Empirical studies about collaboration in competitive software crowdsourcing

• Main Author: Machado, Let?cia dos Santos
• Other Authors: Prikladnicki, Rafael
• Format: Others
• Language: English
• Published: Pontif?cia Universidade Cat?lica do Rio Grande do Sul 2018
• Subjects: Software Engineering; Software Crowdsourcing; Collaboration; Barriers; Characteristics; Communication; Competition; Software Development; Crowd; Platforms; Challenges; Engenharia de Software; Colabora??o; Barreiras; Caracteristicas; Comunica??o; Competi??o; Desenvolvimento de software; Desafios; CIENCIA DA COMPUTACAO::TEORIA DA COMPUTACAO
• Online Access: http://tede2.pucrs.br/tede2/handle/tede/8256

Submitted by PPG Ci?ncia da Computa??o ([e-mail]) on 2018-08-22T17:04:33Z No. of bitstreams: 1 LETICIA_SANTOS_MACHADO_TES.pdf: 7715415 bytes, checksum: 7658ad49d27e7fdfea1a76aeee1ca834 (MD5) === Approved for entry into archive by Sheila Dias ([e-mail]) on 2018-08-24T18:53:57Z (GMT) No. of bitstreams: 1 LETICIA_SANTOS_MACHADO_TES.pdf: 7715415 bytes, checksum: 7658ad49d27e7fdfea1a76aeee1ca834 (MD5) === Made available in DSpace on 2018-08-24T19:23:50Z (GMT). No. of bitstreams: 1 LETICIA_SANTOS_MACHADO_TES.pdf: 7715415 bytes, checksum: 7658ad49d27e7fdfea1a76aeee1ca834 (MD5) Previous issue date: 2018-03-28 === Software Crowdsourcing (SW CS) ? uma estrat?gia emergente de desenvolvimento de software onde um grande n?mero de pessoas tem se engajado para contribuir em v?rias atividades de software. Tal estrat?gia (baseada na multid?o), tem sido utilizada pelas empresas que est?o buscando aumentar a velocidade de seus esfor?os em desenvolvimento de software. SW CS est? geralmente estruturado em torno de plataformas que permitem que um solicitante submeta uma tarefa e conecte-a com uma multid?o de pessoas que ir? pr?ver solu??es para a tarefa. Essas plataformas geralmente exploram uma abordagem competitiva para realiza??o da tarefa: membros da multid?o, independentemente, criam uma solu??o para a tarefa enquanto competem uns contra os outros em busca de uma premia??o financeira ao final da tarefa entregue. Uma vez que a competi??o pode reduzir a colabora??o, recentes estudos, surpreendentemente, indicam que a colabora??o existe em plataformas de SW CS. Estes estudos t?m focado em dois aspectos. O primeiro, em problemas de colabora??o entre plataforma e solicitante com rela??o a atribui??o da multid?o e as tarefas a serem desenvolvidas nos desafios de competi??o (aloca??o e submiss?o de tarefas) e, o segundo aspecto, relacionado ao impacto da colabora??o entre membros da multid?o e a qualidade das solu??es submetidas. Outros aspectos referentes a colabora??o entre os membros da multid?o ainda s?o amplamente inexplorados. Nessa tese, nosso objetivo ? identificar barreiras e caracter?sticas de colabora??o enfrentadas pelos membros da multid?o em SW CS competitivo. Para alcan?ar este objetivo, n?s conduzimos m?ltiplos estudos utilizando diferentes m?todos de pesquisa divididos em duas fases: explorat?ria e avaliatoria. Para a fase explorat?ria, os dados coletados foram obtidos a partir de: (i) partes envolvidas em projetos de SW CS (solicitante, multid?o e plataforma) atrav?s de entrevistas semi-estruturadas com profissionais e empresas, (ii) estudos selecionados atrav?s da revis?o da literatura e; (iii) estudo emp?rico sobre como desenvolvedores colaboram entre si em uma plataforma de SW CS competitivo ? TopCoder. A barreira de colabora??o mais frequente encontrada est? associada a falta de comunica??o apropriada entre as partes. Baseado nessa barreira decidimos na fase avaliat?ria conduzir uma (iv) an?lise qualitativa do principal canal de comunica??o utilizado pela multid?o: f?rums hospedados na plataforma TopCoder e, finalmente, (v) realizamos um survey destinado aos desenvolvedores que competiram na TopCoder para avaliar a influ?ncia da colabora??o no desempenho da tarefa. Os resultados obtidos nos estudos avaliat?rios sugerem que a colabora??o entre os membros da multid?o est? correlacionada com a entrega de solu??es de software vencedoras nos desafios de SW CS. === Software Crowdsourcing (SW CS) is an emergent software development strategy where a large number of people have been engaged to contribute in several software activities. Such strategy (based on the crowd), has been used for companies who are seeking to increase the speed of their software development efforts. This strategy is usually structured around platforms that allow a requester submit a task to be performed and connect with the crowd that assigned and provide a solution for the task. These platforms usually explore a competitive approach: members of the crowd independently create a solution while compete against each other by monetary rewards for task completion. While competition usually reduces collaboration, some recent studies surprisingly indicate that there is collaboration in SW CS platforms. These studies have focused on two aspects. First, collaboration concerns between platform and requester in terms of crowd?s assignment to the challenges (task allocation and submission) and second, the impact of the collaboration among crowd members in the quality of the submitted solutions. Other aspects of the collaboration among crowd members have been largely unexplored. In this thesis, our goal is to identify collaboration?s characteristics and barriers faced by crowd members in competitive software crowdsourcing. To achieve this goal, we have conducted multiple studies, using mixed research methods divided in two phases: one exploratory and one evaluatory. For the exploratory phase, we used data collected from: (i) the three involved parties in SW CS projects (requester, crowd and platform) through semi structured interviews with practitioners and companies, (ii) studies selected via literature review; and (iii) an empirical study about how developer collaborated with each other in a SW CS competitive platform ? TopCoder. The most frequent collaboration barrier was associated to lack of proper communication among the parties. Based on this barrier we decided, in the evaluatory phase, to conduct a (iv) qualitative analysis of the main communication channel used by the crowd: forums hosted on TopCoder platform and (v) a survey aimed at developers who had competed on TopCoder to assess the influence of collaboration in task performance. Our results from these evaluatory studies suggest that collaboration among crowd members is correlated with delivering winning solutions in SW CS challenges.