Transition Rather than Balance: Organizing Constraints for Collective Creativity in Pharmaceutical Development

When it comes to organizing for collective creativity, management scholars typically advise balancing between constraint and freedom. However, this suggestion to balance is often too enigmatic in practice, neglects the dynamics inherent in creative processes, and predominantly highlights the inhibiting aspects of constraints. Thus, the present article aims to rethink this balancing proposition by asking how constraints unfold during collective creative processes. Based on four cases of pharmaceutical development, it illustrates how collective creativity is enabled by constraints from two distinct sources, namely restraint and contingency. The article further shows how and why these constraint sources alternate along the development trajectory. While constraints can enable collective creativity, phases of constraint transition are eventually necessary to revitalize the development process. Building upon these findings, the article concludes that organizing constraints for collective creativity is a matter of transition rather than balance.

Yet, this notion of balancing constraint is often too enigmatic to help organize collective creativity in practice (Caniëls & Rietzschel, 2015;Ortmann & Sydow, 2018;Rosso, 2014). For instance, it remains unclear if and how management should balance between distinct types of constraints (Acar et al., 2018). Furthermore, balance suggestions discount the continuous constraint handling involved with creative processes (Joas, 1996;Lombardo & Kvålshaugen, 2014;Ortmann & Sydow, 2018). Finally, recommendations to balance between constraint and freedom highlight the inhibiting aspects of constraint and severely disregards its enabling characteristics (see Lampel et al., 2014;Ortmann & Sydow, 2018;Rosso, 2014). Thus, the present study aims to rethink the prevalent suggestion to balance constraints for collective creativity in management theory.
To do so, constraints are not examined as static parameters that can enhance or hinder creativity (Caniëls & Rietzschel, 2015), but as dynamic variables themselves, that change and develop during collective creative processes (Fortwengel, Schüßler, & Sydow, 2017;Ortmann & Sydow, 2018). The goal is hence to understand how constraints unfold during collective creative processes. To answer this question the paper uses creativity biographies (inspired by innovation biographies; Butzin, Rehfeld, & Wiedmaier, 2013) from collective processes of pharmaceutical development. Based on four cases, it shows that constraints differ not only in type (Acar et al., 2018), but also regarding their source. Thereupon, the paper illustrates how collective creative processes alternate between phases of enablement and transition regarding constraints. These findings lead to three theoretical contributions.
First, creativity research typically scrutinizes the types of constraints as external factors that hinder or enhance creativity (Acar et al., 2018;Caniëls & Rietzschel, 2015;Lampel et al., 2014;Rosso, 2014). This paper suggests that the sources of constraints illustrate vital aspects regarding the creative process not depicted by type alone. Second, some studies already point to the inextricable interweaving of constraint handling and creativity (Joas, 1996;Lombardo & Kvålshaugen, 2014), stating that transformation of constraints is part of creative processes (Ortmann & Sydow, 2018). Refining this notion, the present article illustrates that phases of collective creativity and phases of constraint transition are distinctly separate. Collective creativity thrives when constraints are taken seriously and are hence sustained by the involved participants. Only after creative insight has been created, do participants try to alter the (now obstructing) constraints. Third, building upon these insights, and upon the notion that constraints enable and restrict creativity (Giddens, 1984;Ortmann & Sydow, 2018;Rosso, 2014;Sonenshein, 2016), this article concludes that constraint transition rather than constraint balance is beneficial to organize for collective creativity.

| Collective creativity
In most of the organizational and management literature, the study of creativity, as the production of novel and valuable ideas, is based on a social-psychological ontology, which examines individual creativity in social context (Amabile, 1988;Amabile & Pillemer, 2012;Woodman, Sawyer, & Griffin, 1993). In contrast, a relational ontology aims to understand the collective origins of valuable novelty (Thompson, 2018, p. 245). There, creativity is conceptualized as an emerging property of ongoing collective processes, which cannot be attributed to specific individual parts and variables (Hargadon & Bechky, 2006;Sawyer, 1999;Sawyer & DeZutter, 2009). Following this relational ontology, collective creativity, as understood in this article, is not about individual creativity within collectives (Kurtzberg & Amabile, 2001), but rather about the emergence of creativity from networks and its embeddedness in social groups (Sawyer & DeZutter, 2009, p. 81). Thereby, the focus of analysis lies on interactions, which "yield creative insight, but cannot be attributed to particular individuals" (Hargadon & Bechky, 2006, p. 484), that is, on the "processual and interactional mechanisms whereby creativity emerges" (Sawyer, 2015, p. 181).
For example, Sarah Harvey (2014, p. 325) argues that group creativity emerges during efforts to integrate contradictory perspectives into a unique and shared understanding (see also Drazin, Glynn, & Kazanjian, 1999). Hence, collective creativity makes another aspect of the complex phenomenon creativity salient. It neither negates nor denies individual creativity, but is instead interested in collective aspects of the generation of valuable novelty (Hargadon & Bechky, 2006, p. 498). In fact, individual and collective creativity coexist and even stimulate one another (Cirella, 2016).

| Constraint and creativity
To remain neutral and inclusive, constraint is here defined as restrictions, limitations, or confinements within prescribed boundaries (Rosso, 2014, p. 553). There are numerous types of constraints in the literature, like structural, resource, and temporal constraints (Lampel et al., 2014); product and process constraints (Rosso, 2014); or capability and coupling constraints (Hägerstrand, 1970) to name a few prominent ones. This paper follows the convincing taxonomy from Acar et al. (2018), which synthesizes these diverse categories into three constraint types, namely input constraint (resource availability), output constraint (outcome requirements), and process constraint (procedural regulations).
Initial management studies regarding individual creativity theorize constraint as antithetical and harmful to creativity (Amabile, 1988).
Instead, freedom and intrinsic motivation are conceived as propellants of creativity, which thrives under slack and freedom (Amabile, 1996).
However, a total lack of constraints can also lead to creativity decrease: for instance, a lack of input constraints can make creators too comfortable (Csikszentmihalyi, 2014), while tight input constraints can stimulate innovation and stretch employees' imagination to find unique solutions (Shalley & Gilson, 2004). Hence, constraints do not only obstruct creativity, but can also pose an obstacle that challenges participants to come up with novel solutions (Marguc, van Kleef, & Förster, 2015). They can inspire and challenge ingenuity by making things difficult, instead of making things easier (Honig, Karlsson, & Hägg, 2013;Lampel et al., 2014;Ortmann & Sydow, 2018).
A similar ambivalent role of constraint is depicted regarding collective creativity, which increases under careful planning of phases and deadlines, well-communicated tasks, and assignment of specific roles, whereas too rigid and formal structures are inhibiting as well (Cirella, 2016). Constraints limit group creativity, but they can also enable it, if contributors are able to accept and embrace them (Rosso, 2014): for instance, process constraints can guide and structure the collective creative process toward better results, as is the case with brainstorming sessions (Stroebe, Nijstad, & Rietzschel, 2010). Likewise, standardization (e.g. routine) and collective creativity are conceptualized as a complementary duality (Elsbach & Hargadon, 2006;Gilson et al., 2005;Sonenshein, 2016).
Moreover, normative constraints can promote free expression of ideas and foster collective creativity as well (Goncalo, Chatman, Duguid, & Kennedy, 2014). Yet again, too tight constraints limit the creative possibilities. For instance, too rigid output constraints limit the possibilities and acceptance of creative surprises (Austin et al., 2012).

| Organizing collective creativity-Balancing constraints?
While these studies all provide very valuable insights on the influence of constraint on creativity, their practical advice for the management of creativity is basically to adjust for compromise: "when constraint trumps freedom or freedom trumps constraint, creativity will suffer" (Rosso, 2014, p. 579). Essentially, management must strike a balance between "underand over-organizing" to foster creativity (Chen, 2012).
However, this insight leads to an all too simple message concerning constraint management for creativity: "tight enough, but not too tight; difficult enough, but not too difficult" (Ortmann & Sydow, 2018, p. 915). In summary, the literature suggests balancing constraint, meaning to establish and keep an optimized middling amount of constraints to organize for collective creativity (see also Cirella, 2016;Lampel et al., 2014). This suggestion rests on the notion that constraints and freedom are mutually exclusive. Yet, balancing constraints raises three distinct issues.
First, balancing constraints entails practical issues for creativity management. Basically, implementing a constraint balance in practice is somewhat enigmatic (Caniëls & Rietzschel, 2015;Ortmann & Sydow, 2018;Rosso, 2014). It remains unclear how such a balance could be achieved, especially regarding different constraint types: through a limited amount of constraints or only partial enforcement?
By leaving some aspects "free", while heavily constraining others?
While balancing constraints seems intuitive, these practical questions are still unresolved in detail.
Second, the balancing suggestion challenges the notion that change is inherent to creative processes. Ideas continually evolve as participants gain knowledge and produce failure (Brinks, Ibert, Müller, & Schmidt, 2018). Similarly, constraints constantly change during creative processes (Joas, 1996). For instance, Ortmann and Sydow (2018) illustrate constraint dynamics during creative processes using Nietzsche's notion of "dancing in chains". During the creative process (dancing), participants put some constraints (old chains) down, while simultaneously establishing (forging) new ones (Ortmann & Sydow, 2018, p. 903). Consequently, creation is never ad nihilum and always results in new constraints. Similarly, Lombardo and Kvålshaugen (2014) illustrate how constraint handling is inextricably intertwined with all creative action and how constraints are shattered during creative action for novel problem solutions. However, such a dynamic perspective on constraints during creative processes conflicts with the standstill implications employed by the suggestion to balance.
Third, arguing for a balance between constraint and freedom discounts the enabling aspects of constraints. The notion to balance constraints falls back to a simplistic understanding that constraint restricts creativity, while freedom enables it (see Amabile, 1996). It rests on the assumption that constraints are necessary for control, but unwanted for creativity, making them a "necessary evil", that should only be employed in the minimal required dosage (Arrighi et al., 2015a;Gilson et al., 2005). However, constraints restrict and enable creativity (Giddens, 1984;Ortmann & Sydow, 2018;Rosso, 2014). Their enabling aspects, though, are mostly neglected in balance suggestions.
Motivated by these three issues, the present article aims to revisit the notion of balancing constraints to foster collective creativity by looking at the unfolding of constraints during collective creative processes of pharmaceutical development.

| METHODOLOGY
This study is conducted by means of a multiple-case design of pharmaceutical developments (Eisenhardt, 1989;Yin, 2014). Every case illustrates the development of a novel and valuable pharmaceutical compound ex-post as a collective process with a tangible creative outcome in a longitudinal account (Brinks et al., 2018(Brinks et al., , p. 1749). This was realized by constructing creativity biographies, thick narrations on the development of a pharmaceutical compound from idea to patent, which were reconstructed from multiple retrospective interviews with essential participants. These creativity biographies follow the methodology of innovation biographies (Butzin et al., 2013), yet they end inquiry at patenting, instead of market implementation. This change was made to focus on the development period (and thus creativity) of the innovation process.
Cases were drawn from pharmaceutical developments, because first, they are large-scale, highly expensive, long-time projects; second, they involve multidisciplinary and complex collaboration spanning wide ecologies; and third, they are characterized by tight control as well as experimentation (Dougherty & Dunne, 2011;Drazin et al., 1999;Sundgren & Styhre, 2003). Therefore, pharmaceutical development cases suitably represent collective creative processes involved with constraint.

| Data collection
To start, suitable cases of creative pharmaceutical development needed to be identified. This was accomplished by selecting tangible creative outcomes in the form of novel and valuable pharmaceutical patents. The novelty of these patents was acknowledged based on recognition (e.g. awards, citations) by the professional field. In turn, the value was verified by comparatively exceptional investments in the development and commercialization of these patents. A list of possible cases was set up through extensive desktop research, thoroughly scanning patent applications, highly ranked scientific publications, industry newsletters, and results of innovation competitions for patents that match the criteria of novelty and value. This search was limited to patents from Germany to support comparability. Another aim was to select cases originating from diverse forms of collaboration, including public research, startups, joint ventures, and international pharmaceutical companies as to achieve a higher level of generalizability (Langley, 1999, p. 706). The resulting list of cases was then presented to an industry expert for confirmation.
Next, the scientists listed on the patent application of the selected cases were contacted. In case of a positive response, narrative interviews about the development process were conducted.
Thereby, the goal was to let the interviewee narrate the development process as freely as possible, by using an open introductory question about the history of the development. Follow-up questions inquired more specifically about decision making, imposed limitations, and aspects of governance (e.g. how collaboration was organized). At the end of every interview, the story was retold by the interviewer to verify or correct the understanding of the process trajectory. Afterwards, further contributors were approached for additional interviews to complete the initial narration. This process was repeated until a clear description of the respective case was established. In addition, four industry insiders from different organizational backgrounds were interviewed about the process of commercial pharmaceutical development in general to supply context information.
A total of 28 interviews were conducted altogether. Initial interviews typically took about 120 minutes, follow-up interviews about 70 minutes. All but two interviews were recorded and transcribed verbatim. Due to the sensitive information present in pharmaceutical development, names of persons, products, and procedures were rendered anonymous. As the interviews were conducted in German, all quotes cited in the findings have been translated into English.

| Data analysis
All interviews from one case were "bundled" into a single creativity biography, a thick narration comprised of overlapping subjective process descriptions (Butzin et al., 2013). Thereupon four "in-depth" cases and three "supplementary" cases were built. A case was classified as "in-depth" when detailed empirical material (at least three interviews), a clear understanding of the development process, and an accountable narration were available. Interviews from supplementary cases as well as those with industry insiders on more general aspects of development were used as context material (a detailed summary of interviews and cases can be found in Table 1). The reconstructed creativity biographies from the in-depth cases were then used as data themselves (Vaara, Sonenshein, & Boje, 2016) to conceptualize phases and detect patterns (Langley, 1999, p. 692).
Initially, the idea of analysis was to understand how different constraint types alternate along the development trajectories. Due to this deductive approach analysis was conducted using qualitative content analysis (Mayring, 2015). Following Acar et al. (2018), the three constraint types-input, output, and process constraint-were used as a coding heuristic to identify when certain aspects of development were constrained. However, it became apparent in coding that all types of constraints pervade every development process, but through different "implementations". For instance, every development process is continuously characterized by input constraints, alternating between constraint resulting from budgeting of resources and constraint from scarcity of resources. These differences could not be fully grasped by a deductive analysis based solely on constraint types.
Subsequently, the aim of the analysis shifted to find an inductive dimension that depicts this notion of "implementation". The ensuing data aggregation was based on grounded theorizing in order to construct analytical dimensions from the collected data (see Gioia, Corley, & Hamilton, 2013). Shifting from data to analysis to theory, it appeared that constraint source is a useful notion to sort and aggregate the data. Source denotes how constraints are implemented and enforced, rather than what aspect of development is constrained, and can be divided into constraint from restraint and constraint from contingency (see Figure 1). The term source as well as the notions of restraint and contingency were established based on the data. In fact, source (or origin) of constraints is little examined in the literature (Acar et al., 2018).
Restraint denotes constraints that derive from control (input constraint), guidelines (process constraint), and objectives (output constraint Contingency denotes constraints resulting from scarcity (input constraint), inadequacy (process constraint), and ignorance (output constraint), like lack of funding, incompetence, inaccessibility, and legal uncertainty. In phases of contingency, the development is based on informal network and community coordination lacking formalized structure and resources. There is no definite deadline, no explicit management plan, no prescribed objective, no legal ramifications, and no contractually enforced partnerships, but also no access to funding, little infrastructure, hardly any human resources, and no legal protection. This source of constraint is denoted as contingency here, because the actual constraints derive from a paralyzing experience of too much possibility without any guiding structure (Luhmann, 1984, p. 152). Hence, constraint from contingency is an unintended consequence of aiming for "freedom" in development.
Lastly, with constraint source and the underlying concepts of restraint and contingency as a coding heuristic, every creativity biography was analyzed again individually using qualitative content analysis. Thereby, the goal was to identify distinct phases regarding constraints during the creativity biographies, by examining (1) the constraint source, (2) the stability of the involved constraints, (3) present (inter-)actions, and (4) the emergence of collective creativity. During the analysis two phases emerged from the data: enablement and transition (for an overview, see Table 2). The qualities of these phases were sharpened through a cross-case comparison.

| Enablement
Enablement describes phases in which constraints are consistently implemented through a steady source, either restraint or contingency.
Thereby, participants accept, embrace, and sustain the present constraints to enable the emergence of collective creativity. Yet, how constraints enable and restrict the creative process depends heavily on their underlying source: If based on restraint, constraints enable complex practices of collaboration with high demands of formal coordination and resources. If based on contingency, in turn, they enable informal practices building on flexibility. It is important to stress, however, that phases of restraint and contingency are not coupled with notions of exploitation and exploration (March, 1991). Both, during restraint and during contingency, participants explore new ideas as well as exploit existing concepts.

| Restraint
Phases of enablement with restraint as a constraint source are characterized by "clear objectives. There is a process which must be addressed, which gives sense to our factory" (LungTreat-3: 7). These objectives "must be a challenging task […] but also […] formulated in a way that you can take on the problem" (CancerStop-2: 4). In addition, there is restraint concerning the involved procedure of development: "the general approach is clear. It is already established" (NatureComp-4: 106). That does not mean that every aspect is defined to the last detail, but that "the steps are clear, the procedure is clear, but the execution is open" (LungTreat-2: 124). Still, development is supervised and controlled: "When you live and work in this context of a company, with investors, with its hierarchies. Well, there is a financial plan and they [management] look relatively closely that everything is in budget" (NatureComp-4: 73).
Yet, these restraints not only limit the creative process, but also enable coordinated execution of the diverse practices involved with pharmaceutical development, which demand a high level of technology, infrastructure, personnel, and knowledge. Restraints synchronize and align these crucial aspects by supplying ways and means "to get down to business" (HeartComp-2: 98). A formally restrained frame supplies "a bigger shovel once you have discovered a golden nugget"  But everything is subordinated to the objective" (HeartComp-2: 33-34).
Especially in moments of crisis collaborators need formal restraint "to push through problems and unexpected events, and not throw the molecule away" (LungTreat-2: 114). This is the enabling aspect of restraint and the benefit for collective creativity. Restrained, structured, and interlocked collaboration is viewed as a functional frame to execute pharmaceutical development and to enable the emergence of creativity: "That is […] the area, which you early described as organized and structured. On the other hand, in such a highly formal process you have the creativity to describe and research the molecule" (LungTreat-2: 46).
While there are severe limitations in restraint, there are also creative possibilities and the freedom to go into detail, to refine, and to enhance development. Thereby, individual constraints can occasionally be altered to meet the refined procedures and ideas; the underlying constraint source, however, is stable. Due to these enabling features, restraints are not opposed by collaborators: "we did everything in-house and thank god for that" (LungTreat-2: 110-112). In turn, a breakdown of or independence from restraint is viewed negatively by the participants in these phases of development: "Everything is harmonized, all the SOPs, standard operating procedures, are coordinated. But then [when they break down] you start to sweat and you try to get everything back to the way it was when it functioned" (HeartComp-2: 88). That way, however, constraints from contingency also enable participants to develop fundamental ideas, overcome conceptual boundaries, make sense of unexpected results, and combine expertise with outsiders. Exactly because there are currently no specific goals or procedures, participants within an informal and unrestrained collaboration can develop an approach which pushes the boundaries of novelty. Similar to phases of restraint, the participants follow and accept these constraints from contingency:

| Contingency
"Everybody plays with your idea, without a task, and without specification.
[…] and the advantage is, without contracts, there is no goal. You do not ask a specific question and get a yes or a no. Everybody does what he or she wants. And all of a sudden you use every brain of every scientist for free" (LungTreat-1: 11).
Participants cannot express a specific task, because they themselves have no concrete idea. Instead, informal contributors engage with the already established team to think of ways of sensible interpretation. Thereby, collaborators look "beyond" their current collaborative setting to find expertise that helps to advance development: "It is part of the creative process to be open. And the group was open and said: We admit it, we lack certain know-how. Now we go interdisciplinary and look for people even if they are spread worldwide" (NatureComp-4: 88).

| Transition
Phases of constraint transition consist of alteration in the source of constraints. When novel insights conflict with present constraints, it is not enough to change constraint types, but the source of constraint needs to be altered to enable new possibilities for development. Thus, during these phases of transition, work on the actual idea is paused, and participants aim to change the underlying constraint source. This transition differs depending whether the source alternates from restraint to contingency or vice versa.

| From restraint to contingency
Unexpected results, as well as failure, repeatedly emerge during phases of restraint "and every interpretation of scientific insight leads to a readjustment or confirmation of your objectives" (NatureComp-3: 80).
Even confirmation can lead to perplexity for the participants: "we saw it worked.
Such ambitions to follow up on newly emerging insights can fundamentally challenge and contradict the restraints that are embedded through the established deadlines, budgets, routines, rules, norms, and roles, which serve as systematic guides for collaboration. Examples from the cases include: the interpretation of new results requires expertise from outside the legal regime imposed on participants (LungTreat); the strategic management plan does not cover experimentation on the emerging insight (NatureComp); a failure during development points to potential for discovery, yet a close deadline and spent budgets inhibit further experimentation (HeartComp). In those instances, the initially enabling constraints become a choke-hold for further development and an obstacle that needs to be overcome.
Therefore, participants aim to suspend, repeal, or circumvent the restraints to follow a new path. However, this transition cannot be achieved by changing certain individual types of constraint. Instead, the underlying source must be altered to initiate phases of enablement again.
Participants argue, for instance, to suspend execution of SOPs, to authorize non-standard procedures, and to include outsiders. Yet, loosening restraint formally is seldom successful, since any deviation "brings pearls of stress-sweat to the forehead of every management type" (LungTreat-1: 9). Thus, participants try to loosen constraints informally. This is accomplished, for example, through unauthorized after-hour work under the protection and discretion of an informal secret: "You must, from such a process, which is probably rather diffuse, develop a good filter-mechanic to connect those few remarks, ideas, approaches, which enable the innovative advance, meaning the big leap" (NatureComp-3: 182).
To address these problems from contingency, participants establish new restraints. Over time, participants synchronize the creative process through defining objectives, limiting participation, imposing legal regimes, setting deadlines, and establishing standard procedures.
They converge on specific restraints that enable the execution of

| CONTRIBUTIONS
The core findings in the present study show how collective creative processes alternate between phases of enablement and phases of transition regarding constraints. During phases of enablement, the present constraints enable collective creativity, based either on restraint or contingency. Yet, the enabling possibilities are limited.
Thus, participants eventually try to change the underlying constraint source. With these findings, the present study makes three related theoretical contributions.
First, organizational and management studies focus heavily on the type of constraint, as the aspect of collaboration or development being restricted (Acar et al., 2018;Lampel et al., 2014;Rosso, 2014).
However, there is hardly any insight into how constraints are imposed and enforced and, moreover, how different forms of their implementation influence collective creativity. The inductively found implementation-based distinction between restraint, as self-imposed limitations and restrictions, and contingency, as unintended constraints based on missing rules and resources, can service as a starting heuristic to reflect these aspects. Thereby, constraints are not external factors that influence creativity, but internal to the mode of collaboration. Hence, constraints are not detached from the creative process, but instead depend on interactions and coordination. Therefore, changing the underlying constraint source is only possible by also changing the way collaboration is organized and conducted.
Second, transformation of constraints and creativity are considered to be closely connected in the literature (Brinks et al., 2018;Joas, 1996;Ortmann & Sydow, 2018). For instance, Lombardo and Kvålshaugen (2014)  Third, organizational and management studies commonly theorize that constraint and freedom constitute a mutually exclusive dualism: the more constraint, the less freedom (Amabile, 1996;Davis & Scase, 2000;March, 1991;Sonenshein, 2016). Based on this assumption, the typical management advice is to balance constraint and freedom to optimize for creativity (Chen, 2012;Cirella, 2016;Davis & Scase, 2000;Gilson et al., 2005;Lampel et al., 2014;Rosso, 2014;Shalley & Gilson, 2017). While constraint is employed to maintain control, freedom is utilized for creativity (Amabile, 1996;Davis & Scase, 2000). In contrast, this study argues for transition instead of balance. For one thing, deviation from or loss of self-imposed restraints does not lead to radical freedom, but rather also induces unintended constraints from contingency. Hence, balancing constraint and freedom does not seem like a helpful notion to organize for collective creativity. Furthermore, suggestions to balance fall back to a simplistic understanding of constraints as only limitation. Yet, during restraint and during contingency, novelty emerges and in both instances participants are enabled as well as restricted by the given constraints (see Giddens, 1984; Ortmann & Sydow, 2018). In addition, constraints eventually conflict with new development paths, either because they restrain novel approaches or because too much contingency limits focused collaboration. Either way, constraint transition is necessary. This transition cannot be accomplished by changing a few selected constraints, but must instead target the underlying constraint source. Therefore, organizing constraints for collective creativity is not a matter of balancing constraint in one optimized constellation. Instead, constraint transition can service as a practicable way to organize for collective creativity.

| IMPLICATIONS AND LIMITATIONS
Building on these contributions, management practitioners and scholars should critically reflect on the notion of balancing constraint to foster collective creativity, leading to three implications for the management of collective creativity. First, there are no per se optimal constraint compositions for collective creativity. All constraints potentially restrict and enable collective creativity (Ortmann & Sydow, 2018;Rosso, 2014), however not indefinitely and not always. What starts out as an inspiration or guidance can turn into an obstacle over time. Hence, transition of constraints is necessary to "revitalize" collective creative processes. Second, currently beneficial constraints should not be "balanced", but fully enforced, since they enable whatever is necessary at that moment. In turn, they should change drastically when those constraints eventually become hindering, again putting a strong emphasis on the transition of constraint source along the creative process. Third, transitions of the source of constraints are resource-and time-intensive phases that pressure participants as well as management.
They are crucial moments during development that can make or break the creative process. Moreover, these transitions can be enacted by participants and are not only initiated by external circumstances. Therefore, management can support the creative process by easing or inhibiting transitions of constraint source at a favorable time. A possible way to advance this research is thus to understand in more detail how management can support or inhibit constraint transition.
The limitations of the present article result from the specifics of the chosen empirical field. By drawing data and building cases from the pharmaceutical industry, the insights are limited to long, expensive, and elaborate creative processes. While data from a single empirical field is advantageous for comparing cases, it limits the validity of the findings to similar fields of development. Especially the need for restraint might not be necessary in cheaper and shorter areas of development with less demand for technology and infrastructure (e.g. arts, music, and literature).