Three editions of the GamifIR1 workshop have shown that a better theoretical underpinning of gami cation design is necessary to advance the state of the art. This was primarily motivated by Sebastian Deterding's keynote[1] and accepted papers at the last GamifIR[2] in 2016. This workshop aims to nd AI and data-driven opportunities for building up and developing gami cation design theory. It took place on 20 September 2017 in conjunction with the Mindtrek 2017 conference in Tampere, Finland [3]. Six full papers were selected by the programme committee from a total of eight submissions.
The call for papers solicited submissions of position papers as well as novel research papers addressing problems related to data-driven gami cation design including topics such as:
Gami ed systems that exploit data mining, machine learning and AI techniques.
Insights on game design elements built upon empirical data that can expand the catalog available to gami cation designers and practitioners. Personalized gami ed systems that exploit physiological, psychological, environmental, emotional and social data to provide tailored game elements to users with di erent characteristics.
Domain-dependent gami ed services and applications addressed to contexts like health, learning, workplace, security, crowdsourcing, and so on. Field evaluations of gami ed systems in speci c contexts of use, and new techniques to envision, design and assess gami cation design techniques. Theoretical re ections and ethical considerations on the future of gami cation enabled by the increasing availability of data.
Each submitted paper has been peer-reviewed by three members of the programme committee consisting of experts drawn from di erent communities guaranteeing a mix of industrial and academic backgrounds. Accepted papers include:
Robin Brouwer and Kieran Conboy. A Theoretical Perspective on the Inner workings of Gami cation in the Workplace. Sami Hyrynsalmi, Kai Kimppa, Jani Koskinen, Jouni Smed and Sonja Hyrynsalmi. The Shades of Grey: Datenherrschaft in Data-Driven Gami cation.
Michael Meder, Till Plumbaum and Sahin Albayrak. A Primer on Data Driven Gami cation Design.
Marigo Raftopoulos. Data-Driven Gami cation Design: An Enterprise Systems Perspective from the Front Line.
Dorina Rajanen and Mikko Rajanen. Personalized Gami cation: A Model for Play Data Pro ling. 3
After a brief welcome and introductory recap of the last three GamifIR workshops we started the presentation and discussion session. During and after the paper presentations we discussed di erent aspects of player types. For instance, we talked about how goals drive motivation and di erent user types have di erent goals. But there exist not only ten or 20 di erent types of goals, there are millions of goals and needs to be assigned to di erent types of player and user groups. It is also important to consider already existing incentives and rewards when interpreting behavior driven by gami cation because there could always exist side e ects by motivation outside the gami cation application like bonus system in workplace environments. Thus, the environment or the context is very important for analysis.
Another aspect we discussed was that an application or system creates a ordances. The gami ed system facilitates need or goal ful llment, but without the user having a congruent goal or need the system is not motivating, only through a combination of actual need and facilitated ful llment of that need can motivation arise. Robin Brouwer underlined that he disbelieves in a basic set of game design elements that always works. Instead, you always need to design something in line with the context in which the game elements are placed. In order to optimize this interplay between context and design elements you need a designer for at least the initial design!
Furthermore, we had a discussion on the necessity of pre-development insights about intended users for the gami cation design or if it is possible to assign a set of game design elements based on users behavior data maybe after a short machine learning phase. This resulted in a discussion about how to detect engagement drop-o s by speci c player or user types to create a ordances to re-engage them. Maybe di erent phases of user engagement and user experience exists and it would be very interesting to know how much exist and whether we could detect them automatically? 4
We concluded that time or timing is very important for successful gami ed systems but it is hard to detect and implement the right user journey or user phases and behavior sections: Do the right at the right time! It is not clear if we need player types as a gami cation design starting point or not. We had di erent opinions and long discussions about this. Another approach could be to just ask the user about her contexts and goals (inside the application) and later target on di erent types and moods. We agreed that we need user feedback for evaluation of di erent machine learning approaches. This could be general ratings by the users or deduced ratings on the gami ed application.
However, to be able to classify the ndings in datadriven gami cation design we need to develop objective measures of success, like the level of gameful experience (emotion, immersion, well-being, etc.), to evaluate data-driven gami cation design. DataDriven Gami cation Design should provide more insights on the di erent actual behavior patterns of different player types maybe without knowing or naming the types. Beyond that it would be interesting to compare actual behavior of di erent user types to theoretically intended behavior of self-assigned types e.g. within a player type tests. Another important dimension additionally to the player type dimension might be the behavior change on di erent time phases.
For another workshop on DDGD we would expect submission on research result about data-driven generated player types, adapting challenge level, di erent user phase detection and rst insight on adapting a gami cation design automatically. 5
We would like to thank Mindtrek [
Sebastian Deterding, University of York (UK)
Ashok Ranchhod, University of Southampton (UK) Albert Weichselbraun, University of Applied Sciences Chur (CH)
We thank all PC members and authors of submitted papers for making DDGD 2017 possible.