Yesterday (Tuesday 29th of April) graduated student Frederik Mijnhardt received the Internet Impact Afstudeerprijs (master thesis price) from Google. Google Netherlands decided, for the first time this year, to grant a price to the best master thesis to encourage research on economic effects and possibilities of Internet.
First price went to Mijnhard’s master thesis “Using crowdsourcing for enterprise software localization”. He received out of the hands of Rogier Klimbie, Google’s Policy Manager Benelux, the 5000 euro award winning price.
In his thesis Mijnhardt looked at the possibility of outsourcing work via the internet (crowdsourcing) vs. the tradition outsourcing and hiring of employees. CA Technologies, an international software company, has a translation department at which professional translators from within the company as well as outside the company work. Due to the fact that the workload fluctuates heavily the costs and the time to have a product translated are high.
Mijnhardt did research on the possibility of outsourcing the translation work on the Internet. The main reasons why the industry has not accepted crowdsourcing as a feasible option yet are the quality of the work delivered by the crowd and the difficulty of measuring this quality. With this in mind Mijnhardt introduced the AVI-unit; a quality assurance method for complex crowdsourcing tasks, that he tested with 120 participants. As the picture (below) shows, the crowd not only translates texts but the results are again checked by the crowd (double crowdsourcing). This feedback gives the opportunity for the quality of the translations to be collaboratively improved.
The results of the research give us three things. Firstly the quality of the translation improved significantly with the AVI-unit in both the after-verification and after-improvement phase. Secondly the AVI-unit needs to be repeated for a second time to have actual professional quality translations. Thirdly being selective in picking only high ranked users and give them specific training and education will further increases quality.
The AVI-unit is the first mechanism in which the ‘crowd’ itself improves the work without the company interfering. The resulting quality is sufficient and can lead to savings up to 70%. Quality assurance methods seem to have the future in at least translation jobs and will be further developed to enhance quality even more. The fact that CA Technologies is working on the implementation of the AVI-unit at this very moment supports this perspective.*
Broader looking, there is a great change that crowdsourcing will become the new outsourcing which will lead to employees being people that work from digital platforms from all over the world. Of course not for all jobs this will be the case, therefore the questions arises; where else except translation jobs will quality assurance methods take over?
The newspaper article with the top three of the Internet Impact Afstudeerprijs:
*As CA Technologies is implementing the mechanism, the thesis is unfortunately not publically available (to protect patent possibilities).