The Lens of Intrinsic Skill Atoms: A Method for Gameful Design

The idea that game design can inspire the design of motivating, enjoyable interactive systems has a long history in human-computer interaction. It currently experiences a renaissance as gameful design, often implemented through gamification, the use of game design elements in nongame contexts. Yet there is little research-based guidance on designing gameful systems. This article therefore reviews existing methods and identifies challenges and requirements for gameful design. It introduces a gameful design method that uses skill atoms and design lenses to identify challenges inherent in a user’s goal pursuit and restructure them to afford gameplay-characteristic motivating, enjoyable experiences. Two case studies illustrate the method. The article closes by outlining how gameful design might inform experience-driven design more generally.


Design for Motivation and Enjoyment
In recent years, human-computer interaction (HCI) has become increasingly interested in designing for enjoyment and motivation (e.g. Blythe, Overbeeke, Monk, & Wright, 2004;Jordan, 2002;Zhang, 2008). This is fueled by multiple currents: With growing affluence and stable institutions, societies are shifting toward postmaterial values of self-realization, self-expression, and experience (Inglehart, 2008). Together with the commoditization of utility and usability, this leads consumers to choose products and services based on experience (Pine & Gilmore, 2011)-and enjoyment or "fun" are among the most desired experiences. With the rise of digital networked technology, user engagement is integrated into value chains, from user-led innovation to user-generated content and word-of-mouth marketing. Hence, organizations have to motivate "produsage" on a continuing basis (Bruns, 2008). Finally, as interaction design expands into contexts like self-management, personal finance, health, or sustainability, behavior change and its motivation become frequent design goals (Fabricant, 2009). Simply put, motivation describes the psychological "processes that direct and energize behavior" (Reeve, 2009, p. 8), whereas enjoyment describes a complex of positively valued experiences and connected physiological-affective-cognitive states (Vorderer, Klimmt, & Ritterfeld, 2004). In search of foundations for designing motivating, enjoyable experiences, HCI has turned to multiple fields, especially psychology (Hassenzahl, 2010;Zhang, 2008). Yet although psychology provides ready constructs for theorizing and measuring motivation and enjoyment, it is also highly generic and analytic: It says little about potential specifics of interactive systems, nor does it easily translate into guidance how to design motivating, enjoyable interactive systems in "messy" real-world settings.

Game Design as Inspiration
Early on, researchers like Thomas Malone and John M. Carroll pointed out that video games are interactive systems purpose-built to afford motivation and enjoyment. Therefore, HCI should study games to tease out the specific "motivational techniques" (Carroll, 1982) and qualities that make games "enjoyable interfaces" (Malone, 1982), and how to design for them in nongame systems (Carroll & Thomas, 1988). A similar intuition underlies Csikszentmihalyi's (1990) study of flow, arguably one of the most influential models of enjoyment. Because activities like gaming are "designed to make optimal experience easier to achieve" (Csikszentmihalyi, 1990, p. 72), one should "explore ways in which even routine activities can be transformed into personally meaningful games that provide optimal experiences" (p. 51).
However, these early intuitions did not catch on in HCI. As the field broadened to matters of experience in the mid-2000s, scholars started studying the user experience of games (Bernhaupt, 2010), as well as methods for evaluating their usability and playability (Isbister & Schaffer, 2008). Yet none of this work was adapted to make nongame systems more engaging (see Hassenzahl, 2003, andNiebuhr &Kerkow, 2007, for two exceptions). Where HCI took inspiration from game design, it has been for intuitive forms of information display (Chao, 2001;Weisscher, 2001).
Instead, researchers began studying playfulness, especially in the context of interactive art (e.g. Arrasvuori et al., 2011;Fernaeus et al., 2012;Gaver, 2009;Morrison, Mitchell, & Brereton, 2007). Yet following play scholars like Caillois (2001; cf. Barr, 2007;Deterding, Dixon, Khaled, & Nacke, 2011), playfulness captures only one of two modes of play, namely, paidia, or free play: the unstructured, curiosity-driven exploration and recombination of behaviors, objects, and meanings. Valuable as research on playfulness is, it has not unpacked and sometimes explicitly opposed the second pole, ludus, gaming, or gamefulness: the rule-structured, challenging pursuit of goals. Yet it is this pole to which Malone, Carroll, and Csikszentmihalyi were pointing. Using the motivating, enjoyable qualities of gaming has a long tradition as serious games and serious gaming: the design and use of full-fledged games and gamingrelated practices for nonentertainment purposes (Michael & Chen, 2005). Yet again, the serious games movement has not explored the transfer of game qualities or game design into other contexts.
It is only in the past few years that practitioners and researchers have become interested in affording the motivating, enjoyable qualities of gameplay in nongame systems. This trend has become known as gameful design or gamification. Following Deterding et al. (2011, p.11), who suggest a 2 × 2 matrix of ludus/paidia and wholes/qualities to differentiate game-related design practices, and means/ends to distinguish gamification from gameful design, game design and toy design aim to create ludic and paidic wholes. Playful design aims to afford paidic qualities. Gameful design is defined by the end of affording ludic qualities or gamefulness (the experiential qualities characteristic for gameplay) in nongame contexts. In contrast, gamification describes the means of using game design elements in nongame contexts, typically for the end of gameful design. Industry publications promise "revolutionary" gains in customer and employee motivation through gamification (Paharia, 2013). According to a recent review, the majority of empirical studies shows at least partially positive results (Hamari, Koivisto, & Sarsa, 2014). However, there has been little research on methods for gameful design (Deterding, Björk, Nacke, Dixon, & Lawley, 2013). Several design frameworks have emerged in the industry but also been critiqued as the indiscriminate implementation of a small set of design patterns that fails to afford gaming-characteristic experiences (e.g. Bogost, 2011;Deterding, 2011;Nicholson, 2012;Robertson, 2010).

Article Goal and Outline
To make good on Malone's, Carroll's, and Csikszentmihalyi's early intuition and address the lack of research-based methods, this article introduces a gameful design method that has been iteratively developed and refined in 19 projects and workshops across domains. After a conceptual analysis of challenges and requirements for gameful design (Section 2), a review of existing gamification methods identifies their commonalities and shortcomings (Section 3). Section 4 presents and justifies an alternative approach: teasing out the learnable challenges inherent in a user's goal pursuit, to then iteratively prototype and improve skill atoms that restructure these to afford motivation and enjoyment. This is illustrated in two case studies (Section 5). The final Section (6) discusses wider ramifications of the method for HCI.

CHALLENGES FOR GAMEFUL DESIGN
As stated, the goal of gameful design is to afford the motivating, enjoyable experiences characteristic for gameplay in nongaming contexts. This goal can be broken down into four questions: (a) What are the enjoyable, motivating experiences characteristic for gameplay? (b) What game structures and processes afford these experiences? (c) How does game design create these structures and processes? (d) How can we integrate this into design methods for nongame systems?

Motivating and Enjoyable Qualities of Gameplay
"Fun" and "enjoyment" are widely used labels for the main desirable experiences and driving motives of gameplay, yet they are also notoriously elusive and overdetermined constructs (Taylor, 2006, pp. 67-92). Multiple models have attempted to explain them in terms of other constructs (like flow or immersion), or deconstruct them into subcomponents (see Boyle, Connolly, Hainey, & Boyle, 2012, for a review). These models variously try to capture gameplay experience (Poels, de Kort, & Ijsselsteijn, 2007), game enjoyment (Klimmt & Blake, 2012;Malone, 1981;Sweetser & Wyeth, 2005), or gaming motivation (Ryan, Rigby, & Przybylski, 2006;Sherry, Lucas, Greenberg, & Lachlan, 2006;Yee, 2006). Unfortunately, many fail to clearly separate and relate (a) enjoyment as an experience and cognitive-physical-affective state; (b) interactional, physiological, and psychological processes that give rise to (a); (c) motivational processes; (d) structural features of games that afford (b) and (c); and (e) principles how to design games with these features. This is already true for flow theory, on which several models draw (Bowman, 2008). In addition, many models use the same terms with different meanings. For instance, the nine dimensions of digital game experience by Poels et al. (2007) entail enjoyment, flow, and control, whereas flow theory suggests that flow is the experience of enjoyment and that control is an aspect of and/or precondition for flow.
One framework that evades these issues is the player experience of need satisfaction (PENS) model (Przybylski, Rigby, & Ryan, 2010;Rigby & Ryan, 2011), itself based on self-determination theory (SDT; Deci & Ryan, 2012). SDT argues that human beings are intrinsically motivated to seek out and engage in activities that satisfy three innate psychological needs: autonomy, the experience of acting with volition and willingness, in congruence with one's own goals, needs, values, and identity; competence, the experience of one's (growing) ability to achieve desired change in the world; and relatedness, a sense of intimate connection with others. PENS explains game selection as motivated by spontaneous interest grounded in previous experiences of need satisfaction and/or the acute deprivation of needs that leads individuals to want to replenish them. Ongoing gameplay is motivated by the actual experience of need satisfactionthis is what is meant with the activity being autotelic (its own goal) or intrinsically motivated. Game enjoyment and immersion are explained as experiential signatures of need satisfaction: What players label "enjoyment" or "fun" is the experience of need satisfaction, which leads to spontaneous attentive absorption and emotional investment in gameplay, constituting immersion (Ryan et al., 2006). Put differently, "fun" is an everyday language label for the experience of enjoyment, arising chiefly from need satisfaction.
Several reasons speak for using PENS to model gaming enjoyment and motivation. SDT is a well-established theory of human motivation with empirical support across contexts (Deci & Ryan, 2012). Need satisfaction accounts not only for a significant part of gaming enjoyment and motivation (e.g., Przybylski et al., 2010;Reinecke et al., 2012) but also for positive experiences in life (Sheldon, Elliot, Kim, & Kasser, 2001) and with interactive products (Hassenzahl, Diefenbach, & Göritz, 2010). Hence, SDT finds increasing usage in research on hedonic and motivational experiences (Hassenzahl, 2010;Zhang, 2008). Finally, PENS can account for constructs of competing models such as flow theory or Yee's (2006) motivations for online play (Przybylski et al., 2010).
More recent models complement PENS with mood management theory (users selecting media to optimize arousal and affect) and appreciation (the satisfaction of eudaimonic needs for meaningfulness and insight), and have found first empirical support for their superior explanatory power (Oliver & Raney, 2011;Reinecke et al., 2012). Still, PENS remains their bedrock; provides a well-operationalized, robust nomological network that accounts for most gaming-characteristic motivations; avoids the conceptual conflations just outlined; and offers an explanatory model for the linkage of game features, motivational processes, and enjoyment (Rigby & Ryan, 2011).

Structures and Processes Underlying Gameplay Motivation and Enjoyment
Formally, most common definitions of games agree that a game consists of one or more interconnected challenges a player is trying to overcome, which emerge from the player taking actions in pursuit of goals, and rules and objects/opponents that make attaining those goals challenging (Juul, 2005, pp. 29-43). Challenge here means a specific kind of difficulty or resistance requiring nontrivial effort and skill to overcome-such as hand-eye coordination in shooter games. Most game design literature agrees that a game's challenge is at the heart of its gameplay experience (e.g. Fullerton, 2008, pp. 77-78, 86-91, Schell, 2009. Challenges give rise to motivating, enjoyable experiences through multiple routes: First, overcoming nontrivial challenges creates the pleasurable experience of competence (Rigby & Ryan, 2011). Second, by choosing to tackle a challenge for the sake of enjoyment (provided no other consequences are attached) and by choosing which challenges to tackle with which strategies (provided multiple are offered), players perceive themselves as acting with high autonomy (Deterding, 2014a;Rigby & Ryan, 2011). Third, overcoming challenges collaboratively and/or in service of significant (virtual) others satisfies relatedness needs (Rigby & Ryan, 2011). Fourth, the outcome of a nontrivial challenge is by definition uncertain, stoking curiosity, interest, and arousal (Malone, 1981;Silvia, 2012). Fifth, following mood management theory, nontrivial challenges bind attention and cognition, leaving few resources to ruminate on unpleasant thoughts (Reinecke et al., 2012). This is what Csikszentmihalyi (1990) referred to as the "deep but effortless involvement that removes from awareness the worries and frustrations of everyday life" (p. 49). Several heuristics for designing challenges and their component goals, rules, and so on, have been developed (Isbister & Schaffer, 2008). For instance, challenges should be balanced relative to the player's perceived current ability such that they appear neither too hard (inducing anxiety) nor so easy that they generate no uncertainty before nor competence upon overcoming them (Csikszentmihalyi, 1990, pp. 49-53, 74-77).

The Systemic, Emergent Quality of Gameplay Enjoyment and Motivation
Notably, challenge, motivation, and enjoyment are systemic, emergent properties. A historical look into serious game design can help clarify this point: The first generation of "edutainment" games assumed that instrumental activities like learning entail inherently unpleasant properties and games inherently enjoyable properties (Egenfeldt-Nielsen, 2007). By analogy, sugar has inherent chemical features that, when imbibed by a living being with the respective dispositions (taste receptors, etc.), will infallibly result in the experience of sweetness. The resulting design paradigm has been called "chocolate-covered broccoli" (Bruckman, 1999): Presumed-inherently unenjoyable learning is made palatable by adding presumed-inherently enjoyable gameplay, either through inserting learning tasks into an entertainment game or by "spicing up" learning tasks with game feedback like the sound and sight of a popping balloon.
Due to lacking success, this inherent-additive paradigm has been largely abandoned. The current generation of serious games subscribes to emergent theories of enjoyment, and as a result, systemic design paradigms (Egenfeldt-Nielsen, 2007;Squire, 2006). By analogy, the taste of a piece of pastry depends not on its sugar content but on the mixture and preparation of all ingredients, and how the resulting whole suits the sensitivities of the eater, her sociocultural background, current satiety, and so on. This aligns with current views of user experience as subjective, holistic, emergent, situated, and dynamic (Hassenzahl, 2010, pp. 6-31). In this paradigm, all activities can be motivating and enjoyable. It is just that good games are purpose-built to optimally afford motivation and enjoyment, whereas, for example, most contemporary formal education is not (Gee, 2004).

Game Design for Motivation and Enjoyment
In game design, the emergent, systemic quality of game enjoyment has been formalized in the mechanics, dynamics, aesthetics (MDA) model (Hunicke, LeBlanc, & Zubek, 2004): A game's mechanics-the rules specifying possible actions-form a system that players interact with, giving rise to interactional dynamics, which in turn give rise to experiential aesthetics. For instance, the mechanics of Monopoly state that players roll two dice and move their game piece the rolled number of fields. If they land on a field owned by another player, they have to pay that player, which may require selling their own fields to generate the necessary cash. This leads to the dynamics of a slowly widening poverty gap, which engenders the aesthetic of frustration for the losers, as losing becomes foreseeable early on but is dragged out. Challenge is a fundamental player-system dynamic: As much as a game might explicitly suggest goals, unless players intentionally pursue them, neither gameplay nor challenge come about (Björk & Holopainen, 2005, p. 317). Balancing of challenge similarly depends on the dynamic relation of game and player states.
The systemic, emergent quality of game experience has resulted in "iterative design" (Zimmerman, 2003) or "playcentric design" (Fullerton, 2008) as the dominant game design method, characterized by iterative experiential prototyping: The designer defines an experience she wishes to evoke and then ideates first mechanics that might generate it. Because both player-system interactions and resultant player experiences are highly emergent, they cannot be reliably predicted. Therefore, the designer builds and tests rough functional prototypes to observe what dynamics and aesthetics actually emerge, evaluates how and why these diverge from the intended experience, ideates promising design changes, and revises and tests the prototype again. This cycle is repeated until the desired experience emerges.

Bridging Game and Interaction Design
To summarize, central components of gaming motivation are basic psychological needs for autonomy, competence, and relatedness, and a central component of gaming enjoyment is the experience of having these needs satisfied. As noted earlier, gameful design is specifically concerned with the ludic, gameful pole of gameplay. The chief enjoyment and motivation of such ludic gaming is the sense of competence (effectance, control, mastery) that arises from overcoming challenges (Klimmt & Blake, 2012)-a view shared by game designers (e.g. Koster, 2004). Malone (1981) and Csikszentmihalyi (1990) similarly have articulated challenge as a central dynamic for optimal experiences. In fact, Hunicke et al. (2004) suggested that challenges are core to any gaming experience. Evoking different experiences and motives means designing different challenges "tuned" to them.
Gameplay experiences are doubly emergent: They are aesthetics that emerge from the dynamics of a player interacting with a game, which in turn is a systemic whole of its mechanics. Specifically, they arise from players tackling and overcoming well-balanced challenges constituted by a system of goals, rules, actions, objects/opponents, and feedback. Game design therefore targets an experience and then engages in an iterative cycle of ideation, prototyping, testing, and analysis until the delta between intended and actual experience is closed.
With this thumbnail sketch of game-characteristic experiences, structures, processes, and design methods regarding motivation and enjoyment at hand, how does one translate them into interaction design-how to move from game to gameful design? Conceptually, gameful design is defined as affording gameful experiences in any nongame context . Practically, it is almost always a form of design with intent (Lockton, Harrison, & Stanton, 2010) that also has to support functional needs-like a printer that both allows printing and motivates users to print less. In short, gameful design attempts to design systems that serve specific functions and uses and facilitate both through motivating, enjoyable experiences.
Closest to gameful design in this regard are serious games: games designed and/or used for non-entertainment purposes. Analyses of the challenges of serious game design thus provide a useful entry point into the challenges of gameful design.

Tension of Experiential and Instrumental Goal
First, both face an inherent goal tension: delivering on an experiential goal (enjoyment, motivation) to serve another instrumental goal. The literature typically articulates this as a trade-off, balancing, or constraint: Entertainment games can focus exclusively on enjoyment. In serious games, enjoyment must be balanced with, is constrained by, and ideally converges with concerns for instrumental outcomes and requirements such as assessment or realism (Harteveld, 2011;Isbister, Flanagan, & Hash, 2010;Khaled & Ingram, 2012, Michael & Chen, 2005Winn & Heeter, 2006). 302

Synthesizing User Research and Requirements for Game Design Ideation
Second, to bring instrumental goals and requirements into the design process, both serious game and gameful design have to bridge interaction and game design. Interaction design is primarily concerned with supporting people's work and everyday needs (Rogers, Sharp, & Preece, 2012, p. 9). It has developed robust methods for understanding the goals, needs, and contexts of system owners (stakeholders deciding over a system) and end users (stakeholders using the system; Rogers et al., 2012, p. 171) and assessing how well systems support these. Although there are increasing efforts around "experience-driven design" (Desmet & Schifferstein, 2011), these have been mostly concerned with formative research identifying and modeling what experience to design for, not how to get from the identified target experience to a design that affords it.
Game design in contrast is concerned with just that: designing interactive systems around experiential goals. Iterative design provides a robust, formalized method for doing so. (Where HCI has worked on experience design methods, suggestions again revolve around iterative prototyping; Buchenau & Suri, 2000). However, as Khaled and Ingram (2012, p. 76) observed, whereas game design provides little guidance on "early stage design research with end users (players), or how to fold these insights back into the design process," interaction design "unsurprisingly sheds little light on how to move from user insights to game mechanics and game concepts." In the design work informing the present article, this lacking bridge was also by far the most frequent issue. Gameful design needs a design synthesis (Kolko, 2010)-shared artifacts, practices, and understandings fertile for generating ideas, and able to guide the following iterative refinement.

Mobilizing the Professional Vision of Game Design
This bridge from research to ideation is part of a larger required epistemic and practical bridging: Design teams need a shared language, a joint way of seeing to articulate, evaluate, and coordinate concerns, observations, and actions. Such a "professional vision" (Goodwin, 1994) entails (a) coding observed phenomena into objects of knowledge, (b) highlighting certain objects as relevant and certain problems as problematic, and (c) material representations that ground perception, discourse, and interaction. Yet "ways of seeing" and their epistemic objects do not travel easily: They are indexically embedded in and constituted by the total web of embodied experiences, practices, tools, and discourse of their profession. Many game design patterns (like "zone of influence") not only draw on further game design patterns (Björk & Holopainen, 2005) but they also only make sense within the context of a game. To straightforwardly speak of the "zone of influence" of a shopping application is simply meaningless. If the main tactic of gameful design is to apply game design elements (i.e., gamification), then these elements have to somehow be mobilized, translated such that they can travel to interaction design.
In serious games, this mobilization typically happens through multidisciplinary teams that include game designers; joint work over time establishes a shared way of seeing (Winn & Heeter, 2006). Yet this is by definition resource intensive, is friction laden, and has to happen with each team anew. Ideally, interaction designers should be able to create gameful systems on their own. This asks for an accessible formalization of the epistemic objects of game design that can be easily integrated into standard interaction design practice.

The Challenge of Challenge
Taken at face value, interaction design and game design are anathema. A core tenet of interaction design is to make systems usable, to remove any unnecessary challenge. Game design, as seen, is about the deliberate creation of challenges. After all, "playing a game is the voluntary attempt to overcome unnecessary obstacles" (Suits, 2005, p. 55). As Cook (2008) put it, if a usability engineer were to redesign Super Mario Bros. with the sole goal of usability, the result would be a single button, labeled "Rescue Princess." By easing the avowed instrumental goal into a single button click, the actual experiential goal of enjoyment is destroyed. This issue was already pointed out by Malone's (1981)

distinction between toys and tools:
Toys can be defined as systems used for their own sake with no external goal (e.g., computer games, puzzles, etc.). Tools can be defined as systems used as a means to achieve an external goal (e.g., text editors, programming languages, etc.). The requirements for good toys and tools with respect to challenge are mostly opposite. . . . In a sense, a good game is intentionally made difficult to play to increase its challenge, but a tool should be made as easy as possible to use. (pp. 359-360) Psychologically, things we call "games" and "toys" we typically engage with autotelically (for their own sake), whereas things we call "tools" we typically engage with exotelically, for the sake of some instrumental purpose (Csikszentmihalyi, 1990). Autotelic, autonomous engagement is a defining part of people's conception of "playing" (Deterding, 2014a). However, as Csikszentmihalyi's (1990) workplace studies show, "working" with tools is often autotelic as well. Conversely, exotelic, inautonomous engagement with games and gamified applications leads people to label it "work" and impedes enjoyment and performance (Deterding, 2014a;Mollick & Rothbard, 2013). It ultimately depends on the users' situational framing-afforded but not determined by objects and context-whether they experience an activity as "work" or "play" (Deterding, 2014a). Reframing, therefore, is itself an important part of gameful design (Deterding, 2014b).
Of course, game design is not about indiscriminately adding challenge: Good games combine difficult gameplay challenges (evoking competence) with easy-to-use interfaces (as whatever gives access to gameplay; Juul & Norton, 2009). Yet the fact remains that the value of challenge in interaction and game design is diametrically opposed.
Serious games resolve this issue through intrinsic integration and transfer. Intrinsic integration means that learning material is embodied in the game world as the most "fun-to-play" core gameplay (Habgood & Ainsworth, 2011). In such "endogenous games" (Squire, 2006), overcoming the core challenge of the game requires and supports acquiring the concepts and skills to be learned. This makes pedagogical sense because gameplay embodies good, experientially grounded, explorative, scaffolded learning by doing (Gee, 2004;Squire, 2006). Evidence supports that intrinsically integrated serious games are motivationally and educationally more effective (Habgood & Ainsworth, 2011). Serious games can work with intrinsic integration because they achieve their instrumental goal through transfer (Mestre, 2005): Mastering game challenges affects beliefs, concepts, attitudes and skills, which in turn affect ultimate activity beyond gameplay, for example, choosing healthier food. Learning (and game challenges) are separate from and thus don't add friction to the ultimate target activity.
In sharp contrast, gameful systems have to both directly support end user activity (be easy to use) and facilitate it through enjoyment and motivation. Hence, "tacking on ad hoc complications will not evoke fun" (Carroll & Thomas, 1988, p. 21). Few users would enjoy having a ticket vending machine gamified by adding a puzzle challenge to the already puzzling purchase process. The notion of intrinsic integration has to take a different bent here. Rather than creating new challenges that evoke enjoyment (entertainment games) or embody learning goals (serious games), gameful design has to identify the challenges already inherent in the user's pursuit of her needs and restructure them in a motivating manner (cf. Csikszentmihalyi, 1990, p. 51). That is, what are the already-present challenges of her activity that in the given constraints can neither be hidden or automated away nor made any easier? To engender competence, these inherent challenges are ideally skill based: something a user can learn and improve in (cf. Cook, 2008). Creating an online classified advertisement, for instance, involves the inherent challenge of writing persuasive copy, which is a learnable skill. Figuring out where to enter the written copy in an online form is neither inherent nor skill based. Having to solve a math equation to unlock the form field would be skill based-one can learn mathematics-but not inherent to the activity.
This means that gameful design has far less creative control over challenge. Entertainment game designers can create whatever challenge they consider most conducive to enjoyment. Gameful designers usually can neither fully control nor predict the goals (and entailed challenges) of their users. In Super Mario Bros., designers create goals and environments. With a word processor, overall context and user expectations chart a corridor of likely goals and activities, but the actual goals and objects a given user works on at a given time are widely divergent (Pagulayan, Keeker, Fuller, Wixon, & Romero, 2008, p. 743). Hence, gameful design will often have to be underspecified, open to customization, or support dynamic personalization.

Summary
This section established the motivating and enjoyable qualities of gameplay, the structures affording them, how game design creates them, and the challenges of translating this into interaction design. We can summarize the main findings in six requirements for gameful design: 1. Design for basic need satisfaction: Enjoyable, motivating experiences characteristic for gameplay are competence and, to a lesser extent, autonomy, relatedness, curiosity, arousal, and attentive focus. Gameful design should target these experiences. 2. Design around inherent skill-based challenges: Games afford these experiences through nontrivial challenges arising from a system of goals, objects, rules, actions, and feedback. Because gameful design typically involves a system that supports existing functional activity, it ought to contour the skill-based challenges already inherent in said target activity, rather than adding new ones. 3. Design for systemic emergence: Because enjoyment, motivation, and challenge are emergent properties of a systemic whole of user-system interaction, gameful design therefore has to design and test such whole systems. 4. Formative research: Because gameful design needs to support existing user goals and needs, it requires formative research that identifies their goals and needs as well as the challenges inherent to their pursuit. 5. Design synthesis: Formative research has to be synthesized into a form that informs the ideation and evaluation of gameful design ideas. 6. Epistemic mobilization: The epistemic objects of game design have to be translated into a form that is intelligible and easily adoptable for interaction designers.

Academic Methods
With these requirements in hand, we can review existing work on gameful design. Academic research on gamification and gameful design methods is still in its infancy (Deterding et al., 2013). Nicholson's (2012) "user-centered theoretical framework for meaningful gamification" articulates useful design values (user centricity, transparency, personalization) but no actual method. The social gamification framework by Simões, Redondo, and Vilas (2013) similarly remains a declaration of objectives and future outcomes. Gears and Braun (2013) merged object-oriented systems analysis and design with SDT and the theory of 16 basic desires (Reiss, 2004) into a business gamification process. A "role-motivation-interaction framework" articulates the interactions and possible underlying motivations for each user role, which then shall guide the choice of fitting game design patterns. Yet not only are SDT and basic desires incompatible theories (Reiss, 2004), the framework doesn't provide guidance how to identify fitting patterns.
The four-step gamification process by Aparicio, Vela, Sánchez, and Montes (2012) tries to identify the main user task and a "transversal objective." This appeals to "games with a purpose" (Von Ahn & Dabbish, 2008), where an overt gaming goal (guess what word another user might think of when seeing this image) covertly solves human information tasks (tag images). Designers should select game mechanics that match both main and transversal objective and support autonomy, competence, and relatedness. However, many gameful applications do not feature diverging main and "transversal" objectives (e.g., fitness apps); Aparicio and colleagues also don't explain how they matched mechanics with basic needs, or how to choose the right mechanics. Popa (2013) suggested using a traditional user-centered design (UCD) process involving research-based personas (Pruitt & Adlin, 2006) to create "gamey personas." These include information about a user's personality type (using the BrainHex player typology), emotions the user enjoys in playing games, and the user's experiential goals, and they serve as brainstorming triggers. Apart from basic issues with player typologies (see next), there is again little guidance how to move from experiential goals to a design solution.

Industry Methods
Arguably the earliest documented industry method is "smart gamification" by Kim (2010). She suggested to first identify the preferred "engagement style" of the target audience based on four "social engagement verbs" and the system owner's goals. Then designers create a "player persona" entailing a "player journey" with goals and activities on three stages: newcomers, regulars, and enthusiasts. Preferred social engagement verbs are mapped to fitting stages, as are game design patterns that display user progress. These should be connected to rewards like virtual items or feature unlocks. Next, customization options for the main "social objects" (e.g., user profiles) are chosen. Finally, for each supported social verb, a "social engagement loop" is devised: a feedback loop of a call to action, social action, system feedback, and underlying motivation.
The majority of later methods follow in Kim's tracks with slight amendments. Zichermann and Cunningham (2011), Werbach and Hunter (2012), Kapp (2012), Kumar and Herger (2013), Paharia (2013), and Burke (2014) all suggest a design process of more or less the following form: 1. Identify system owner goals. 2. Identify trackable behaviors of end users that support these goals; quantify their relative contribution in a metric. 3. Profile and segment end users using player typologies (usually "Bartle Types"). 4. Select and specify game design patterns: a. Translate the quantified system owner value of end user behaviors into point values displayed back to users. b. Articulate an ordered sequence of explicit goals for end users, consisting of sets of behaviors or point thresholds (quests, challenges, levels). c. Define feedback to display upon single user actions ("engagement loop") as well as reaching point thresholds or goals (achievements, badges, leader boards), including potential rewards (virtual items, customization options). d. Choose additional game design patterns. 5. Playtest. 6. Build and deploy. 7. Use analytics of user behaviors to monitor system performance and guide the improvement and release of new content and features.
An interesting alternative, Dignan's (2011) "game frame" suggests to identify the core user activity to be facilitated, create a player profile entailing its main hindrances and motivational drivers, choose long-term goals for the user, identify skill(s) required, and determine a "resistance" that makes achieving the objectives difficult, as well as resources, objects, and spaces with which the user engages. Finally, designers need to define a skill cycle, select rewards and punishments, and playtest and refine. Two further edge cases worth mentioning are Ferrara (2012) and Kapp (2012), who both overtly present models for gamification yet in the end revert to educational serious game design.

Characteristics and Issues
One may safely assume a dark figure of unpublished methods in use in the industry, but the high similarity of the publicly available methods and their reach through widespread deployments (Paharia, 2013) and training programs (Werbach & Hunter, 2012;Zichermann & Cunningham, 2011) suggest that they can be taken as representative. Beyond their basic procedural commonality, several further commonalities are worth teasing out.

Little Formative Research
In the reviewed methods, formative research is neither deeply integrated nor well specified. Although all profile end users, typically using personas, the majority recommends imaginary rather than data-driven personas-a common problematic practice (Chang, Lim, & Stolterman, 2008). Data collection methods are glossed over at best. Some explicitly recommend basing personas on educated guesses (e.g., Burke, 2014, p. 102;Werbach & Hunter, 2012, pp. 91-93;Zichermann & Cunningham, 2011, p. 21). Partial exceptions are Paharia (2013, pp. 192, 208-213) and Kumar and Herger (2013). They recommend using interviews and site visits but provide no guidance on how to conduct these to identify motivations or player types.

Reliance on Player Typologies
Almost all reviewed methods portray classifying end users by "Bartle Type" as a useful activity (Kapp, 2012, pp. 132-137;Werbach & Hunter, 2012, pp. 92-93;Zichermann & Cunningham, 2011, pp. 21-24): "Understanding the type of player will help you choose the game mechanics that will be most appealing to your target audience." (Kumar & Herger, 2013, n.p.) Bartle (1996) suggested that most players of Multi-User Dungeons (MUDs) stably prefer one type of enjoyment in MUD play: Achievers enjoy achievement within the game context, explorers exploration of the game world, socializers socializing with others, and killers imposing themselves upon others. Kim's (2012) "social engagement verbs" are also based on Bartle but (she claims) better capture people's main "social engagement style," namely, competitive, explorative, cooperative, or expressive.
A general problem of typologies is that they are far less context-specific than data-driven personas. Building personas from typologies, as suggested by Kumar and Herger (2013) or Popa (2013), loses the very advantage personas bring in contrast to typologies (Pruitt & Adlin, 2006, p. 99). Game designers therefore often warn of relying on player typologies (Schell, 2009, p. 111).
Second, as Bartle (2011) himself argued, the gamification literature frequently misconstrues player types: It ignores his later refinement of eight player types and falsely assumes that player types suggest what kind of activity players should receive points and badges for, whereas only achievers care about the latter to begin with.
Third, Bartle's (1996) typology, based on the unsystematic review of one bulletin board regarding one MUD, has been superseded by a plethora of alternative typologies, based on empirical studies with representative samples and rigorous methods (Hamari & Tuunanen, 2014). Research into game genres beyond MUDs regularly highlights motivations not accounted for by Bartle (e.g. Sherry et al., 2006).
Fourth, the applicability of player types beyond MUDs is an unproven, fraught hypothesis. To assume that a classification of MUD players captures motivations in any context beyond MUDs is as much a non sequitur as assuming that a typology of milk drinkers captures people's reasons for watching movies. This also holds for Kim's (2012) social engagement verbs: covering "modern social gaming and social media," they won't account for, for example, solitary activities.

Appeals to Motivational Psychology
The reviewed methods frame gamification "as a means to design systems that motivate users to do things" (Werbach & Hunter, 2012, p. 30). SDT is by far the dominant theory appealed to, with the notable exception of Zichermann and Cunningham (2011, pp. 15-34). However, SDT is chiefly received through popularized representations (most notably Pink, 2009), leading to partially erroneous representations. It is also variously mixed with other models into new, idiosyncratic, untested models of motivation (Burke, 2014;Dignan, 2011;Gears & Braun, 2013;Kumar & Herger, 2013;Paharia, 2013).
More important, the reviewed methods portray gamification as the choice of preexisting elements: "Putting all these [game] elements together is the central task of gamification design, and having knowledge of these game elements will make your gamification project compelling" (Werbach & Hunter, 2012, p. 81). In essence, they recommend a pattern-based approach to design (Seffah & Taleb, 2012) and subscribe to an inherent-additive paradigm of motivation and enjoyment: Activities are made enjoyable by adding inherently enjoyable game elements. "Once you have nailed that [core value proposition] down, the gamification mechanics can layer on top of it and drive engagement, activity, and loyalty around it" (Paharia, 2013, p. 69).
Thus, although the reviewed methods commonly label the catalogued elements "mechanics" and mention the MDA approach, their portrayal ignores its systemicemergent thrust. For instance, Werbach and Hunter (2012, p. 80) mistake mechanics and dynamics as "orders of abstraction." Kim (2010, p. 20) misidentifies dynamics as "time-based patterns" such as "reward schedules" rather than emergent properties. The majority of elements listed by the reviewed methods are in fact interface design patterns for displaying goals, reputation, and progress feedback .
An important facet of this neglect of systemic emergence is that most methods directly classify design patterns by motivation, assuming a deterministic one-toone pattern-effect relation (Aparicio et al., 2012;Kim, 2011;Kumar & Herger, 2013;Paharia, 2013;Zichermann & Cunningham, 2011). However, the motivational "functional significance" (Ryan, 1982) of an event results from situated, active interpretation-for example, whether feedback is perceived as informational (supporting competence) or controlling (thwarting autonomy). As Antin and Churchill (2011) outlined, a simple design element like a badge can be functionalized for at least five different motivational processes (goal-setting, group belonging, etc.).
Also, by definition, design patterns are "invariant solution[s] to address a recurrent design problem" (Seffah & Taleb, 2012, p. 95). They allow to quickly create best practice solutions in a well-explored design space. This efficiency is an important selling point of gamification (Paharia, 2013). Yet whatever efficiency and reliability gains there may be, they are necessarily bought with foregoing the exploration of novel, potentially more effective solutions. That is, the reflexively denied "indiscriminateness" of gamification is inherent to the pattern-based approach of the reviewed methods.

Lacking Guidance in Game Design Pattern Choice
The reviewed methods provide little guidance in choosing and customizing patterns for a given context-a common problem of pattern-based design (Seffah & Taleb, 2012). In other words, the presumably central design step in gamification is glossed over. Kim (2010) suggests simply picking mechanics from a list. Zichermann and Cunningham (2011, pp. 36, 142-151) first state that gamification always involves points, levels, achievements, and rewards, to then literally recommend throwing two dice on a table of 12 game mechanics. If the result doesn't fit the problem, one should "leverage that creative dissonance" (Zichermann & Cunningham, 2011, p. 79). Werbach and Hunter (2012, p. 82) emphasize the importance of picking the right mechanics and claim that engagement loops "should give you the skeleton of your system" (p. 99), but again provide no guidance how to devise engagement loops or how to move from loops to elements.
In this, the reviewed methods also fail to synthesize formative research into a useful ideation and evaluation basis. In some instances (Werbach & Hunter, 2012;Zichermann & Cunningham, 2011), player personas and player type classifications remain token exercises the results of which are never manifestly used later. Kumar and Herger (2013) suggest choosing mechanics that fit player types but without any specification what mechanic fits what type. The exceptions here are Paharia and Kim. Paharia (2013, pp. 218-219) recommends walking through scenarios of personas interacting with a system, at each step asking what pattern from a master list of 10 patterns might apply (then again, the remainder of his design process seems to suggest always using some form of goals, points, and achievements; Paharia, 2013, pp. 213-228). Kim (2010) used the four social engagement verbs to derive explicit recommendations which interaction functionality to offer at which stage of a user's engagement with a system. But as with all other reviewed methods, this functionality-verb (or pattern-motivation) link is never really justified (let alone empirically tested).

No Iterative Prototyping
Reviewed methods vary in whether they involve an explicit evaluation phase, and whether this phase is separate from postdeployment monitoring. But in all, iterative prototyping and evaluation of alternative design directions are noticeably absent. Neither Kim (2010) nor Paharia (2013) even mention iterative prototyping. Kumar and Herger (2013), Zichermann and Cunningham (2011, p. 73), and Burke (2014, p. 124) chiefly recommend postlaunch monitoring and development of a gamified system. Werbach and Hunter (2012) spend exactly one sentence on the matter: "The best way to tell if your system is fun is to build it and test it and refine it through a rigorous design process" (p. 99). The same holds for Dignan (2011).

Data-Driven Design
In tune with current lean startup and big data trends, the reviewed methods emphasize the quantitative tracking and monitoring of end user engagement. Granular data on individual user behaviors are in fact a major value proposition of gamification (Paharia, 2013). There is one crucial difference, though: Lean startups (Ries, 2011) extensively use data already in the iterative design of prototypes or "minimum viable products." The reviewed methods suggest tracking and monitoring only after deployment.

Summary
Gamification, as manifested in the reviewed methods, is a pattern-based, datadriven design approach grounded in an inherent-additive paradigm of motivation and enjoyment. By foregoing formative user research for problematic player typologies, and iterative design exploration for patterns and postlaunch tracking, it appears to be optimized mainly for speed and cost. It also fails the identified requirements for gameful design: 1. Design for basic need satisfaction: Although most reviewed methods appeal to basic needs, they typically blend popularized versions of SDT with other models into untested "homegrown" motivation models. 2. Design around inherent skill-based challenges: With the partial exception of Dignan (2011), none of the reviewed methods focuses on skill-based challenges inherent in the user's goal pursuit.

Design for systemic emergence:
The pattern-based, inherent-additive paradigm of the reviewed methods is anathema to systemic emergence. 4. Formative research: With the partial exception of Popa (2013), formative research is either not sufficiently specified or replaced with tokenist categorizations of users into "Bartle Types." 5. Design synthesis: No reviewed method provides sufficient research-grounded instruction for the presumed-essential step of selecting fitting game patterns. Kim's (2010) social engagement verbs come closest but, like all reviewed methods, lack validation. 6. Epistemic mobilization: In grafting on game design concepts, translation errors abound, ranging from minor ones (misrepresenting "mechanics") to the fundamental issue that central game design processes like iterative prototyping are absent.
In sum, none of the reviewed gamification methods is set up to create systems affording the motivating, enjoyable experiences characteristic for gameplay or to mobilize the professional vision of game design.

A METHOD FOR GAMEFUL DESIGN
Analyzing the role of challenge (Section 2.5), we concluded that a promising strategy for gameful design is to restructure challenges inherent in the user's goal pursuit into a systemic whole that optimally affords enjoyable, motivating experiences. The following section explicates this strategy into a design method, the lens of intrinsic skill atoms. As the name suggests, the method draws on three concepts: design lenses, skill atoms, and intrinsic integration. The latter we already encountered: Gameful design should focus on challenges inherent in the user's goal pursuit. The next subsections explain the two remaining concepts and the resulting design process.

Design Lenses
Design lenses were initially developed by Schell (2009) for game design, but Scott (2010) quickly adapted them to transfer game design concepts into interaction design. As Schell (2009) put it, a lens is "a way of viewing your design" (p. 732-733). Practically, it combines a memorable name, a concise statement of a design principle, and a set of focusing questions that allow the designer to take on the "mental perspective" (Scott, 2010) of the lens. By binding together a principle with focusing questions, a lens becomes self-contained. This enables lenses to "travel" between designs and domains. By using questions rather than guidelines, lenses instruct designers how to see a design space. They provide terms to code observed phenomena into epistemic objects and highlight certain objects as relevant and certain problems as problematic. Thus, lenses instantiate part of the "professional vision" (Goodwin, 1994) of their discipline. Take the following example (Schell, 2009, p. 181):

Lens #32: The Lens of Meaningful Choices
When we make meaningful choices, it lets us feel like the things we do matter. To use this lens, ask yourself these questions: • What choices am I asking the player to make? • Are they meaningful? How? • Am I giving the player the right number of choices? Would more make them feel more powerful? Would less make the game clearer? • Are there any dominant strategies in my game?
This lens instructs designers to code a design space in terms of meaningless versus meaningful choice. It then highlights as problematic an imbalance of offered choice or the presence of dominant strategies. The lens doesn't prescribe specific solutions how to offer meaningful choice (like a pattern does), nor is it limited to identifying issues (like a heuristic would). It provides a design guideline that is both generative and evaluative, and devised to be self-contained. This makes design lenses an ideal starting point for gameful design.
Existing game design lenses (Schell, 2009) unsurprisingly speak to games and game designers. Take "dominant strategy" in the previous example: It is a game theoretic term describing a choice where one option is reliably better than the other, effectively rendering the choice meaningless (Schell, 2009, p. 180). Using this term presupposes that the designer is embedded in the vernacular of game design. Also, existing game design lenses give no clear starting points for analysis, and many are not easily applicable beyond games. To address these issues, we rearticulated game design lenses into motivational design lenses that are fully self-contained and organized by motives (Deterding, 2015; see Figure 1; Supplementary Material 1). Second, we suggest to start gameful design by viewing the given design space through the lens of skill atoms-which focuses design on inherent challenge and provides a base anchor and vocabulary for all lenses.

Juicy Feedback
Varied, unexpectedly excessive sensual positive feedback on small user actions and achievements can instil a sense of competence even without a big challenge to overcome. Its unexpected variety also stokes curiosity.
What are the small steps of your action (e.g. approaching, touching, lifting, moving, releasing in "drag and drop")? When does a user achieve a goal? How might you exaggerate the auditory, visual, and tactile feedback at these steps and achievements, going unexpectedly "over the top"? Is there a material or creature whose sensual properties might inspire your feedback (like anemones or bubble wrap or dough or ...)? How might you add some variation to that feedback to keep it interesting over time?

Skill Atoms
The concept of skill atoms (Cook, 2007) stems from an ongoing effort in game design to develop a formal "game grammar" (see Dormans, 2012, pp. 49-65, for a review). As an "atom," it consists of smaller reoccurring particles yet cannot be broken into these without losing its systemic "gaminess." A skill atom describes a feedback loop between user and system that is organized around a central challenge or skill: A user takes an action, which forms an input into the system's rule engine that determines state changes of its tokens, which get put out as feedback to the user, which she integrates into her mental model of the system. Through repeated interaction-multiple run-throughs of the atom-the user masters its skill: training hand-eye coordination, understanding the rules, and so on (Cook, 2007).
Because Cook's original model articulates process phases more than structural features designers can work with, it was slightly amended. A skill atom as understood here (Figure 2) consists of goals, actions and objects, rules, feedback, emergent challenge, and motivation. Note that in digital systems, rules are rarely explicitly accessible: They are implied in what actions, objects, and feedback are offered to the user.

Skill Atom Components
• Goals: System states the user attempts to achieve. Goals are typically explicitly suggested by the system but must be actively pursued by the user to be goals. • Actions: What the user can do to approach her goals. • Objects: Entities the user acts upon; their configuration embodies the system state.
• Rules: Specifications what actions the user can take and how they affect the system state. These may be algorithms, humanly enacted rules, physical laws, or a combination thereof. • Feedback: Sensory information that informs the user of system state changes resulting from her actions or autonomous system processes; entails immediate feedback on each action and progress feedback on the user's accumulated progress. • Challenge: The perceived challenge of achieving the user's current goal, posed by the current system state relative to the user's perceived current skill. • Motivation: The psychological needs energizing and directing the user to seek out and (continue to) engage with the system-typically competence.
To give one example: In the game chess, the goal is to deliver checkmate to an opponent. The actions are moving her figures on the game board (objects). The rules specify which figures a user can move when and how, and how this affects the game state (a pawn being exchanged for a queen, etc.). The user receives feedback in the form of changing figure positions on the board. The inherent challenge of chess is choosing the move that will put the user in the most advantageous position toward checkmating her opponent, assessing different strategies and predicting likely countermoves by the opponent. The motivation is typically competence need satisfaction.

The Lens of Intrinsic Skill Atoms
Combining design lenses with skill atoms and intrinsic integration, we arrive at the following (Figure 3): This lens articulates the basic structural components of a gameful system and thus sits at the center of our gameful design method. Two modes can be distinguished: The innovating mode serves to create a new system around a target user need, whereas the evaluating mode serves to improve an already-existing system. Both identify the inherent challenges of the target activity and ideate and iteratively prototype and test (new or amended) skill atoms that structure those challenges in a motivating, enjoyable form, involving additional design lenses.

Design Steps in Gameful Design
The design process entails five steps ( Figure 4): First, the designer determines goals and parameters of the design. Formative research identifies user needs, motivations, and hurdles underlying the target activity. If gameful design is appropriate, the designer moves on to synthesis and ideation. In innovating mode, triplets of activity, challenge, and motivation are teased out and translated into brainstorming triggers for new skill atoms. In evaluating mode, the lens of intrinsic skill atoms is used to tease out the skill atoms already entailed in the existing system. Further design lenses are then used to identify issues and generate ideas for improvement. In both modes, the designer storyboards (new or revised) skill atoms and iteratively builds, tests, and evaluates prototypes of them.

a. Define target outcome and metrics
Designers first define the target outcome of the design intervention and how it can be quantified-for example, reducing household energy consumption as measured by household electricity meters.

b. Define target audience and activity
Desk research, interviews, observation, or other forms of data gathering identify activities of target audiences that contribute to the desired outcome. The resulting list of audiences and activities is prioritized by (a) presumed effect on the target outcome and (b) which is best facilitated by improving motivation. Based on project scope, one or several activities may be targeted.

a. Translate user activity into behavior chains
To structure research, a behavior chain analysis (Lynch, Chapman, Rosenthal, Kuo, & Linehan, 2006) or similar method (like customer journey mapping) deconstructs complex activities into chains of behaviors by different actors. For instance, if the target outcome is weight loss, one supporting activity would be healthy eating. A behavior chain analysis might then uncover how healthy eating is preconditioned on both not satiating with unhealthy snacks and cooking healthy meals, which is preconditioned by planning meals and buying healthy foods, and so on.

b. Identify user needs, motivations, and hurdles
With this model in hand, motivations and hurdles for the target activity and its behaviors are identified. Motivations and hurdles are elicited with an amended form of laddering interviews (Vanden Abeele, Zaman, & De Grooff, 2012), mitigating typical issues of subjective reports (like postrationalization) by activating memories of and focusing on recent actual events. Thus, for each target activity or behavior, participants are instructed to first recall and describe in detail a recent actual engagement. Only then are they asked why they did so, instructing them to stay with the recalled engagement. For each reason given, the interviewer follows up with "Why is that important to you?" and repeats this multiple times until a terminal motive is reached. This is repeated for motives and circumstances that kept the participant from the activity (see Deterding, 2012, for a full account).

c. Determine gameful design fit
The analysis of needs, motivations, and hurdles allows one to check whether gameful design is an effective and efficient strategy to achieve the target outcome. This is the case if the following questions can be answered in the positive (cf. Werbach & Hunter, 2012, p. 49): • Does the activity connect to an actual user need? • Is lacking motivation a central issue or opportunity (and not, e.g., poor usability)? • Does the target activity involve an inherent challenge with a learnable skill? • Is affording experiences of competence the most effective and efficient way of improving motivation (and not, e.g., defusing fears)?

a1. Innovating mode: Formulate activity-challenge-motivation triplets
For each targeted activity or behavior, motivations and inherent skill-based challenges are identified. The first guiding questions from the lens of intrinsic skill atoms can be used for this: • What motivations energize and direct the activity? • What challenges are inherent in the activity? What challenges can be removed through automation or improving usability? What challenges remain that the user can learn to get better at?
Results are presented as clusters in the form Activity > Challenge > Motivation and serve as the main input for ideation.

a2. Evaluating mode: Identify skill atoms of existing system
In evaluating mode, the full skill atom lens is used to tease out the implicit skill atom(s) of the existing system. In the case of analyzing an existing car assembly training, for instance, the analysis would ask, What are the goals currently articulated? What actions and objects are learners given? What rules and feedback exist? What challenge and motivation (don't) emerge from this implied skill atom as a result?

a1. Innovating mode: Brainstorm using innovation stems
One main challenge during developing and using this method was providing a design synthesis that enables non-game designers to create apt ideas. A generally useful means for idea generation is innovation stems (VanPatter, 2012). Instead of simply asking people to "have ideas," innovation stems provide prompts that frame the design problem in an inspiring manner. A known commercial game that entails the same challenge as the target activity proved an especially effective stem. For instance, assembling cars and the early electronic game Simon share the core challenge of memorizing sequences. Forced association of two elements also proved useful-specifically, insight combination (Kolko, 2010) of one synthesis result and one design lens. Figure 5 presents the innovation stems that have been most effective in past projects and workshops.
Innovation stems are used as inputs for facilitated brainstorming (Paulus & Coskun, 2012). The facilitator reads one innovation stem aloud and then gives all participants a target number of ideas (e.g., 10) to solitarily write on sticky notes in a set time (e.g., 2 min). Sticky notes are then posted on a wall and explained to the group, allowing individuals to add additional ideas in response.

a2. Evaluating mode: Evaluate skill atom and generate ideas with design lenses
If designers are interested in small amendments, they can ideate promising design changes by applying motivational design lenses (Deterding, 2015) to the identified motives and skill atom components of the target audience existing system. The designer walks through each component of the skill atom and applies the lenses that pertain to that component and/or identified motivations.

b. Prioritize ideas
Ideas are clustered into affinity diagrams and clusters are prioritized, for example, using group dot voting based on likely effectiveness and efficiency.

c. Storyboard concepts
Prioritized ideas are sketched into initial concepts, allowing the same idea to be developed into multiple concepts. Experiments with supporting this step have led to a template storyboard with six panes and instructions to visualize goal, actions and objects, rules, feedback, challenge, and motivation (Supplementary Material 2). This structured forcing of the systemic whole of a skill atom proved useful for non-game designers.

d. Evaluate and refine concepts with design lenses
If desired, sketched concepts can be evaluated and refined using design lenses (Supplementary Material 1). Again, this has proven useful especially for designers new to game and gameful design.

Iterative Prototyping
Concept sketches are translated into working, low-fidelity prototypes and playtested (see Fullerton, 2008, pp. 248-271, for a detailed guide). Failing concepts are discarded, promising ones iteratively developed. The only rule for the kind and fidelity of prototypes is that they should enable testers to actually interact with the core challenge of the skill atom. Paper and craft prototypes often suffice for this. Once the actual "feel" and usability of individual interactions is tested, digital interactive prototypes are usually in order, using, for example, in-browser prototyping with Javascript or one of the many available game and app prototyping toolkits. For instance, in a project involving gameful learning with online videos, one challenge was betting on an answer related to the content of a running video, with diminishing returns and ability to revise one's bet the closer the video was to completion. A video playing in the browser, a question written on paper, some poker chips for betting, and a timer on a mobile phone sufficed to playtest this core challenge.

Evolution of the Method
The presented method was developed and refined in 19 design projects and training workshops with more than 300 participants (Supplementary Material 3). Participants were product managers, creative technologists, user experience designers, domain experts, and digital advertising creatives. Briefs included financial selfmanagement; online video exploration; car assembly, analytics software, and travel agent training; posting classified ads online; using cloud storage; using social business networks; online dating; and online question-and-answer platforms.
Project teams spanned from two to 12 members, involving typically one gameful designer and, on the client side, product managers, user experience designers, and subject matter experts. Typically, an initial client briefing was followed by establishing project goals and constraints, and either reviewing existing or conducting new user research, which was then synthesized into behavior chains, activity-challengemotivation triplets, and/or skill atoms. Ideation and prototyping was done either by the gameful designer or jointly with the client team in 1-or 2-day sprints typically resulting in playtested prototypes accompanied by a design document outlining goal, rationale, and specifics of the prototyped concept.
Training workshops usually lasted a full day and alternated presentation and design, involving one facilitator and 12 to 120 participants, split into teams of four to six. After an hour-long introduction to gameful design, participants received a brief specifying a target outcome and audience. A workbook offered instructions for each step of gameful design, a template skill atom and storyboard, and a set of 26 design lenses. All participants had access to pens and sticky notes for brainstorming and craft materials for prototyping. Each design step was first explained in a 15-min presentation. Participants then had 10 to 45 min to exercise its instructions. 1 Projects and workshops were bookended with structured feedback gathering on the method. Participants individually answered four questions on sticky notes: (a) "What did you take away?" (b) "What remained unclear?" (c) "What would you change?" (d) "What would you keep?" Answers were clustered on a wall and discussed in the group. Both sticky notes and discussion results were recorded for analysis, together with informal field notes. After each project/workshop, main issues of the method were identified, promising changes ideated, and tested in the next iteration.

321
The initial version of the method simply presented participants with a skill atom template and examples for each component of the atom (e.g., CityVille quests as a form of goals). As a result, participants often imitated whatever game example they thought of, typically creating a competition between users, trying to motivate them with rewards that appealed to material self-interest or social status. There was no concern for underlying user needs or for the challenge inherent in the target activity.
The first revision introduced laddering interviews to sensitize participants to user needs and instructed them to identify the core challenge of the activity. To avoid a myopic focus on components, participants had to develop a high-level idea before refining its individual components, using a storyboard. Yet participants found it difficult to move from identified motivations and challenge to a high-level idea. Concepts mostly mimicked the example presented by the instructor.
In the next iteration, explicit synthesis and ideation steps were introduced, with the innovation stem "How might a game around <challenge> look?" This proved to be too abstract for participants without game design experience. This resulted in the present iteration that introduced commercial games with comparable core challenges and design lenses as parts of the innovation stems. The following case studies demonstrate what kinds of results it produces.

Innovating Gameful Design
The client of this case was an online dating platform for heterosexual adults with more than 11 million users across Europe (2011). Apart from me, the design team included the company's head of product, three product managers, and its user experience design lead.

Strategy
The initial brief was to increase overall user engagement, as measured by the frequency of targeted user actions. The team first prioritized five high-level activities (registration, initiating contact, etc.) for design. The target audience was the total platform population, and the target channel was the online desktop version of the platform.

Research
Drawing on existing in-depth user interviews by the company, each high-level activity was broken into behaviors with connected motivations and hurdles. Checking against the criteria for gameful design fit, three activities were found to (a) support platform goals and user needs, (b) suffer from lack of motivation, and (c) involve a 322 Deterding FIGURE 6. Profile of target activity "initiate contact". learnable skill, one of which is focused on here: initiating contact. Strategy and research resulted in short profiles of each target activity (see Figure 6).

Synthesis
The overt user need was finding a partner, and the underlying motivations were relatedness and curiosity. Research also indicated countermotives: fear of rejection, disappointment, and failing to write the "right" introductory message. Using the lens of intrinsic skill atoms, the challenge of this activity was identified as choosing and crafting effective introductory communications. Thus, the triplet for ideation was as follows: Activity: Initiating contact > Challenge: crafting an effective introduction > Motive: curiosity (+), relatedness (+), fear of failing (−) The lens further suggested that challenge was poorly balanced: The perceived stakes and difficulty of the activity were enormous and users' perceived ability small, creating a sense of anxiety palpable from the user research data. The connection between activity and goal was unclear for some users, who expected their counterparts to initiate contact. Available actions were poorly articulated: The platform offered a large number of contact options, with no clear indication which were proper or when they are promising. Also, feedback was highly unclear, as users never knew whether a nonresponse was due to a poorly phrased message, the receiver disliking the initiator, the receiver being busy, and so on. Feedback did not make transparent that one typically needed to initiate contact with multiple users to get a response. Browsing through the available lenses, templates, constrained choice, and actionable feedback were chosen as lenses that seemed to speak directly to these identified issues.

Ideation
The whole design team brainstormed over mixed forced pairings of stem parts, such as How might we alleviate fear of failing with constrained choice? How might we spark curiosity in writing an effective introduction? How might we use templates to make writing an effective introduction more enjoyable? How might we spark curiosity with templates?
A facilitator posed a single innovation stem to the team, and the team then had 2 min to note as many ideas as possible, after which the facilitator moved on to the next stem. For initiating contact, 26 ideas were generated. These included suggesting random, over-the-top starting phrases; a scaffolded series of recommendations how to write messages that would be progressively unlocked as users adhere to them; a counter comparing the user's current total of sent messages with the average number of messages it takes to receive a response; and a random challenge generator that would combine a piece of information from the other user's profile with an absurd constraint (e.g., "ask about his/her holiday without using the letter 'a'"). Team members then distributed three points among ideas for each target activity to mark favorites, prioritizing 10 ideas for storyboarding. A total of 20 storyboards were developed with some ideas receiving multiple treatments. The following concept, "picture match," was eventually not chosen for implementation but can therefore be freely shared (see Supplementary Material 4 for the storyboard).
"Picture match" was inspired by the forced pairing of lacking competence (motivation) and constrained choice (design lens). In short, it is a gamelike challenge offered as an additional contacting form. The initiator takes the following actions upon the following objects: She randomly generates a series of postcards depicting various places, objects, and people. She then writes a question like "Where would I most like to be right now?", orders the postcards accordingly (from most to least fitting), and sends the result as an invitation. When the recipient opens the invitation, she sees the initiator's question, a link to her profile, and the postcards as a scrambled heap next to an array of slots. Her goal is to put the images in the order in which the initiator put them. The inherent challenge is to infer the initiator's preferences based on her profile. After submitting a guess, the recipient can order the images according to her own preference. After the initiator receives her response and guesses her order, both guesses and orders are revealed. The reveal screen entails invites the user to respond with follow-up messages on each (mis)matching couple of images.

Prototyping
The interaction was later prototyped and playtested: two oversized cardboards presenting a "mad lib" form with blanks for questions and photos, and a result screen, papers with brief fake user profiles for the designers, pictures printed from an online image search for "holiday photos," and paper cutouts for screen overlays and messages. Results indicated that the core interaction was working. However, "cheesy" stock images were felt to potentially reflect badly on the sending user (suggesting building a good image pool). Also, coming up with a good question was seen as potentially too effortful or anxiety inducing (suggesting that the system could suggest sentences).

Evaluating Gameful Design
In this case, the client was an online social network for business contacts with more than 12 million users (2012). The design team consisted of me and a user experience designer, with occasional help from the larger user experience design team.

Strategy
The brief was to increase registered users through contact requests by existing users, measured by successful invites sent per user. Using existing data-based personas, the target audience was identified as tech-savvy young professionals who had just started their career: intent on (online business) networking but new to it. We focused on the network's desktop web presence.

Research
We created a behavior chain of the existing web presence for users sending contact requests. Using laddering interviews with target audience members, we identified motivations and hurdles for each step (see Figure 7). The main identified motives were being able to quickly retrieve contact information of even older contacts and social status gains in being visibly connected to "relevant" persons. The network also offered a free month of premium services for successfully inviting a number of new registrants.
We determined that adding contacts was a partial fit for gameful design: Of all involved behaviors, writing fitting invitation messages (and to a lesser extent, choosing the right people to invite) were the only ones that involved any challenge and learnable skill.

Synthesis
We translated choosing whom to invite and writing invitations into a skill atom (Figure 8) whose individual components we evaluated with design lenses. It immediately became apparent that the current interface did not provide goals that structure user activity and provide a sense of progress, something all the more important as the

FIGURE 8. Skill atom of inviting contacts.
Text message "request sent"

Motivation
? Goal Rules number of contact discovery options (address book, Facebook contacts, etc.) and lists of invitees were perceived as overwhelming (lens of interim goals). Also, the system did not offer any calls to action what to do next after having added contacts (lens of next best actions).
The actions and objects of choosing whom to invite failed to limit, highlight, and structure choices into clear and easily doable chunks (lenses of bite-sized actions and limited choice). Instead of one uninterrupted one-click loop of action and feedback (lens of microflow), users had to wade through long lists, marking/unmarking contacts to invite, constantly moving the cursor across the whole screen. Contacts (and marking them) were presented in a standard, bland web form look that did not invite interacting (lens of sensual objects).
The feedback on sending a contact request consisted of a single text line stating "contact request sent." It was neither juicy, nor varied, nor surprising (lenses of juicy, varied , and surprising feedback). It was also protracted and insufficiently helpful (lenses of immediate and actionable feedback): The user could wait days to see whether an invitation was successful, and only if she intentionally clicked her way multiple steps through the menu to see a list of outstanding invitations. Users likewise received no progress feedback of any kind (lens of graspable progress).
The challenge of inviting contacts in the existing interface was an exercise in pure diligence, with no increasing complexity (lens of scaffolded complexity) or variation (lens of varied challenge). Nor were users introduced to and guided through inviting contacts (lens of onboarding). Finally, neither feedback nor the overall skill atom appealed to the underlying motivations of signaling social status or "saving" contacts to an autorefreshing contact list (lens of appeal to motives).

Ideation
We generated ideas for selected issues, using the focusing questions of the respective lenses, noting down and clustering a total of 25 ideas. One cluster revolved around social comparisons to reduce the fear of impropriety and appearing "spamming," for example, displaying "XX% of your contacts have synchronized their address book to always keep their contacts up to date" together with a linked call to action. Other clusters revolved around appealing to motivations, juicy feedback, and guiding and chunking contact options to create a sense of progress.

Prototyping
Using presentation and prototyping software, we created clickable prototypes for the user flow of a newly registered user adding contacts that integrated the developed ideas. We focus here on one single interface element that illustrates the kinds of ideas generated. Figure 9 shows one feedback overlay displayed on successfully inviting users through importing contacts from other platforms. It showcases an appeal to motives ("34 addresses that will soon be automatically kept up to date for you") and next best action ("synchronize contacts from your address book?"), but also juicy feedback: the number of invited contacts dissolves into an animation of butterflies that X invitations sent to GMail be automatically kept up to date for you.
Want to synchronize contacts from your adress book next? Close fly across the interface. The more contacts invited, the bigger the number displayed, and the more butterflies rise up. Also, butterflies change unexpectedly with the kind and number of invitations, creating varied feedback. Animated butterflies fly and "disappear" into the "contacts" wing of a larger butterfly in the sidebar, which visualized progress in the four dimensions of a "well-rounded" profile: contacts, profile completion, events, and group memberships. This provided interim goals and graspable progress (see Supplementary Material 5).

Evaluation of the Method
Gameful design aims to facilitate users' pursuit of their work and everyday needs through interactive systems by affording the motivating, enjoyable experiences characteristic for gameplay. To address the lack of research-based guidance on gameful design methods, we analyzed the characteristics of gameplay experience, how game structures afford them, how game design creates these structures, and how to translate this into interaction design, arriving at a list of six criteria (see next).
A review of existing gamification methods showed them to be pattern-based, data-driven, and wedded to an inherent-additive paradigm of motivation and enjoyment, failing the criteria. In response, we presented an alternative method for gameful design, the lens of intrinsic skill atoms: Its core premise is to identify the skill atoms of the challenges inherent in the user activity and restructure them to optimally afford motivating experiences, using design lenses and iterative prototyping. This method meets all six criteria: Designing for basic need satisfaction, specifically competence (because these are the motivating, enjoyable experiences characteristic for gameplay): Laddering interviews identify user needs and underlying basic needs. The skill atom incorporates these as target experiences and is purpose-designed to afford competence need satisfaction through overcoming challenges.
Designing around inherent skill-based challenges (because overcoming challenges is the core gameplay dynamic affording competence, and because adding artificial challenge to a system supporting functional needs engenders frustration): The method explicitly teases out the target activity's skill-based challenges. The skill atom is organized around this inherent skill-based challenge, and it serves both as ideation input and evaluation focus.
Designing for systemic emergence (because challenge, motivation, and enjoyment are emergent affordances): The skill atom brings together the systemic whole that affords a central challenge and motivation. Design lenses target qualities of the systemic whole of an interaction. Through iterative experiential prototyping, the method helps observing and designing toward actual emergent challenges, motivations, and experiences.
Formative research (because gameful systems need to support given user needs, activities, and challenges inherent in them): The method explicitly includes formative research with specific instructions how to elicit user needs and inherent challenges in a given activity.
Design synthesis of formative research into a form useful for ideation and prototyping (because such bridging has been lacking, even in serious game design): The skill atom and activity, challenge, motivation triplet synthesize research into a form useful for ideation, complete with specific instructions how to do so.
Epistemic mobilization of game design (because its epistemic objects otherwise remain unintelligible to interaction designers): The lens of skill atoms offers a conceptual tool to view a given activity as a game(like) systemic whole. Design lenses further articulate insights and concepts of game design into a self-contained form that is easily intelligible and applicable for interaction designers.

Limitations and Future Work
Although the presented method has been tried and refined in many practical settings, it lacks a formal empirical evaluation of its utility and usability. This is the next intended step.
Second, the method intentionally focuses competence experiences through challenge, although the spectrum of motivating and enjoyable experiences in gameplay is larger by far. Like game design, the method in principle allows skill atoms to be "tuned" to different motives, using design lenses. Still, the structure of skill atoms remains chiefly concerned with competence. Connected to this, the method intentionally focuses on gameful design not playful experiences, although entertainment games regularly afford both. In recent projects, participants were invited to alternatively create playful "possibility atoms" organized around an explorable possibility space, together with supporting lenses. Yet it remains to be seen whether the presented method can be expanded to fully accommodate for playful and motivational design writ large.
Finally, gameplay enjoyment and motivation are deeply contextual (Deterding, 2014a). As noted, the social reframing of situational engagement with a system is likely just as important as its material restructuring to facilitate gameful experiences (Deterding, 2014b). However, we are just at the beginning of understanding the contextual dimension of game enjoyment.

Outlook and Ramifications: Toward Interaction Aesthetics
In 1988, Carroll and Thomas (1988, p. 23) called for "a research program in fun and motivation." A quarter century later, Hassenzahl (2010, p. 43) replied: "User Experience is this program." Yet when it comes to actually designing for fun and motivation, game design has long been (more) forthcoming than HCI. Together, they give us the necessary pieces: here the formative research and evaluative assessment, there the experience-centric design process. Gameful design is a first example of how to put HCI and game design together. Thus, its larger promise is to offer a blueprint for experience-driven design in general.