AN APPROACH FOR IMPROVING THE SOCIAL ASPECTS

OF THE SOFTWARE DEVELOPMENT PROCESS

BY USING A GAME THEORETIC PERSPECTIVE

Towards a Theory of Social Productivity of Software Development Teams

Murat Yilmaz

and Rory V. O’Connor

Lero Graduate School in Software Engineering, Dublin City University, Dublin, Ireland

Lero, the Irish Software Engineering Research Center, Dublin City University, Dublin, Ireland

Keywords:

Software process improvement, Team dynamics, Game theory, Mechanism design, Research framework,

Software engineering economics, Productivity, Social aspects of development.

Abstract:

As software development is considered a form of knowledge based social activity, investigating social in-

teractions and behaviors of individuals and teams constitutes a starting point for improving organizational

performance and productivity. Therefore, a software development organization is regarded as a form of social

network, which may be more efﬁciently structured based upon participants’ skills, roles and capacities to ex-

change information. This paper aims to propose a research framework for modeling development activities in

terms of social interrelationships. It investigates ways for improving the productivity of the software develop-

ment process to several social issues (e.g. team formations, interpersonal conﬂicts, social loaﬁng) that affect

the group setting during the software development process. An industrial survey will be conducted to vali-

date the proposed framework. This survey will be applied to three medium size software organizations; (i) to

measure social aspects of productivity, (ii) to assess the effectiveness of our approach, (iii) and to improve the

welfare of the software development organization. Semi-structured interviews with practitioners, and expert

reviews with the managers will be used to evaluate the results. Ultimately, based on the principles of our game

theoretical approach and collected data, we propose a research framework, which will beneﬁt future research

by drawing a road map that establishes a body of knowledge, speciﬁcally on software development teams and

organizations.

1 INTRODUCTION

This paper aims to establish a research framework

based on the importance of understanding soft-

ware development as a knowledge-based social activ-

ity (Dittrich et al., 2002). A knowledge-based view-

point suggests that the intangible value of an orga-

nization is created as a deployable knowledge as-

set (Grant, 2002), where it is formed from the out-

come of the interactions of participants. This value is

somehow stored by its teams and individuals in terms

of intellectual (Rus and Lindvall, 2002) and social

capital (Lin, 2002). Therefore, the structuring of indi-

viduals’ roles in a software organization with respect

to the quality of information exchange among the in-

dividuals should be considered more important than

several other factors affecting the productivity of a

software development organization. However, appro-

aches (e.g. game theoretical) that employ the roles of

individuals and analyze the outcomes of their social

interactions in team-based activities have not yet been

investigated in the ﬁeld of software development.

An initial exploration of the importance of social

aspects of software development (Acuna et al., 2005)

highlights the fact that modeling individuals’ inter-

actions (Acuna and Juristo, 2005) and their motiva-

tions (Beecham et al., 2008) to achieve economic and

social outcomes is beneﬁcial for maximizing the op-

portunities for improving the software development

process. There is a body of research that has rein-

forced the view that social networks (i.e. network of

social interactions) can be considered as an important

factor for investigating organizational issues, in par-

ticular when they are used for considering the struc-

ture and dynamics of a software development orga-

nization (Ohira et al., 2006; Nielsen and Tjornehoj,

Yilmaz M. and V. O’Connor R..

AN APPROACH FOR IMPROVING THE SOCIAL ASPECTS OF THE SOFTWARE DEVELOPMENT PROCESS BY USING A GAME THEORETIC

PERSPECTIVE - Towards a Theory of Social Productivity of Software Development Teams.

DOI: 10.5220/0003492900350040

In Proceedings of the 6th International Conference on Software and Database Technologies (ICSOFT-2011), pages 35-40

ISBN: 978-989-8425-76-8

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

2010). Furthermore, the ability for a software de-

velopment organization to understand the factors that

hinder productivity is essential in improving the com-

petitive advantage for the business success (Chemu-

turi, 2009), where it is not possible to think about

productivity regardless of social issues (i.e. matters

that affect individuals and teams) affecting the devel-

opment process.

The remainder of this paper is organized as fol-

lows. In section 2, we provide a brief overview of

research methodologies. In section 3 we establish a

context for our research framework. Based on the re-

search questions and the context of the research, in

section 4, three hypotheses are proposed. Finally, sec-

tion 5 summarizes the research and section 6 shows

future work and expected outcomes.

2 A BRIEF OVERVIEW OF

RESEARCH

METHODOLOGIES

Although there are several different research method-

ologies used in software engineering research, all can

be classiﬁed in two main categories: quantitative re-

search (e.g. survey research, experiments, and simu-

lations) and qualitative research (e.g. grounded the-

ory, ethnography, action research) (Dawson, 2002).

Based on the idea of seeing the human as an in-

strument and considering their experiences, qualita-

tive research aims to investigate participants actions

and words by explanatory patterns of meaning (Stake,

2010). It studies mental attitudes and social behaviors

frequently recorded as information from the partici-

pants own words and deﬁnitions, and classify them

from their natural work (settings) environment. Qual-

itative researchers use techniques like interviews, in-

dividual experiences, case studies and focus groups to

capture data about the values and emotions of people

for investigative observations. The collected data can

be documented in a contextual framework for con-

ducting a closer observation of words and view of the

participants and further inspection (Hesse-Biber and

Leavy, 2010).

On the other hand, quantitative research is based

on the idea of quantifying the interrelations between

different types of variables (i.e. independent, depen-

dent) using numerical techniques. It is corrobora-

tive and based on conventional and systematical pro-

cesses to gather data, which describes the information

by cause and effects relations. Quantitative research

aims to create theories, hypotheses and mathematical

models based on empirical investigations (Balnaves

and Caputi, 2001).

A blending of different qualitative and quantita-

tive methods and approaches (or their variants) as a

technique to improve the validity of research ﬁnd-

ings and conclusions is called mixed method research.

Triangulation (Cook and Campbell, 1979) is one of

the most well-known mixed method design strate-

gies. This approach advocates that multiple types of

data and methods can support hypotheses, incident or

events. In other words, the name triangulation is a

metaphor which describes the use of multiple meth-

ods in a research: a combination of two research tra-

ditions (Jick, 1979); i.e. theories, data sources, meth-

ods or techniques to study a research question or to

measure a single construct.

3 CONTEXT OF CURRENT

RESEARCH

As it is suitable for empirical or applied software en-

gineering research, our approach is an attempt to in-

vestigate the inﬂuence of social factors over the soft-

ware production process. In the hope that we will be

able to improve the social structure of a software or-

ganization and therefore to maximize the productivity

of software teams. With these in mind, our goal is to

develop a team composition framework (i.e. a game

theoretical team interaction model) based on the de-

tails (e.g. interactions and types) of the participants.

To facilitate the visualization of our model fur-

ther, on the one hand, we need qualitative analysis

techniques for gathering scenarios of interaction from

participants. On the other hand, we need quantitative

methods for establishing a rigorous mathematical for-

mulation for investigating several factors, creating a

scenario based team formation model, and use them

concurrently for testing our theories and hypotheses.

By combining these two approaches systematically,

our key idea here is to build a more efﬁcient perspec-

tive.

3.1 Game Theoretic Personality Types

Before introducing game theoretic personality types,

we ﬁrst brieﬂy survey some similar concepts used in

personality proﬁling in research.

The theory of personality types was introduced by

Carl Jung which is based on the classiﬁcation of peo-

ple with distinctive set of characteristics found in in-

dividual’s personalities. Based on the theory of psy-

chological type (Jung et al., 1991), which states that

there are four types of personality dimension, a ques-

tionnaire called Myers-Briggs Type Indicator (MBTI)

ICSOFT 2011 - 6th International Conference on Software and Data Technologies

was created. An aim is to operationalize Jung’s work

in order to understand how people make decision re-

garding to their personalities and categorize them in

16 different psychological types (i.e. types are con-

sidered a combination of four preferences) (Myers

et al., 1999). Moreover, Keirsey has conducted em-

pirical investigations on the human psychology based

on a variant of Jung’s theory (Keirsey et al., 1984). In

particular, he worked on a temperament sorter about

personality types where he also found out 16 differ-

ent forms. Unlike Myers who investigates people’s

feeling, his research is directly focused on observable

behaviors of humans.

The correlation between personality of actors and

their productivityﬁrst indicated by Weinbergand later

investigated by Shneiderman for its role among the in-

teractions of software development (Weinberg, 1971;

Shneiderman, 1980). Da Cunha and Greathead claim

that types of personality and skills of software devel-

opment personnel can be aligned to advance the pro-

ductivity (Da Cunha and Greathead, 2007). Beecham

et al. identify several social factors effecting soft-

ware engineer characteristics (Beecham et al., 2008).

For example, software practitioners prefer challeng-

ing tasks and therefore recognition in their work, sta-

bility in their organization, and most importantly they

prefer to be socially identiﬁed with their team or

group (Beecham et al., 2008).

Although several research was conducted on per-

sonality types most of these works use MBTI. (e.g

(Capretz, 2003; Da Cunha and Greathead, 2007;

Gorla and Lam, 2004)), there is no body of research

for identiﬁcation of personality types applicable to

game theoretical approaches. We suggest that, based

on our game theoretic model, these types should

be compatible with our model for identiﬁcation of

people’s behavior with respect to several situations.

Therefore, we propose creation of socio-types regard-

ing to variety of interactions based on a game theo-

retical setting. Therefore, we start our framework by

hypothetical game theoretic personality types (GTPT)

as; (i) cooperative, (ii) competitive, (iii) socially co-

operative, (iv) socially competitive, (v) individualis-

tic.

4 RESEARCH QUESTIONS AND

HYPOTHESES

In response to the problems highlighted, in this sec-

tion we develop a list of research questions and hy-

pothesis so as to conduct our research and we employ

techniques that can be used to evaluate our conceptual

propositions with respect to the accuracy of data col-

lected. We test our hypotheses and analyze the data

gathered from selected software organizations as the

study of practical cases.

Research Question 1. Will assigning individ-

uals to the appropriate roles with respect to

their GTPTs improve software team produc-

tivity?

Research Question 2. Can a correlation be-

tween productivity and social productivity be

identiﬁed for software development?

The goal of the ﬁrst research question is to ini-

tiate a framework to expand a mapping of technical

roles and align it with participants’ game theoretic

types. This question stems from the fact that when

characteristics of individuals and their technical roles

are better blended, the productivity of team will in-

crease (Acuna et al., 2006; Capretz, 2003).

Brandenburger and Nalebuf (Nalebuff and Bran-

denburger, 1996) have argued that there is a visible

interconnection between the actors of production pro-

cesses (e.g. client and vendor) which should be high-

lighted for developing better responses to the market

demands. In the light of this argument, the second

research question distinctively seeks possible conﬁg-

uration of intellectual (i.e. human and social) capi-

tal of software development organization to improve

production rate of software development. Therefore,

a particular aim here is to expand the deﬁnition of

software development productivityto include the con-

cerns regarding to social issues for individuals and

teams in a software development organization. In par-

ticular, we deﬁne social aspects affecting the produc-

tivity as social productivity, i.e. the production rate of

software development increases if we give due con-

sideration to maximizing the social relations. This re-

search is concerned with identifying the relationship

between productivity and its social aspects and hence

it has been developed to support this endeavor.

To seek answers to these questions, we have es-

tablished and formalized our ﬁrst hypothesis guided

by our research agenda.

Hypothesis 1. Improving the social structure

of a software organization will improve the

productivity of software development teams.

The second group of research questions seek to

uncover the ways to improve the formation of soft-

ware teams by using GTPTs. We suggest that the

information exchange problem among software team

members needs to be identiﬁed and converted into a

game theoretic view by combining these information

with game theoretic types of interacting individuals.

Research Question 3. Can organizing in-

dividuals regarding to their personality types

AN APPROACH FOR IMPROVING THE SOCIAL ASPECTS OF THE SOFTWARE DEVELOPMENT PROCESS BY

USING A GAME THEORETIC PERSPECTIVE - Towards a Theory of Social Productivity of Software Development

Teams

help to improve the social dynamics of a soft-

ware organization by designing a mechanism?

Research Question 4. Can we formulate a

model and simulate software team composi-

tions and improve the productivity of software

development teams?

The second hypothesis relies on the argument that

it should be possible to use our concept of game theo-

retical personality types to improve several aspects of

software development teams.

Hypothesis 2. Software development teams

that are formed by using GTPTs will perform

better than those structured with an ad-hoc

method.

The next group of hypothesis and research ques-

tions aim to facilitate hypothesis 2 so as to expand to

include social network analysis while seeking ways

for maximizing the collective outcome. Therefore,

the structure of hypothesis 3 and the subsequent two

research questions will be built on hypothesis 2. The

expected achievement here is to ﬁnd techniques to vi-

sualize a software organization and behaviors and test

the distinctive GTPTs have a possible impact on so-

cial network structure. For example, by investigating

and visualizing the social structure of an organization;

(i) key participants can be identiﬁed, (ii) the informa-

tion ﬂow inside the social network can be analyzed,

(iii) the outcome of the interaction of different actors

can be highlighted,..etc.

Research Question 5. Can we collectively

use both social network analysis and person-

ality types together to maximize the software

development team productivity?

Research Question 6. Does revealing and ex-

amining the positions of participants in a so-

cial structure can help us to improve the team

productivity?

A frequently asked question at social and eco-

nomic research is the impact of revealing the posi-

tions of participants on tiers of a network. Therefore,

we suggest it is important to show the inﬂuence of in-

dividuals and their social circles at a conceptual level.

Hypothesis 3. A combination of social net-

work analysis together with the individuals

personality types leads us to understand the

potential requirements for structural improve-

ments of software teams.

Based on an information exchange economy, we

consider that the organizational structure of software

development is an intellectual effort based on in-

terconnected networks of individuals, their actions

and relationships. Consequently, the third hypothesis

states that a social network analysis should be helpful

to investigate the impacts of participants interactions,

and as a result how these social interactions and the

structure affect team formations and the productivity

of software development. Therefore, we plan to use

SNA with GTPTs together for team optimization and

to maximize the outcomes of a production process.

In conclusion, our aim is to generalize our theoretical

ﬁndings with empirical observations from the ﬁeld.

5 THE RESEARCH

FRAMEWORK

Figure 1: The framework for the proposed research.

Here, we formalize a research framework to conduct

our research (see ﬁgure 1). As outlined earlier in this

paper, the primary goal is to create a team produc-

tivity model for software development organizations

based on social and economical factors and therefore

to analyze the impact of social factors on the produc-

tivity of software organizations from a game theoret-

ical perspective. We introduce the concept of social

productivityand further we use it to identify the social

factors and its relationship with software productivity.

ICSOFT 2011 - 6th International Conference on Software and Data Technologies

Initially, we begin the ﬁrst part of the research by

creating a survey instrument based on game theoret-

ical personality types of individuals and determine

their types and degree of interactions. We are con-

ducting a research on the game theoretical personal-

ity types and further we will have them reviewed by

experts from the ﬁeld for evaluation. Secondly, we

continue our work by analyzing the correlation be-

tween productivity and social productivity by using

regression techniques and to construct an empirical

relationship model.

Thirdly, we initiate our investigation of several sit-

uations and based on this information we evaluate the

identiﬁcation of our game theoretic personality types.

To store this information for future use, we plan to

create situational context cards. Fourth, we claim that

an empirical investigation of the notion of personal-

ity types and their variants of compositions need to be

identiﬁed by their location and role in a social struc-

ture of a software company. Therefore, we formal-

ize the social structure of a software organization by

using sociometric techniques such as social network

analysis for examining relation of personality types

and other social aspects such as task roles inside an

information exchange network.

Fifth and ﬁnally, we aim to simulate the output of

this research by creating a team composer software

prototype for better composition of teams. We argue

that a simulation (which has been used frequently to

create social and economic modeling (Gilbert, 2008))

of team compositions will be a suitable way to per-

form several scenarios to investigate the social dy-

namics of software development. In other words, we

suggest simulation is a feasible approach for captur-

ing interactions for this emergent phenomena from

the viewpoint of its interacting units. Because it sup-

plies important information on a system in virtual

form, we can simulate dynamic scenarios in a vir-

tual landscape. It will, therefore, allow us to organize

more productive software teams.

6 CONCLUSIONS AND FUTURE

WORK

A signiﬁcant challenge in the software engineering re-

search is to model software development as a social

and economic activity. Based on the deﬁned tasks of

development, an organization may have several types

of participants with distinctive roles. As we have pro-

posed, one way of improving the social productivity

of a software organizationis to improve the productiv-

ity by exploring people’s interactions, behaviors and

to design its outcomes.

The focal point of this research is based on game

theoretical investigation of a software development

projects as it aims to construct a model for software

development as a (cooperate) game of incomplete in-

formation. As it has been observed that organization

of software development can be very sensitive to sev-

eral types of market forces (Madhavji et al., 2006),

organizations can beneﬁt from a model of a software

organization for better decision making about their

social and organization issues. Further, we will com-

pare our game theoretic personality types with Myers-

Briggs Type Indicator (MBTI) (Myers et al., 1999)

and Belbin’s team roles (Belbin, 2010) so as to iden-

tify how our particular part of research intersects with

the previous body of knowledge.

We aim to conduct our research on three medium

size software companies. Initially, we interview some

experts from these companies. However, there are

several parameters may hinder our expected results.

For example: variations of team size and teams work-

ing in different domains. Therefore, our prelimi-

nary planing includes testing our approach on three

software teams consisting of forty people within the

same company for acquiring more homogeneous re-

sults. Several social situations and scenarios can be

captured at different levels of software development

organization. And, these scenarios can be depicted

by using use case diagrams. Moreover, the work-

ing scenarios can be adapted to a simulation model

and rigorously analyze to gather more information

about potential alternative cases. This study aims to

use computer simulation techniques to perform sev-

eral hypothetical scenarios based on a team environ-

ment. To improve the productivity of software devel-

opment teams, we compare virtual events with actual

outcomes that are exercised by team members on sce-

nario based situations. Furthermore, we suggest that

situational outcomes of a computer simulation should

also be discussed with management teams of targeted

software companies by conducting expert reviews.

ACKNOWLEDGEMENTS

This work is supported, in part, by Science Foun-

dation Ireland grant number 03/CE2/I303-1 to

Lero, the Irish Software Engineering Research Cen-

tre (www.lero.ie).

AN APPROACH FOR IMPROVING THE SOCIAL ASPECTS OF THE SOFTWARE DEVELOPMENT PROCESS BY

USING A GAME THEORETIC PERSPECTIVE - Towards a Theory of Social Productivity of Software Development

Teams

REFERENCES

Acuna, S. and Juristo, N. (2005). Software process model-

ing. Springer Verlag.

Acuna, S., Juristo, N., and Moreno, A. (2006). Emphasiz-

ing human capabilities in software development. Soft-

ware, IEEE, 23(2):94–101.

Acuna, S. T., Juristo, N., Moreno, A. M., and Mon, A.

(2005). A Software Process Model Handbook for In-

corporating People’s Capabilities. Springer-Verlag

New York, Inc.

Balnaves, M. and Caputi, P. (2001). Introduction to quan-

titative research methods: An investigative approach.

Sage Publications Ltd.

Beecham, S., Baddoo, N., Hall, T., Robinson, H., and

Sharp, H. (2008). Motivation in software engineer-

ing: A systematic literature review. Information and

Software Technology, 50(9-10):860–878.

Belbin, R. (2010). Management teams: why they succeed

or fail. Butterworth-Heinemann.

Capretz, L. (2003). Personality types in software engineer-

ing. International Journal of Human-Computer Stud-

ies, 58(2):207–214.

Chemuturi, M. (2009). Software Estimation Best Practices,

Tools & Techniques: A Complete Guide for Software

Project Estimators. J. Ross Publishing.

Cook, T. and Campbell, D. (1979). Quasi-experimental de-

sign and analysis issues for ﬁeld settings. Chicago:

Rand Mc-Nally College Publishing Co.

Da Cunha, A. and Greathead, D. (2007). Does personality

matter?: an analysis of code-review ability. Commu-

nications of the ACM, 50(5):109–112.

Dawson, C. (2002). Practical research methods: a user-

friendly guide to mastering research techniques and

projects. How To Books Ltd.

Dittrich, Y., Floyd, C., and Klischewski, R. (2002). Social

thinking-software practice. The MIT Press.

Gilbert, G. (2008). Agent-based models. Sage Publications,

Inc.

Gorla, N. and Lam, Y. (2004). Who should work with

whom?: building effective software project teams.

Communications of the ACM, 47(6):79–82.

Grant, R. (2002). The knowledge-based view of the ﬁrm.

The strategic management of intellectual capital and

organizational knowledge, pages 133–148.

Hesse-Biber, S. and Leavy, P. (2010). The practice of qual-

itative research. Sage Publications, Inc.

Jick, T. (1979). Mixing qualitative and quantitative meth-

ods: Triangulation in action. Administrative science

quarterly, pages 602–611.

Jung, C., Baynes, H., and Hull, R. (1991). Psychological

types. Routledge.

Keirsey, D., Bates, M., and Company, P. N. B. (1984).

Please understand me: Character and temperament

types. Prometheus Nemesis Del Mar, Calif.

Lin, N. (2002). Social capital: A theory of social structure

and action. Cambridge Univ Pr.

Madhavji, N., Lehman, M., Perry, D., and Ramil, J. (2006).

Software evolution and feedback. Wiley Online Li-

brary.

Myers, I., McCaulley, M., Quenk, N., and Hammer, A.

(1999). MBTI manual. Consulting Psychologists

Press.

Nalebuff, B. and Brandenburger, A. (1996). Co-opetition:

A Revolution Mindset That Combines Competition

and Co-operation: The Game Theory Strategy That’s

Changing the Game of Business. Doubleday Book,

Newy York.

Nielsen, P. and Tjornehoj, G. (2010). Social networks in

software process improvement. Journal of Software

Maintenance and Evolution: Research and Practice,

22(1):33–51.

Ohira, M., Ohoka, T., Kakimoto, T., Ohsugi, N., and Mat-

sumoto, K. (2006). Supporting knowledge collabora-

tion using social networks in a large-scale online com-

munity of software development projects. In Software

Engineering Conference, 2005. APSEC’05. 12th Asia-

Paciﬁc, page 6. IEEE.

Rus, I. and Lindvall, M. (2002). Guest editors’ introduc-

tion: Knowledge management in software engineer-

ing. IEEE software, pages 26–38.

Shneiderman, B. (1980). Software psychology: Human fac-

tors in computer and information systems (Winthrop

computer systems series). Winthrop Publishers.

Stake, R. (2010). Qualitative Research: Studying How

Things Work. The Guilford Press.

Weinberg, G. (1971). The psychology of computer program-

ming. Van Nostrand Reinhold New York.

ICSOFT 2011 - 6th International Conference on Software and Data Technologies