David vs. Goliath: Responses to
Domination Strategies in PC and Server OS Markets
Vedat G. Diker and Jochen Scholl
Nelson A. Rockefeller College, University at Albany - SUNY
Milne 300, 135 Western Avenue, Albany, NY 12222, USA
Tel: (518) 442 3937 Fax: (518) 442 3398
vd7606 @csc.albany.edu JocScholl @ aol.com
Abstract
Can free, open source, or non-proprietary software gain influence and market
share in a monopoly situation as it exists in today’s PC and server markets? The paper
tries to capture major dynamics in these markets and presents the surprising finding
that Microsoft’s dominance may be gone in the not too distant future.
I. Problem Identification
In the modern PC and server software arena, the position of the market
dominator Microsoft seems to be perpetual. Microsoft controls more than 90 percent
of the markets of Operating Systems and of major productivity tools. However, as
examples such as Acrobat Reader / Postscript and Java/Sun suggest, there are
opportunities to establish areas that are not or not completely under the control of the
market leader. Recently, the free or open source software movement known under the
names of GNU (which stands for GNU Not UNIX) and Linux has come to public and
even Wall Street Journal first page attention.
The philosophy underlying the free software movement is radically different
from proprietary approaches. In using a combination of copyright, patent, and trade
secret legislation proprietary software vendors restrict licensees of their products to
very specific and even time limited uses. Free software in contrast is free, i.e., its
source code is open, is can be widely shared and copied infinitely without restriction.
It can be modified and redistributed by anyone. Many scientists and researchers feel
that proprietary software stifles free academic research and the progress of science as
such, and hence favor free software approaches. Free software does not mean free of
charge, however. Rather than charging premiums for the right to use a piece of
software as in the case of proprietary software, free software charges are possible only
for add-on services such as installation, packaging, maintenance, troubleshooting etc.
In other words, not only the philosophies of free and proprietary software but also the
business models are radically different. There are more than 10 million users of free
software as opposed to some 250 million users of proprietary software as of this
writing. Free software already has the highest share in the market of web servers. Can
the free software movement and its business model represented by companies such as
RedHat, Caldera, Slackware, SuSe and others become even more successful? What
can Microsoft do about this? Will proprietary software maintain the upper hand in this
battle?
II. Model Structure
II. A. Overview
The OS (Operating System) Market Model is an effort toward modeling and
simulating the dynamics of the current PC/Server Operating Systems markets. The
model simulates the consequences of the competition between the professionally
marketed, “corporate built”, proprietary Microsoft OSs (Windows 95/98/NT etc.,
labeled “Windows” or “MS Windows” below) and a freely distributed, “alternative”,
open source OS (GNU/Linux) that is being built by numerous “professionally
unrelated” developers spread over the world. The model consists of four major
sectors, three of which represent the broad behaviors of three main stakeholders in the
market, namely OS Developers, Application Developers and OS Customers (PC
Customers). Accordingly, these sectors are called OS Developers Sector, Application
Developers Sector and OS Customers Sector. The fourth sector is the Market Effects
Sector through which the effects of market share between two OSs and domination
efforts by Microsoft.
The OS Developers Sector reflects the dynamics that determine the number of
developers working on developing the two OSs and the system capabilities of these
OSs.
Application Developers Sector is the sector where the dynamics that govern
the number of developers that work on developing applications for the two OS
platforms and the pool of applications that run on these OSs.
In OS Customers Sector, the number of PCs in use, the portions of this PC
pool on which the two systems are installed are calculated, and the choosing behavior
of PC users and the support of PC vendors are reflected.
A diagram of sectors and intermediary variables is given in Figure 1.
OS Developers Sector Application Developers Sector
MS Windows OS Developers Effect of Relative
Linux OS Developers Capability on App
MS Windows OS Capability Developers Flow
Linux OS Capability
Relative Linux OS Capability Effect of Linux Capability
1» on Linux Application
Developers Increase
‘MS Windows App. Developers
‘*MS Windows Application Pool
‘Linux App. Developers
Linux Application Pool
Relative Breadth of MS Windows
Applications
Application Influence
on New MS Windows
OS Suites Fraction
Effect of Relative Application
Developer Willingness towards
Linux on App Developers Flow Application Influence on
Fractional Change from
Windows to Linux
Market Effects Sector Ra _~ ‘OS Customers Sector
Relative Market 7
Market Share Effect on ee Mae PC Units in Use
Developer Willingness MS Windows in Use
Perceived Domination Effort by Meinedowe, Linux in Use
Microsoft Relative Market Share of
Relative App Developer MS Windows
Willingness towards Linux
Figure 1. Sectors and Intermediary Variables.
II. B. Sectors
II. B. 1. OS Developers Sector
The main stocks in this sector are ‘Microsoft OS Developer Employees’,
“GNU/Linux OS Developers’, ‘Windows OS Capability’, and ‘GNU/Linux OS
Capability’.
“Microsoft OS Developer Employees’ represents the number of Microsoft
employees that work on developing Microsoft’s OS platforms. This number increases
1.1% each quarter, and being multiplied by ‘MS Windows OS Developers
Productivity’, gives the “Net Increase in MS Windows OS Capability’. ‘Net Increase
in MS Windows OS Capability’ adds to ‘MS Windows OS Capability’, which
represents the system capability of Windows. The unit of system capability is defined
as OSCU (Operating System Capability Unit), which is an abstract construct. It
represents the power of an operating system platform embodied by its components, its
architectural capabilities, its HW scope, its programmability, its usability, and its
elegance (in an abstract manner). The more capable an OS becomes, the more and the
more diverse applications can be programmed and run on it. ‘MS Windows OS
Developers Productivity’ is a table function of ‘Microsoft OS Developer Employees’.
It is assumed that productivity of each employee decreases as the number of
employees increases.
“GNU/Linux OS Capability’ increases through ‘Net Increase in GNU/Linux
OS Capability’, which is determined by the number of respective developers and their
productivity, as in the case of ‘MS Windows OS Capability’. Here again, the
productivity of each developer decreases as the number of employees increases. There
is a difference in calculating ‘GNU/Linux OS Developers Productivity’, however. The
table function is multiplied by the total effect of scenario elements that are explained
in the following paragraphs.
The number of ‘GNU/Linux OS Developers’ increases through the flow ‘Net
Increase in GNU/Linux OS Developers’. This flow is a function of maximum possible
increase and the effect of a group of scenario elements. Maximum possible increase is
assumed to be the net change in the number of GNU/Linux OS developers if all
scenario elements are at their highest possible level, thus representing the ‘best
possible case’.
The scenario elements that influence ‘Net Increase in GNU/Linux OS
Developers’ are ‘Communication’ such as the Internet as low price backbone for idea
exchange and program (re-)distribution, “Technical Features’ such as. stability,
adaptability, scalability, and elegance, and ‘Philosophy’ such as open source/ free
software and priceworthiness.
There are two main feedback loops in this sector. The first loop is the
reinforcing (positive) loop that involves the variables ‘GNU/Linux OS Capability’ and
“GNU/Linux OS Developers’. As the number of developers that work on developing
GNU/Linux increases, so does ‘GNU/Linux OS Capability’, and as “GNU/Linux OS
Capability’ increases more and more OS Developers choose to work on GNU/Linux,
thus increasing ‘GNU/Linux OS Developers’. (Figure 2).
The second main loop in this sector is the one that involves “GNU/Linux OS
Developers Productivity’, as well as ‘GNU/Linux OS Capability’ and ‘GNU/Linux
OS Developers’. It is assumed that an increase in number of ‘GNU/Linux OS
Developers’ has a decreasing effect on the productivity of each individual developer,
so this loop is a balancing (negative) one. (Figure 2).
The main outputs of this sector to other sectors are ‘GNU/Linux OS
Capability’ and ‘Relative GNU/Linux OS Capability’, which is the ratio of
GNU/Linux OS capability to the total of the capabilities of the two OSs. ‘GNU/Linux
OS Capability’ influences ‘Effect of GNU/Linux Capability on GNU/Linux
Application Developers Increase’, which goes into the Application Developers Sector
and influences ‘Net Increase in New GNU/Linux Application Developers’ there.
‘Relative GNU/Linux OS Capability’ influences ‘Effect of Relative Capability on
Application Developers Flow’. That variable also goes into Application Developers
Sector, and influences ‘Fractional Application Developers Flow from MS Windows to
GNU/Linux’, the fraction of application developers that decide to move from building
Windows applications to building GNU/Linux applications, or vice versa.
, Linux OS
Linux OS Developers
Capability . Productivity
+
G)
Linux OS
+ Developers
Figure 2. Main loops in OS Developers Sector.
II. B. 2. Application Developers Sector
There are four main stocks in this sector: ‘MS Windows Application
Developers’ and ‘GNU/Linux Application Developers’ represent the number of
developers who work on building application for MS Windows and GNU/Linux
respectively. ‘MS Windows Application Pool’ and ‘GNU/Linux Application Pool’
represent the amount of finalized/released applications that run on each of these two
OSs.
The unit used to represent the amount of such applications that run on a given
OS is named VGAU (Value Generating Applications Unit). The ‘quantity’ that this
unit represents can be regarded as a ‘convolution’ of the number of applications run
on a given OS, the ‘significance’ of these applications in terms of the potential value
that can be generated by using them, and the ‘support potential’ of developer
companies that stand behind these applications. So the fancy program “XYZ” written
by moonlight hacker Johnny Foobar is not seen as Value Generating Application
under this definition, while MS Excel or SAP R/3 surely is.
‘Pools’ of applications that run on each OS increase through new application
developments, which is a function of the number of developers and their respective
productivity. The productivity of GNU/Linux Application Developers is assumed to
be higher (roughly twofold) than those of MS Windows Application Developers, due
to factors such as free and open source programming and the rapid and numerous
feedback on public program code.
The application pools decreases through application obsolescence. MS
Windows applications are assumed to get obsolete sooner, due to market conditions
and frequently updated OS and application versions that are incompatible with
previous versions in a number of cases. MS Windows applications have an average
life span of 12 quarters, while the average life span of GNU/Linux applications is 20
quarters. This also reflects the fact that “revving up” open source code applications
runs more smoothly than proprietary (hidden) source code applications.
It is assumed that the total number of application developers increases by 1.5%
each quarter. A certain fraction of the net increase in application developers, which is
determined by the effect of GNU/Linux OS Capability and the combined effect of a
number of scenario elements, goes into ‘GNU/Linux Application Developers’ stock.
The rest of the net increase goes into ‘MS Windows Application Developers’.
The scenario elements that influence ‘New GNU/Linux App Developers’ are
‘Communication’, ‘Technical Features’, and ‘Philosophy’. The Communication
effects here are the same as on the ‘Net Increase in GNU/Linux OS Developers’ in the
OS Developers Sector. However ‘Technical Features’, and ‘Philosophy’ elements
have slightly different meanings (Technical Features) and different dimensions in this
sector than in the OS Developers Sector, and are thus represented by different
variables. Whereas Technical Features refer to elegance, stability, adaptability, and the
scalability in the OS Developers Sector, they refer to robustness, maintainability,
compatibility, and priceworthiness in the Application Developers Sector.
Another structure that influences the distribution of application developers
between the two stocks is the flow of developers from MS Windows to GNU/Linux,
and vice versa. ‘Net Flow from MS Windows to GNU/Linux Application Developers’
is determined by ‘Fractional Application Developers Flow from MS Windows to
GNU/Linux’, which is a function of relative breadth of applications run on each OS,
relative capability of the two OSs, market driven willingness of application
developers, and the effects of the scenario elements explained above.
oy ees. New MS Windows
New Linux Relative Breadth of ;
Application Relevant MS Windows Application
Developers 7 Applications eo 4s) Developers
Linux “a wing
Care indows
Application
Pao PtP ool
+
Linux Fractional Application MS Windows
Application Developers Flow from Application
Developers + MS Windows to Linux . Developers
Figure 3. Main loops governing Application Developers Sector.
This sector is governed by four positive feedback loops. Two positive loops
reinforce the number of GNU/Linux application developers and GNU/Linux
application pool through ‘New GNU/Linux Application Developers’ and fractional
flow from MS Windows application developers to GNU/Linux application
developers. Another two loops reinforce the number of MS Windows application
developers and MS Windows application pool through ‘New MS Windows
Application Developers’ and fractional flow from GNU/Linux application developers
to MS Windows application developers. (Figure 3).
II. B. 3. OS Customers Sector
This sector represents the main OS market dynamics. The two main stocks are
‘MS Windows in Use’ and ‘GNU/Linux in Use’. These stocks increase or decrease
through a co-flow structure driven by the associated flows of the stock ‘PC Units in
Use’. Number of PC units in use increase through ‘New PC Shipments’, which is
based on historical data merely data driven, and decreases through ‘PC Obsolescence’.
“New PC Shipments’ invoke new OS installations. The fraction of new PCs on
which MS Windows or GNU/Linux are installed is influenced by the relative breadth
of applications run on each OS, and the relative number of PC vendors that bundle
MS Windows only or both OSs on their newly shipped PC units.
As PC units become obsolete and go out of use, the OSs that run on those PC
units also go out of use, thus the outflows that decrease the stocks of “MS Windows in
Use’ and ‘GNU/Linux in Use’, namely ‘MS Windows Phase Out Rate’ and
“GNU/Linux Phase Out Rate’ are calculated as functions of ‘PC Obsolescence’, and
relative market shares of respective OSs.
Another structure that influences the number of MS Windows and GNU/Linux
in Use is ‘Fractional Change from Windows to GNU/Linux’, which is determined by
relative breadth of existing MS Windows and GNU/Linux applications, and a number
of scenario elements such as reliability, user friendliness, price and distribution
concerns. Under reliability the stability and robustness of the OS are subsumed, user
friendliness refers to the ease of use e.g. given by a graphical user interface (GUI),
while distribution concerns reflect the ease of distribution and redistribution of
applications. Proprietary applications may not be distributed, while open source / free
applications may.
The feedback loops in this sector do not significantly influence the overall
model behavior. However, certain paths in this sector are parts of larger multi-sector
feedback loops that govern the general behavior of the model. Relative market shares
of OSs influence the application developers willingness towards building applications
for these OSs, which in turn effects relative breadth of applications run on each OSs.
Relative breadth of applications influence the numbers of MS Windows and
GNU/Linux in use through the fractions of new suites and the fractional change
between two OSs. Four main positive feedback loops formed by these effects are
shown in Figure 4.
P- _ New MS
Linux in Use Ss MS Windows in <———— Windows Suites,
+ Fraction
Use
A
GAP Gy
4 Fractional Change
from Windows to -
Relative Market Linus
Share of MS Relative Breadth of
Windows , — Relevant MS Windows
2 _ Relative Application = Applications
Bm Developer Willingness ——®™
towards Linux
Figure 4. Multi-sector feedback loops involving paths from OS Customers Sector
II. B. 4. Market Effects Sector
The effect of relative market share of two OSs on application developer
willingness to build applications for either OS is calculated in this sector. This is done
by calculating the direct effect of market share on developer willingness to develop for
or on any given platform, and the effect of Microsoft's domination efforts on
developers willingness to develop on a platform which is also determined by relative
market shares.
TII. Model Behavior
III. A. Base Case Run
The base case of the model is run under the assumption that all the scenario
elements worked perfectly for GNU/Linux, which we believe is the case today.
Figure 5 shows the behaviors of the main stocks in the model. It is observed
that GNU/Linux OS Developers increase exponentially, while MS OS Developer
Employees increase linearly as intended. The sheer difference between the two OS
developer stocks begins to work in favor of GNU/Linux after the 12" quarter, and
GNU/Linux OS Capability, showing an exponential growth itself, surpasses MS
Windows OS Capability in the 29" quarter.
The increase in GNU/Linux OS Capability affects the distribution of
application developers in favor of GNU/Linux, and as the number of GNU/Linux
Application Developers reaches almost half of the number of Windows Application
Developers, the GNU/Linux Application Pool reaches three fourths of the Windows
Application Pool. The two main reasons why GNU/Linux enjoys a relatively high
number in the application stock compared to the low number of GNU/Linux
application developers are that GNU/Linux application developers’ productivity is
higher, and the life span of GNU/Linux applications is longer.
The dynamics in the Application Developers Sector soon show their effect on
the OS Customers Sector. The increase in the number of MS Windows OS suites in
use begins to slow after the 28th quarter, and by the 33rd quarter this number even
begins to decrease. Meanwhile GNU/Linux OS suites show an exponential growth
behavior, as expected.
OS Developers and Capabilities
30,000 persons
3,000 persons
20,000 OSCUs
15,000 persons
1,500 persons
10,000 OSCUs
0. persons L-
0. persons Pe.
0 OSCUs cam ty}
4 8 12 16 20 2% 28 32 36 40
Time (Quarter)
Linux OS Developers : base —t 4 4 4 4 4 + persons
MS OS Developer Employees : base 2222 2 persons
Linux OS Capability : base 3——3 33 ‘OSCUs
MS Windows OS Capability : base —1 44 4 4 4 Qscus
Application Developers and Pools
200,000 persons
1,200 VGAUs
[21
100,000 persons
600 VGAUs ae
A
|}
—] t |
0 persons Ltr]
0 VGAUs _, fatty
0 4 8 12 16 2 24 2 32 36 40
Time (Quarter)
Linux Application Developers : base. —1—__t____1___i____+ 4 persons
MS Windows Application Developers : base 222 2 persons
Linux Application Pool : base VGAUs
MS Windows Application Pool : base, 4444 VGAUs
Installed OS Base
400M_ OS Units
400M OS Units
300M OS Units
300M OS Units Let
200M _ OS Units
200M OS Units
Lea!
| 2+
,-
100M OS Units 7
100M OS Units
0 OS Units
0 OS Units _, 4 4 =
o 4 8 12 6 20 24 2 32 36 40
Time (Quarter)
Linux in Use : base, —1_4__4__4_4 44 4 4 4 Q§ Units
MS Windows in Use : base OS Units
Figure 5. Output from Base run of the model.
Il. B. Validation Runs
III. B. 1. Extreme Condition Tests
A number of extreme condition runs were performed to test model’s
robustness. As a part of these test, both GNU/Linux OS Developers and GNU/Linux
OS Capability were set to zero, representing a case where GNU/Linux OS does not
exist. The stocks “GNU/Linux Application Developers’, “GNU/Linux Application
Pool’, and ‘GNU/Linux in Use’ stays at zero, as expected.
As another extreme condition test all scenario variables are set to zero,
representing the case that GNU/Linux has no leverage points, such as code sharing
over the Internet, open source programming, etc., at all. As expected, both OS and
application developers of GNU/Linux hardly increase, as seen in Figure 6, showing
that the OS stays more of a ‘cult thing’ which is of interest only a small group of
closely related geeks and hackers. Accordingly, GNU/Linux OS capability and
application pool almost stay zero. The very slight increases observed in Figure 7 are
attributable to the ‘hobby level’ efforts of this small group of programmers. As
expected the number of GNU/Linux OS suites in use remains at zero.
Ill. B. 2. Sensitivity Analyses
Sensitivity runs are another set of validation tests that are performed in order
to test a model’s robustness. One of these sensitivity runs involves “GNU/Linux
Application Developers Productivity’. 200 runs are made for values between 75% of
the original value, and 125% of the original value. As it can be observed in Figure 8,
the model is somewhat sensitive to changes in the value of “GNU/Linux Application
Developers Productivity’, which suggests that this parameter should be further
investigated.
Another sensitivity run involved ‘Weight of Scenario Elements Effect on New
GNU/Linux OS Developers’. 200 runs were made for values between a lower bound
of 50% of the original value, and an upper bound of 150% of the original value. The
graph for GNU/Linux OS Developers in Figure 9 shows that the model is sensitive to
changes in this variable. However that effect does not heavily impact the number of
GNU/Linux OS suites in use as can be seen in Figure 9.
TII. B. 3. Scenario Tests
As the next step in the validation procedure, some scenario tests were
performed. One illustrative example for these tests is where Communication and
Philosophy elements are set to 0.05, 5% of the original value. This represents the case
when fast and inexpensive communication media such as the Internet can be used only
in a very limited manner, and philosophical elements such as open source software
and code sharing are not playing any major role as in the real world. It can be
observed in Figure 10, that increases in key stocks related to GNU/Linux are far from
being as dramatic as in the base. Even though the numbers of GNU/Linux OS suites in
use increase considerably after the 28" quarter. GNU/Linux does not seem to be a real
threat to MS Windows within the selected time horizon. We consider 40 quarters to be
a reasonably long period for analyzing this problem in the PC/server software market.
Linux OS and Application Developers
20
‘0 | } } |otel+
5 gail
[ot
ff 4
=e
jitit
°
° 4 8 12 «16 «20 (hCG
‘Time (Quarter)
Linux OS Developers :scenzero 1+ 1 4 1s 1 4 persons
Linux Application Developers : scenzero. 2222222 persons
Figure 6. Output from extreme condition run: all scenario elements set to zero.
Linux0S Capability and Application Poo!
2 Oscus
0.2 VGAUs fayT
1.8 OSCUs
0.18 VGAUs
1 OSCus
0.1 VGAUs Bi
0.8 OSCUs
0.08 VGAUs
oscus el
veaus 4} +}
0 4 8 12 16 2 24 2 32 36 40
Time (Quarter)
Linux OS Capability : scenzero 111111 1 1 oscus
Linux Application Pool : scenzero. 2222 2 2 2 2 VGAUs
Figure 7. Output from extreme condition run: all scenario elements set to zero.
IV. Conclusions
We better understand the nervous public reactions of Microsoft executives
toward GNU/Linux in recent weeks. The free software movement seems to have the
potential to alter at least some rules of the game in the PC and server markets. We see
prominent companies opening and freeing parts of their proprietary software, and even
some diehards of the proprietary movement such as Apple released key technologies
(e.g. Quicktime) to the public. We cannot, however, predict from this model whether
or not the new, leaner business model has the potential to redefine the entire game.
For answering this question the implications and dynamics of the diverging business
models need to be included into the simulation model. This might be a consequent and
natural step of further investigation. We are, however, confident that the free software
movement will gain even more momentum than can be discovered today. This by
itself is a dramatic development given the monopolistic scenario at the outset. We
suggest that the DOJ trial against Microsoft and the proposed break-up of the
company - as some Microsoft critiques demand - should be given second thoughts. So
far, the arguments for cracking down the company were based on (1) the conjecture of
a market failure in the PC and server markets (with Microsoft having more than 90
percent market share), and (2) the evidence of Microsoft’s monopolistic misbehavior
and unfair practices against other competitors. If the potential of developments as
shown in our model exists under such adverse circumstances, we believe the market
failure argument cannot be maintained.
prod
50% 75% (NN 95% I 100%
prod
50% 75% (NNN 05% 100%
Linux Application Pool Linux in Use
2,000 200 M
1,500 150M
1,000 100 M
500 50M
9 ——__—— 0
0 10 20 30 40 0 10 20 30 40
Time (Quarter)
Time (Quarter)
Figure 8. Output from sensitivity run: 75%-125% of GNU/Linux Application Developers Productivity.
weight
50% 75% (I 05% 100%
Linux OS Developers
60,000
45,000
30,000
15,000
0 10
20
Time (Quarter)
30
40
weight
50% 75% (NNN 95% J 100%
Linux in Use
200 M
150M
100M
5 es sr eer]
° 10 20 30 40
Time (Quarter)
Figure 9. Output from sensitivity run: 50%-150% of Weight of Scenario Elements Effect on New GNU/Linux OS Developers.
OS Developers and Capabilities
30,000 persons
3,000 persons
20,000 OSCUs
15,000 persons
1,500 persons
10,000 OSCUs
+ -
0 persons
0 persons 4
0 oscus tot, plolplel | 34
0 4 8 12 16 20 24
Time (Quarter)
Linux OS Developers : comphil + 4 4 4 4 4 + persons
MS OS Developer Employees : comphil -2 2222 persons
Linux OS Capability : comphil 3 OSCUs
MS Windows OS Capability : comphil|_. 14444 OSCUs
Application Developers and Pools
400,000 persons
1,600 VGAUs
Pee
200,000 persons z gh ah ee
800 VGAUs Pte pe
3
Lay
a3
0 persons pte] tat
o veaus _,|,|.lal4 FT _ |
o 4 8 12 16 20 24 28 32 36 40
Time (Quarter)
Linux Application Developers : comphil_ ———+——+_+__+___+ persons
MS Windows Application Developers : comphil 2222 persons
Linux Application Poo! : comphil 3 a— VGAUs
MS Windows Application Pool : comphil. 14444 VGAUs
Installed OS Base
500M OS Units
500M _ OS Units
375M_ OS Units
375M_ OS Units 2]
250M OS Units |
250M OS Units TFA
125M OS Units 2]
125M OS Units =+
0 OS Units
0 OS Units
+ 4 4 aft}
4 8 12 16 2 24 28 32 36 40
Time (Quarter)
Linux in Use : comphil — 4_§_4-_-4_ _¢- ¢ OS Uniti’
MS Windows in Use : comphil OS Units
Figure 10. Output from scenario test run: Communication and Philosophy elements set to 5%
of original values.
