must make them.
On the output side, the interface asks which variables
should be plotted or printed for given reruns. Once again the
dialog is in English with answers being interpreted for Dynamo.
There are several options for type of output. Dynamo plots and
“tables of data can be sent back to the terminal in which case the
interface formats the output so that an entire plot will fit on
the ‘display screen. The same plot can be automatically enlarged
for output on an 8-1/2" by 11" page to be printed locally via a
printer attached to the terminal or sent to a remote laser
printing device. If higher, quality images are desired the
interface will érive a remote color plotter creating line graphs
from the tabular data Dynamo produces. Here the size of the
resulting graph can be specified along with the type of paper and
ink to be used, A major benefit of this method is that graphs can
be given descriptive labels identifying their origin and content.
The next area of interface development involved tailoring
it to the amount of experience of each user. We began by creating
three levels; beginner, intermediate and expert. Beginners are
given full paragraphs and sometimes pages of explanation of
assumptions. In addition to being a way to run the model this
658
a
makeup systems and markets. The intermediate level provides
shorter pieces of information, generally one or two line
definitions of underlying assumption. While the expert level is
for those who know the model and Dynamo. In order to facilitate
fast access by proficient modelers, brief phrases are used to
express the meaning of dynamo equations. Responses are confined
to numeric changes of model parameters.
An additional aspect of the interface is that it provides a
trail of rerun information that is placed in a historical file
for each user. Reruns can be recalled from the day before or any
other prior work session. As a result users develop and maintain
their own set of scenarios making adjustments only to their
tailored versions of the model. Many bookkeeping and database
headaches are eliminated in this way.
We are currently developing a version of the model which
will use a color video display and quantitize or interpolate the
dynamic, model results. These will then be fed directly into a
financial forecasting model. The end results will be profit and
loss forecasts and discounted cash flows for potential system
buyers and market forecasting or market strategic analysis data
for suppliers. Eventually TECHNOLOGY DYNAMICS will include
routine updates of key industry statistics with full real time
15
color interaction for scenario generation and direct links to
other packages for production of the reports.
THE REACTION
Thus far the response to both the interface and the model
has been overwhelmingly positive. People appreciate the model as
a means of analyzing the complex issues involved in prepress
automation without having to study the intricate lines of code
-which comprise it. Users come away genuinely pleased after
inputing scenarios which they had been considering prior to
exposure to the model but had never been able to spell out
clearly for their peers to see, Even the more conservative
members of the graphic arts community who are fundamentally
opposed to implementation of electronic prepress color systems
have found comfort in their ability to have their beliefs tested
and compared to those with more divergent, more optimistic
opinions. While few completely comprehend the full power and
advantages of dynamic models almost all of our users perceive the
value in logically replicating the behavior of their markets
using their own assumptions.
The color printing business is an extremely subjective one.
It is very difficult to learn how to produce quality work cost
effectively. It takes a well trained eye to interpret what a
customer means when he says, “I want a friendlier blue". Our
659
clients feel that "The World of Blectronic Color Model" increases
their objectivity and has made a positive contribution to the
graphic art industry. In a recent letter Zane Tankel, president
of Collier Graphic Services and an early user of the model wrote,
“We “did find your input most helpful and are certainly
appreciative. Hopefully, our input back to you was equally
helpful. I think there is a most definite need for TECHNOLOGY
DYNAMICS and your computer model is something more people should
take advantage of prior to acquiring any computerized imaging
system. It would certainly be time and money well spent." Tankel
points out another heretofore unrecognized benefit of creating a
friendly interface to Dynamo. As a consequence of having experts
from printing businesses use the model we do indeed find their
input equally helpful in strengthening our understanding of what
is occuring in the industry. The development of the model itself
has been an iterative process for us. Where we see consistent
disagreement with our base run assumptions by knowledgeable
users, we further refine the model to more closely reflect actual
behavior patterns. The cumulative experience of our users has
unquestionably been our best teacher.
THE CONCLUSION
This is an example of the kind of analysis that we can,
through TECHNOLOGY DYNAMICS, provide potential buyers of very
7
through TECHNOLOGY DYNAMICS, provide potential buyers of very
expensive and rapidly changing systems. However we must hasten to
point out two things. First of all, this is not a crystal ball.
It is a strategic support tool to help decision makers do things
that they do in their head already, but a little bit faster, a
little bit better, and in away that's a little bit more
shareable with others involved in making the decision, Second,
the work of getting this kind of tool into the hands of managers
“has only begun. More advanced and user-friendly interfaces are
both necessary and possible. At some point in the future users
should be able to merely speak to dynamic models and have them
respond, We feel TECHNOLOGY DYNAMICS and computers in general
must become "people literate" rather than vice versa.
™D is a new approach to the problem of strategy development
and assessment relative to the displacement of traditional means
of production by computerized systems in a craft-oriented
industry. This kind of modeling will be a uséful tool for users
and providers of, the new technologies.
ABOUT THE AUTHORS
Gregory Van Buren is Research and Simulation Manager at
Inter/Consult. He has designed market research for numerous
studies on new technologies in printing and publishing industries
and has been the project coordinator in the development of
660
18
Inter/Consult's simulation models. A recent Harvard College
graduate, his previous experiences include market research with
Kenyon and Eckhardt Advertising Agency and analysis of energy
policy forecasts with Stone and Webster Management Consultants.
David H. Goodstein is Director of Inter/Consult. He has spent
over ten years consulting with users and vendors on advanced
applications and markets in the evolving world of electronic
publishing. His work is in all areas of the Information
Technology world such as videodisk, CAD/CAM, and videotex. Mr.
Goodstein is a Research Affiliate of M.I.T.'s Visual Language
Workshop and designed Inter/Consult's simulation models.
1. David Kreutzer and Robert Lucadello, "Strategies for Investing
in Electronic Color", D-3357 (Cambridge,Ma.:System Dynamics
Group, MIT,1982)
2. Inter/Consult, "The Future For Electronic Pagination" (Boston,
Mass.:The Institute for Graphic Communications, 1981)
3. Jay W. Eorrester "Information Sources for Modeling the
National Economy", Journal of the American Statistical
Association, September 1980.
4, Alexander Pugh and George Richardson, Introduction to System
Dynamics with Dynamo, (Cambridge, Ma.:The MIT Press, 1981)
5. A. E. Gardner, "Graphic Communications World", (Tallahassee,
Fl.:Technical Information Inc, May 1982)
