Influence of Free Version upon Pay Version of
High Specialty Software Diffusion
Yutaka TAKAHASHI
School of Commerce, Senshu University
2-1-1, Higashimita, Tama, Kawasaki, Kanagawa, Japan
Telephone +81-44-900-7988 Facsimile +81-44-900-7849
Email takahasi@ isc.senshu-u.ac.jp
Abstract
Today software vendors have various ways to handle their products. In contrast to
non-software products, software products’ properties can be changed or controlled at a
low cost; vendors do not need to extend their operations, and well-designed software
can be customized systematically. Moreover, its diffusion is influenced by its market's
externalities. The diffusion structures of software that have single user type, e.g. World
Wide Web browsers, can be expressed mainly as a single reinforcing loop. This is
because users can be considered as homogeneous. However, there are multiple
possibilities of structure, which corresponds to business strategies for software which
has divided users. An example in this category would be highly specialized software, e.g.
simulation software. We attempt to explain diffusion processes and characteristics
concerning externalities of software in this category using System Dynamics models.
This paper shows that multiple versions can more effectively assist diffusion of main
products than strengthening externality effects.
1. Introduction
Today we can find much software that has several versions irrespective of fields, price
range, or the degree of difficulty in acquisition of skills to use them. For example,
Microsoft is handling two versions of an operating system and several versions of an
office suite.
Many kinds of simulation software are also released with multiple versions. Most of
them are given their own characteristics by which customers can distinguish them in
functions, prices, or additional support services.
Some software products have a pay version and a free version. This is an extreme
pattern of price distinction.
Indeed, there are already products that have similar distribution styles; some
simulation software has a free student version that is allowed to be used by full time
students. Its purpose is to help or promote students to know easily how to use or learn
their product. Their products have been already purchased by schools so that they can
eam profit despite free distribution of the student version. It is reasonable for software
vendors to occupy market as soon as possible because software markets have
“Winner-take-all” rule (Frank and Cook 1998).
This free distribution can be considered as a kind of publicity. In addition, this style of
publicity was only employed for new, or rather novel, software or by new or small
companies. Nevertheless, Microsoft, a huge company, is also handling several versions
of their products. However, they have never released free versions of them. Their lower
price versions are not publicity materials. Rather, they are introduced to the market in
order to meet multiple needs. People who need only basic functions can select lower
price products. Moreover, Microsoft can prevent such people escaping to other
companies’ products or giving up using computers. Hence, we can say that Microsoft
expects not that lower price versions assist higher price products but that they should
meet multiple needs and keep their potential customers.
Today we can find some products which have both pay versions and completely free
versions not for advertisement but for one of regular products. For example, Sun
Microsystems released a pay version of an office suite "Star Office." In contrast,
OpenOffice.org also released a free version of it whose name is "Open Office." Their
names are different, but programs are the same. Indeed, a unitary group
(OpenOffice.org) develops them. Other software vendors are also handling both
versions of one product because of strategic reasons.
Sun Microsystems can obtain some benefits brought by externality effects, which is
the phenomenon that one’s choice affects others’ choices regardless of merchandise’s
position in markets (Arthur 1996). The free version can bring more users, and it can
cause to bring more users to the pay version.However, highly specialized software,
including simulation software, has different properties; their customers are clearly
divided into active users who use the software in order to make something and passive
users who use the software to read the result or are beginners. Their inner
communication (an active user to another active one, or a passive user to another
passive one) and inter categories communication can be different. If this is true, it is
possible for them to be influenced by externalities in a different style or degree from the
Sun Microsystems' case.
This paper shows the models of each strategy and examines them in order to explain
differences between strategies to make software popular using free versions.
2. Methods
Irrespective of strategies to distribute software, there are at least three stock flow
combinations: passive users, active users, and data inventories.
The passive users include a kind of research clients and recipients of reports. The
active users consist of model designers, and code programmers who make workable
data or some programs for their clients. The data inventories are accumulated by
activities of active users.
Since diffusion of software is considered to have externalities, regardless of user types
or software variety, both users have direct feedback loops to themselves. More passive
users produce more passive users in the future. It is possible for followers to have more
information on this software. Specifically, followers can consider that they can avoid
problems. This idea incites them to choose this software, regardless of its characteristics.
Likewise, more active users bring more active users. Followers can think that they can
get useful information to operate this software. Highly specialized software sometimes
requires us to take a long time to become skillful. Therefore, it is a general attitude of
active users to choose existing software with more information, not withstanding the
possibility of some problems regarding its capacity.
These two feedback loops are both reinforcing loops. They are expected to have some
effects including path dependency that brings a robust characteristic of systems'
performances (Sterman 2000).
In contrast, the data inventories have no feedback loops that return directly to the data
inventories. The data cannot increase by themselves; change has to be stimulated by the
active users' activity, i.e. marketing activities, data processing, or programming.
These three stock flow combinations are connected with a feedback loop. Increase in
passive users (e.g. clients) leads to growth of active users (e.g. model builders or
programmers). Increase in active users promotes making data. More data stimulates
passive users’ interest; its popularity brings passive users' trust in results of research and
the software can be employed into their work.
In addition, there is another short feedback loop between passive users and active
users; one encourages an increase in its own number. Passive users tend to choose
contractors only from applicants who use famous or known methods because of their
trust. Markets that have more clients attract active users because the business in such
markets can be more profitable.
These basic conditions produce the causal relation diagram in Figure 1.
In order to verify the effect of free version software release, we add one stock flow
__ Passive users =
Data inventories
y
| Active users =>
Figure 1. Causal relationship Diagram
combination concerning diffusion of free version software. This stock flow combination
is connected with passive users' stock flow combination, not with active users. The
reason for this connection is the assumption that most products of free version software
have limitations in their function because of distinction between free versions and pay
versions, which is often based on strategic or marketing requirements, not on cost of
product development. Because of this, we suppose that free versions are expected more
by beginners or newcomers to its market, compared with active or skilful users.
It is usual to provide some support programs or data through vendors' World Wide
Web pages in the Intemet today. Indeed, most vendors already have their own
distribution bases in the Internet. Therefore, they do not need to have additional
distribution costs for release of free versions. Moreover, software can be easily
controlled by programmers whether a function works or not. This means that making
multiple versions does not require additional costs. These conditions of costs are a
unique characteristic of software and different from other non-software products. Hence,
it is possible to eliminate discussion about the cost for free version release.
3. Simulations and Results
In order to substantiate the effects of the activity of free version releases, the model,
coming from the basic conditions without additional stock flow combinations
conceming free version's diffusion, was tested. This model is shown in Figure 2.
It has external inputs that give both users' characteristics which are how quick
performances were changed. New function releases are done each twelve time-units.
Various sets of values in external input variables reveal that changes of combination
pattems of external input values provide no difference in the result. A typical
performance is shown in Figure 3. Both active users and passive users are constantly
decreasing or remain constant. Pulses are given by new function releases.
Next, a diffusion structure including free version (Figure 4) was examined. This
additional structure, which is inside a dotted box, is connected with stock flow
stopping reading rfte
Passive users
indication by a builder
model ingfeasing rate
‘model building
model builder increasing
education stopping building
new function release span
decision to use as a builder
Figure 2: Stock Flow diagram for the model shown in Figure 1.
1: Models 2: Active users 3: Passive users
2000.00 =
105.00
ena
1050. 00
eon
T 1
0.00 30.00 60. 00 90. 00 120. 00
Figure 3: Transition of models, active and passive users of the model of Figure 2
Free version factor
download ra cision to use as a redder
Downloaded Free Verdion
(@
stopping reading rte
Passive per
download word oNgouth
indication By-a builder
stopping \\eading
| reage increas
model increasing rat Models
reader increasing rate
eminar partfipant
model building
educatfon
Swnera abdgping building rate
new function release span (2)
impact of new functions
model builder increasing
stopping building
decision to use as a builder
Figure 4: Stock flow diagram with the effect of diffusion of a free version
| 15 tode t's 2: Active user 3: Passive user
1 300.00 —
rs 105. 00
3 300. 00
1 200.00
2 90.00 4
3 200. 00
4: 100. 00
2 75. 00
3 100. 00
T
0.00 30. 00 60. 00 90. 00 120. 00
Figure 5: Transition of models, active and passive users of the model of Figure 4
combinations of both users; information that promotes increase in demand for free
versions comes from passive users, and information of output goes to both users.
Various sets of external inputs, the same as the basic condition model's simulation,
were examined. A typical result is shown in Figure 5. Clearly different from the basic
condition model, this model indicates increase of both users after a rapid decrease in the
beginning. All results provide the same tendency. Therefore, this additional structure
works effectively in encouraging expansion of user population.
4. Discussion
It is argued for not only highly specialized software but also general that their
diffusion is largely influenced by externalities. The phenomenon that free version's
diffusion assists pay version's diffusion, examined in the previous section, is a case of
result of externality effect.
Highly specialized software has another possible structure, which can cause
externality effects different from non-specialized software: extension of data
inventories.
The relationship between simulation software and its data is similar to one between
particular hardware and its software. From this point, it is possible that data inventories
can cause some change in a whole system’s performance (Fukunaga et al. 1999).
However, they have one different: speed of data inventory growth in high specialty
software market and a number of software titles for particular hardware. The former is
directly linked to user population so that user population and data inventories growths
are nearly simultaneous. In contrast, the latter has time delays between software releases
and hardware sales growths. Therefore, it is meaningful to examine the performance of
the high specialty software market structure.
In order to verify the effect of extension of data inventories, an acceleration structure
of model accumulation is required.
The case that active users make models more actively has been already examined; it
can be expressed by the basic condition model with control of external inputs. This
change of scenario (and value of external inputs) showed no different result from other
simulations using the basic condition model.
There is another possible scenario: converting other software data to relevant software.
This scenario requires extending the basic condition model's structure so that it is
possible to cause a new result. The model is shown in Figure 6. This is also an extension
of the basic condition model (Figure 2), the same as the free version effect model
(Figure 4).
indication by a builder
Models
target models
stopping building
decision to use as a builder
new functioyi release span
Figure 6: The model with the externality effect
One new feedback loop that consists of all stock flow combinations appeared.
Various sets of parameters were examined. As a result of these simulations, this
extension (or scenario change) has no effect on the performance of this system.
The difference between the free version model (Figure 4) and data conversion model
(Figure 6) is not concerned with a number of reinforcing feedback loops but with a
position of each reinforcing feedback loop. Indeed, they have the same number of
reinforcing loops. Moreover, both models have similar stock flow structures. However,
the free version model has a short cycle that consists of "demand of free version," while
the data conversion model has a long cycle that consists of "conversion." Although the
latter includes whole stock flow combination, it has no effect. In contrast, the former
includes only half of the elements in the model. Nevertheless, the elements drive the
whole system.
Consequently, this paper has shown that free version software diffusion without output
products (e.g. models) can assist more effectively the pay version's diffusion than
actively making more output products. It can be also said that the switch of extemality
effects can be hidden in the cases of having multiple user layer, e.g. high specialty
software diffusion.
5. Acknowledgement
I would like to thank Mr. David Watts (Graduate School of Education, University of
Bristol) for his constructive comment. This research is a part of my productions of
“Senshu Research A broad Program 2004.”
References
Arthur, W. B. 1996. Increasing Returns and the New World Business, Harvard Business
Review, July- August 1996.
Frank, R. H. and Cook, J. 1998. The Winner-take-all Society. Simon and Schuster Inc.
1998.
Fukunaga, Y., Takahashi, Y., Tanaka, N., Uchino, A., and Morita, M. System Dynamics
Analysis of Network Extemality in Complex Market Structure Part II,
Proceedings of 1999 International System Dynamics Conference.
Kim, D and Anderson, V. 1998. Systems Archetype Basics. Pegasus Communications.
Ruth, M. and Hannon, B. 1997. Modeling Dynamic Economic System. Springer Verlag.
Senge, P. M. 1992. The fifth discipline. Doubleday.
Senge, P. M., Ross, R., Smith, B., Roberts, C., and Kleiner, A. 1994. The Fifth
Discipline Fieldbook. Doubleday.
Sterman, J. 2000. Business Dynamics. McGraw Hill.
Takahashi, Y and Tanaka, N. 2000. A Construction and an analysis of Lock-in Model
under Effect of Software Distribution. Journal of Japan Society of Business
Mathematics, Vol. 22, No. 1.
Notes
Each software product shown in this paper can be found in vendors World Wide Web
pages below.
Microsoft OS and other products: http://www.microsoft.com
OpenOffice.org: http://www.openoffice.org
Sun Microsystems Star Suite: http://www.sun.com/software/star/staroffice/index.xml
CC BY-NC-SA 4.0