Small Business & Entrepreneurship

A Survey on the State of the Practice in Video Game Software Development

Description
A Survey on the State of the Practice in Video Game Software Development
Published
of 13
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Technical Report A Survey on a State of the Practice inVideo Game Development Jürgen Musil 1  Angelika Schweda 1  Dietmar Winkler 2  Stefan Biffl 2   Institute of Software Technology and Interactive SystemsFavoritenstr. 9-11, A-1040 Vienna, Austria 1 {jmusil, angelika} @computer.org  2 {dietmar.winkler, Stefan.biffl} @tuwien.ac.at  Technical Report No. IFS-QSE 10/04Issued: March 2010  A Survey on the State of the Practicein Video Game Software Development Juergen Musil 1 , Angelika Schweda 1 , Dietmar Winkler 2 , Stefan Biffl 2 Institute of Software Technology and Interactive SystemsVienna University of Technology 1 {  jmusil, angelika } @computer.org, 2 { Dietmar.Winkler, Stefan.Biffl } @tuwien.ac.at Abstract Video Game Software Development is a promising area of empirical research because our first obser-vations in industry environment identified a lack of a systematic process and method support and rarely conducted/documented studies. Neverthe-less, video games - specific types of software prod-ucts - focus strongly on user interface and gamedesign. Thus, engineering processes, methods for game construction and verification/validation, and best-practices, derived from traditional software en-gineering, might be applicable in context of videogame development. We selected the Austrian gamesindustry as a manageable and promising starting point for systematically capturing the state-of-thepractice in Video game development. In this paper we present the survey design and report on the first results of a national survey in the Austrian gamesindustry. The results of the survey showed that theAustrian games industry is organized in a set of small and young studios with the trend to ad-hocand flexible development processes and limitationsin systematic method support. 1 Introduction Video game software development (VGSD) hasmaturated within only five decades from the hobbyof talented computer scientists into a billion dollarindustry that today easily outperforms the moviesindustry in gross revenue. Top game softwareprojects have become artistic, technological and fi-nancial challenges with development costs rangingfrom few million up to 100 million dollars 1 . Ac-cording to the recent PricewaterhouseCoopers me-dia forecast 2 , the video game business is predicted a 1 Digital Battle. http://www.digitalbattle.com/2010/02/20/top-10-most-expensive-video-games-budgets-ever(last visited 03/01/2010). 2 PricewaterhouseCoopers. Global entertainment and media outlook 2009-2013  . PricewaterhouseCoopers LLP. constant annual growth through 2013, seeing videogames on top in home entertainment spending. Itseems that such results draw a promising futurefor digital games, the development and productionpractices of the games industry are far from theprofessional level that sales figures may suggest.Game developers have to face workflow integrationproblems, difficulties in requirement elicitation andworkload fluctuations due to a lack of verified do-main tailord processes and techniques. Althoughthe video games industry seems rather turbulent, itpresents itself as an interesting enviroment for em-pricial software engineering that varies significantlyfrom traditional software domains. Video gamesare interactive entertaiment products of high sys-tem complexity that are built upon multiple com-puter science disciplines including artificial intelli-gence, computer graphics, distributed systems andhuman computer interaction. Their production istime critical and involves multiple heterogeniousdisciplines including experts from non-engineeringdomains (e.g. arts, design, animation, creativewriting). Because games premier purpose is amuse-ment, validation of their entertaining impact relieson qualitative and quantitvative methods 3 . A goodexample demonstrating that software engineeringcan indeed learn things from video game develop-ment is e.g. in the area of rapid development of new software products [25].At the example of a state of the practice surveyin the Austrian games industry, we therefore seekto investigate if empirical/evidence-based softwareengineering can be of assistance in providing so-lutions for problems in VGSD and in identifyingsimilarities and differences between game and non-game software development. Furthermore, the re-sults can (a) support video game producers in in-troducing lessons learn- ed from traditional soft- 2009. 3 Microsoft Game Studios, Games User ResearchGroup. http://mgsuserresearch.com/publications/ (last vis-ited 03/01/2010). 1  ware engineering to increase process and productquality and (b) can be the starting point for amore detailed analysis of video game developmentpractice in a larger context. We are interested if conflict areas in processes and techniques in VGSDcould be found that would be possible candidatesfor techniques from existing best-practice softwareengineering. Also we seek to retrieve hands-on in-formation about games industry-specific problemscalled ”feature creep” and ”crunch time” and theircauses.The remainder of this paper is structured as follows:Section 2 presents related work on video game de-velopment and best practice software engineeringapproaches. Section 3 illustrates the research issueswith focus on best-practices in video game develop-ment. We described the study design of the surveyin section 4 and provide initial results on the surveyin section 5. Section 6 presents the discussion. Fi-nally, section 7 concludes and identifies next stepsfor future work. 2 Related Prior Work This section summarizes related work on VGSDand best-practices software engineering approacheswith a particular focus on: (1) Examining thestate of empirical studies in game software devel-opment processes. (2) Identifying major challengesin game software development and existing conflictareas. (3) Discussing the contemporary importanceof best-practice software engineering. 2.1 Existing Empirical Studies We have reviewed existing literature with particu-lar attention on empirical studies of game develop-ment practices and processes. Unexpectedly, wehave identified a significant lack of developmentprocess-focused publications and empirical studies,after investigating major computer science-relatedacademic platforms (ACM, IEEE, Springer) as wellas game industry/game studies related portals (DI-GRA, Game Studies, Gamasutra). The only cur-rent study among game studios that is methodicallycomparable with our presented work is the Stateof Game Development 2010 Survey  4 from ThinkServices. The difference between our research andthe Think Services survey is that our survey aimsto cover the Austrian game studios and to recheckresults from existing studies. Whereby the ThinkServices survey has been conducted among clients 4 Game Developer Research. State of Game Development 2009-2010 Survey  . UBM Think Services. San Francisco,CA, 2010. of Think Service products and questions aim pre-dominantly demographics, hardware/software toolusage, project budgeting and purchasing behavior.A survey [16] dealing exclusively with game devel-oper demographics has been conducted by the In-ternational Game Developers Association (IGDA)in 2005.Since there is no widely validated body of knowl-edge in VGSD available to date, we therefore alsopropose to follow the five steps of evidence-basedsoftware engineering according to [10, 20]:1. Converting the need for information into ananswerable question.2. Tracking down the best evidence with whichto answer the question.3. Critically appraising that evidence for its va-lidity, impact, and applicability.4. Integrating the critical appraisal with softwareengineering expertise and stakeholders’ values.5. Evaluating effectiveness and efficiency in exe-cuting steps 1-4 and seeking ways to improvethese. 2.2 Video game development Peltoniemi [26] evaluated the maturation process of the game sector and concludes that the game devel-opment sector after decades is still less developedthan the automobile industry in its most turbulentyears of industry formation. One reason is thatvideo game development has become increasinglychallenging over the last decade [9, 12] with globalcompetition of creative concepts and a strong in-crease in systems complexity [15]. Games industryveteran Jonathan Blow has identified the follow-ing obstacles in current video game development[9]: (a) workflow problems in programming, testingand content production (e.g. 3D modeling, gamelevel design), (b) problems because of the concur-rent development for multiple target platforms (e.g.PC, Playstation 3), (c) integration problems dueto a lack of domain-specific software patterns anda dependence on third-party components/game en-gines in order to maintain project timeliness, (d)project risks from trade-offs between game designand product-value based considerations that Blowsummarizes with the questions ”how will this never-implemented feature feel to the end user? Is it go-ing to be worth [...] to implement it?” [9].Petrillo et al. [27] note that many problems of the software industry occur in the electronic gamesindustry as well. In their postmortem analysisPetrillo et al. identified, besides technological prob-lems that have been discussed by Blow [9], also2  scheduling problems and problems of scope:Scheduling problems occur in the form of classicalproject delays and crunch time, but are also oftencontributory caused by technological problems [27]. Crunch time is a video games industry term forphases of extensive overtime work lasting from oneor two weeks up to some months in the worst case.Crunch time is regarded as a serious and endur-ing problem by developers 5 that has led to lawsuitsagainst large game companies [2]. The most recentincident 6 has again confirmed Peltoniemi’s theoryabout the video games industry’s maturation level[26] and nurtures assumptions about deep-rootedproduction process difficulties.Problems of scope occur in the form of unrealisti-cally large project size, design problems, the cut-ting of features and a behavior that is called fea-ture creep [15, 27]. Chandler [12] notes on fea-ture creep that it ”occurs if additional features areadded without adjusting the other project variables(time, resources and quality) to accommodate theadditional work”. Frequent and late changing re-quirements hinder systemic approaches (V-ModelXT) and foster a trend to agile game development(e.g. Scrum) instead.Project scope problems have also been confirmedby Callele et al. [11] who identified the followingareas where VGSD benefits from requirement en-gineering and project management best-practices: ”(1) communication between stakeholders of dis-parate background, (2) remaining focused on thegoal and resisting feature creep, (3) influence of prior work (e.g. building a new game on top of an existing game), (4) media and technology inter-action and integration, (5) the importance of non- functional requirements, and (6) gameplay require-ments” [11]. Callele et al. note that the dominationof non-functional-requirements and especially thecategory of gameplay requirements are character-istic for the game software domain and that theseaspects require further investigation due to a lackof foundation research [11].Another inherent challenge is heterogeneity of disciplines (e.g. arts, engineering) that collabora-tively practice game development and arising in-tegration conflicts [15]. On the example of thegame title Dragon Age: Origins we demonstratethe semantic ambiguity by the usage of the word design  : software system design, user interface de-sign, game design, art direction, animation and 5 Grant, C. Epic’s Mike Capps responds to accusationsof ’exploitative’ working conditions. In Joystiq  . We-blogs Inc. Network. 22. April, 2009. Available athttp://www.joystiq.com (last visited 03/01/2010). 6 Rockstar Spouse. Wives of Rockstar San Diego em-ployees have collected themselves. In Gamasutra  . UnitedBusiness Media LLC. 7. January, 2010. Available athttp://www.gamasutra.com (last visited 03/01/2010). cinematic design, audio/voice-over direction. Al-though the previous list is not exhaustive it stillgives the impression that it resembles more activ-ities required for a hollywood movie, it is a factthat video games have arrived manufacturing com-plexity and budgeting of even such. The signifi-cant increase in production quality over the previ-ous years becomes particularly apparent when com-paring material of the movie Final Fantasy: TheSpirits Within  and the state-of-the-art game title Final Fantasy XIII  . Peltoniemi [26] concludes thatwhat makes the games industry different is the cre-ation of non-utilitarian products and proposes theclassification of the game sector in the category of cultural and creative industries (e.g. movies, mu-sic) that have the following key economic charac-teristics: monopolistic competition and horizontaldifferentiation, hits and misses, increasing returns,gatekeepers, taste formation and experience goods,skewed labor markets, majors and independents[26]. The circumstance that the games industry isderived from the software industry but located inanother industry family that produces a differenttype of goods hinders direct adoption of proven,established software development approaches andis a main reason for many misunderstandings be-tween these professional domains. Concluding, thekey challenges for empirical research in video gamedevelopment can be summarized as:1. Integration problems in workflow due to de-velopment across heterogeneous disciplines de-mand tailored process support for arts and en-gineering domains.2. Requirement elicitation problems because of an emphasis on non-functional requirements,”gameplay” requirements and a dominance of audio-visual, narrative and interactive aesthet-ics in the end product.3. Identification of methods that address domain-specific risk assessment, value-based consider-ation and verification and validation.4. Challenges for product family engineering,since current installments of video games seriesare concurrently developed for multiple (het-erogeneous) target platforms and have harddeadlines (e.g. Christmas season).Empirical software engineering can play an impor-tant role in identifying practices and processes thataddress these challenges.3  2.3 Best-practice software engineer-ing The main goal of business IT software development,e.g. database driven systems and web applications,is the construction of high-quality software prod-ucts [1, 29]. In contrast to game development, busi-ness IT development focuses on databases, businesslogic, and user interfaces enabling the interactionwith underlying systems. Nevertheless, the userinterface is predominantly limited to a functionalbehavior.Traditional software engineering best-practices in-clude (a) software processes, (b) constructive ap-proaches, and (c) analytical approaches for verifi-cation and validation purposes. Software processesprovide sequences of steps for project planning,monitoring and control. Traditional software en-gineering processes, e.g. V-Model 7 and RationalUnified Process [24] focus on separated sequencesof steps for project planning with limitations inaddressing frequent changing requirements. Flex-ible and agile approaches, e.g. eXtreme program-ming and Scrum, support frequent changing re-quirements by providing a flexible structure by pro-viding small and high-efficient teams [5, 28].Nevertheless, methods are necessary to (a) con-struct valuable products and (b) to verify and val-idate deliverables. Products can be based on mod-els [7] and tests [4] within an integrated develop-ment environment. Early definition and executionof test cases based on models enables continuousintegration strategies (CI&T) [14] and can enablefrequent test runs, early availability of componentsand systems functions and foster frequent changingrequirements during project duration. Thus, CI&Tapproaches are promising approaches for VGSD be-cause of high flexible process and method support.Additionally, introducing best-practice software en-gineering approaches can help improving game de-velopment practices. Our research showed thatthere is little research regarding game developmentpractices.As a reaction on the absence of related empiri-cal work and to find out if empirical software en-gineering can be helpful in investigating the ratherunknown domain of VGSD, we consider it neces-sary to extend the existing knowledge by collectinghands-on information about the state of the prac-tice at the example of the Austrian games industry. 3 Research Issues Little research has done to capture game develop-ment practices. Thus we see the need to capture 7 http://www.v-model-xt.de (last visited 3/3/2010). best-practices in the video game development prac-tices in an systematic survey in Austria. Based onthe related work, we have set our focus on workflowintegration problems and identified a set of researchquestions with focus on the state of the practice inthe Austrian games industry. RQ1 - Studio distribution. Since there is no ex-isting information about the distribution of gamestudios available, we need basic data about studiodemographics. Depending if the Austrian games in-dustry is dominated by major or independent stu-dios [26] or a mixture of both, different further re-search steps would need to be taken. RQ2 - Process Support and Method Application  We are interested to see how far development pro-cesses are matured and to what extend agile pro-cesses and techniques like automated testing or con-tinuos integration one applied. Berner et al [6] showthat test automation is still inappropriately han-dled in traditional software industry and we expecta worse situation in the games industry. We arealso interested to see how far automation is usedwith respect to development phases and overall stu-dio size. We further assume that user-interaction-focused techniques like interaction sketching tool-ing and tailored scripting languages could be of par-ticular interest for the studios. RQ3 - Feature creep and crunch time affliction. Petrillo’s post-mortem study [27] sets crunch timeaffliction with 45% of the examined (major) studios(average team size of 22). We assume a higher per-centage of studios suffering under low to mediocrepressure of crunch time, since: (a) the majority of examined studios are seasoned and rely on estab-lished, streamlined production processes, and (b)depending on the game type we expect differentaffliction rates. 4 Study Design This section describes the basic setting of an onlinesurvey to capture the state of the practice in theAustrian games industry. Our study design is basedon the study guidelines of Kitchenham et al. [23],Wohlin et al. [30] and the reporting guidelines of Jedlitschka et al. [17]. 4.1 Survey process The applied survey process is sequential and sep-arated into (1) preparation phase, (2) executionphase and (3) data analysis and evaluation phase.A key requirement that has to be guaranteed dur-ing the whole process is participants’ anonymityand confidentiality of answers.4
Search
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks