1. Introduction The paper “Marginality and Problem-Solving Effectiveness in Broadcast Search” written by Lars Bo Jeppesen and Karim R. Lakhani in 2010 primarily addresses the question which kinds of external solvers are able to generate successful solutions within an innovation contest when problem information is revealed widely and contest participation is unconstrained.
For that purpose the authors distinguish between solvers who possess deep knowledge and experience in the problem domain and individuals who are “marginal” and have knowledge or approaches from analogous domains that may create effective solutions. The authors’ overall objective is to show that the latter – the marginality of external solvers – is a statistically important predictor of problem-solving process. In addition they postulate two distinct ways of being marginal in problem solving, namely technical and social marginality.
The former describes the distance between the solvers` field of expertise and the focal field of the problem, whereas social marginality is associated with being female, as women have been shown to be less involved in high-status science careers (Jeppesen and Lakhani 2010, Cole and Zuckerman 1984). On these basic principles the authors develop two hypothesises, which are the object of research in this paper. On the one hand they claim that successful solution generation in a broadcast search context will be positively associated with increasing technical marginality.
And on the other hand they hypothesize that being a woman, i. e. social marginality, will lead to effective and winning solutions (Jeppesen and Lakhani 2010). Whereas the academic literature of innovation contests is primarily focused on issues such as the optimal tournament design, incentives for participation, award size, entry criteria or the optimal size of the solver pool, there still exists a high lack of scientific knowledge about what determines who will be a successful solver and who not (Jeppesen and Lakhani 2010).
Another driver that motivates the authors to conduct this study results from research which has shown that individuals who possess high experience and strong knowledge in the problem domain and who are close to the problem often encounter difficulty in solving novel problems (Allen 1970, Duncker 1945, Lovett and Anderson 1996, Luchins 1942, Sorensen and Stuart 2000). By identifying some general characteristics of problem solving Marengo et al. 2000) argue, that differences in perspectives (internal representation of the problem) might turn out to be an even more powerful problem-solving strategy than the decomposition of a problem. In addition, high variances in perspectives among problem solvers lead to similar high variances in the heuristics used by the solvers. As a result, the potential of finding novel solutions increases dramatically and the advantage of marginality arises (Jeppesen and Lakhani 2010).
This view is also supported by research studies in the sociology of science literature that have shown that inventions are usually generated by marginal individuals or so called outsiders (Ben-David 1960). Subsequently, this research paper is encouraged by two main drivers: a lack of scientific studies about how the characteristics of a problem solver impact on the chances of winning the respective innovation tournament, on the one hand, and empirical evidence that marginality increases the chances of reaching breakthrough innovations, on the other hand.
Prior research in economics suggests that having many solvers work on an innovation problem will lead to a lower equilibrium effort for each solver, which is undesirable from the perspective of the seeker. Consequently, the same authors argue that it is optimal to restrict the number of participants to reduce this effect (Terwiesch and Xu 2008). Based on this academic progress, Jeppesen and Lakhani (2010) derive an additional benefit of open up the external innovation process by including previously excluded marginal participants in the problem-solving process.
Hence, and in my point of view, their results contribute to the existing literature by suggesting both the removal of entry barriers for participation for non-obvious individuals and the facilitation of self-selection of problem solvers. The authors provide empirical evidence for the effectiveness of non-restricted innovation contest and with it they also give new reason for further research in this domain. In addition, the findings are also particularly useful for managers who are facing the issue of designing innovation contests in order to receive appropriate and profitable solutions. . Research Methodology The impact of marginality on the successfulness of problem-solving has been measured by considering a unique data set of 166 science challenges involving over 12,000 scientists at InnoCentive. com (IC). For quantifying the marginality effect the authors needed problem as well as solver information. Therefore they used both, the problem and solver information stored in IC? s database and their self-developed online survey that was sent to the 993 individuals that had submitted solution proposals to IC? s broadcasted challenges.
From the IC database they got valuable information as for example the primary scientific discipline of the challenge, whether or not the problem was solved by the seeker and also the scientific interests of the solver and their name. Within a 20-minute, online, web-based survey the participants who had submitted solutions to IC problems were asked about their degree of knowledge of the problem, the effort they put into creating the solution, their motivation for participating in a contest and exerting effort without a guarantee of reward, and demographic information including gender or educational achievements (Jeppesen and Lakhani 2010).
This acquired data provides the foundation for testing the two hypotheses which have already mentioned and is used in a two-stage Heckman probit model, that takes into account the self-selection decisions of individuals as well as the success in broadcast search problem solving (Jeppesen and Lakhani 2010). Using a two-stage endogenous self-selection model is necessary to avoid biased coefficient estimates due to omitted variables that affect both the decision to participate and the resulting outcome (Hamilton and Nickerson 2003).
Hence, the first stage of this model measures in which way the independent variables such as gender, number of previous problems opened, solver interest and problem-discipline match or the award value influence the solver`s decision to submit a solution (dependent variable). But more interesting, in a second step, the study actually examines who becomes a winning solver. Therefore the two independent variables expertise distance, which captures a sense of technical marginality and gender as a proxy for socially marginalized individuals has been applied to the dependent variable solver is winner through a regression analysis.
Whereas the dummy variable gender was quite easy to obtain, the other determinant expertise distance is the result of the answers to a survey question and therefore a self-made estimation of individuals’ perceived distance between the problem and their own field of expertise (Handberg 1984). Thus, the purpose of the second stage is to show if these two determinants of marginality, which are both related to the hypotheses, correlate positively with the success of problem-solving.
As a result, the coefficient of expertise distance is positive and significant at the 5% level, whereas the regression analysis has also shown that women are considerably more likely (at the 1% level) to create winning solutions (Jeppesen and Lakhani 2010). The authors used a well-structured and comprehensible methodology so that the research purposes and findings could be easily followed. Starting the analysis with the adaption of the self-selection decisions of individuals to the regression analysis the authors avoid biased coefficient estimates, which could distort the resulting outcomes of the whole research study.
Due to the fact, that we are considering variables (as for example the gender) that affect both the decision to participate and the resulting outcome, ignoring the self-selection decisions would lead to the so called problem of endogeneity (Jeppesen and Lakhani 2010). Hamilton and Nickerson (2003) argued that failure to statistically correct for endogeneity can result not only in biased coefficient estimates but, more importantly, in faulty conclusions about theoretical propositions.
Therefore, using techniques to correct for endogeneity is extraordinarily important, but frequently missed by a lot of researchers. By using the Heckman probit model the authors completely meet this important requirement. Even though the empirical research gives great insights about the marginality effect on successful problem solving, there are also certain limitations to this study and there are a few concerns about the methodology. These issues will be discussed in the following.
According to the research analysis the independent variable expertise distance has only been measured through survey respondents’ self-evaluation on their perceived distance between the problems they were facing during the innovation contest and their respective fields of expertise. Therefore, doubts arise if solvers are able to precisely define the technical marginality on their own and there may be concerns that the self-evaluation may create errors in this measurement (Jeppesen and Lakhani 2010). Consequently, the evidence of the first hypothesis has to be considered cautiously.
Furthermore, research findings could be biased due to problem holder firms’ procedures to select winning solutions. In the most cases these problems should have been solved by the seeker’s own R&D department before they were broadcasted to external individuals. After spending a lot of time and effort on finding appropriate solutions without success, companies may tend to select external solution proposals that appear to be quite different from their internal attempts. As a result, the winner selection process might be biased due to so called “out-of-field” solutions (Jeppesen and Lakhani 2010).
According to the second hypothesis the authors just have considered the gender for measuring social marginality. However, social marginality depends not only on being a woman. Moreover, factors such as institutional affiliation, academic degree or age determine whether people could be defined as being socially “out of the circle” or not (Jeppesen and Lakhani 2010). Related to the research study let us consider the situation in which many low skilled men had been among the participants who had submitted solution proposals without getting rewarded.
Assuming that individual education influences social marginality (Stanworth and Curran 2007, Jeppesen and Lakhani 2010) we can argue that the low skilled men in our example are socially marginalized. In this case successful solution generation would be negatively associated with social marginality. Hence, in my point of view the validity of hypothesis two should be tested by taking more social marginality influencing variables into account, because the overall objective is to show that marginality, in general, leads to problem-solving effectiveness in broadcast search. Furthermore, with regard to the generalizing of the findings, i. . the external validity, both postulated hypothesis have to be considered critical, due to the self-selection process of individuals which is an essential characteristic of IC? s innovation contests: A possible interpretation of technical marginality would be that the higher the distance between the solvers` field of expertise and the focal field of the problem the merrier is the submitted solution proposal. However, generalizing this hypothesis to the extreme makes definitely no sense and was rejected by McLaughlin`s (2001) concept of “optimal marginality” (Jeppesen and Lakhani 2010).
According to this, optimal marginality implies having access to knowledge of the problem as well as access to resources outside the specific field of problem. The finding of technical marginality by Jeppesen and Lakhani, however, is conditional on the self-selection by solvers into solving a problem. Due to the fact that the solvers have evaluated their relevant knowledge to the problem by themselves before submitting a solution proposal, those solvers that assessed their expertise as to far away from the focal field of the problem may never have participated.
But also the assumption that being female is related with being a wining solver should be treated cautiously and may be not generalized to all existing conditions. Rather Cole and Zuckerman 1984 argued with the “productivity puzzle” that woman are generally less productive than men in science. Thus, the gender effect provided by Jeppesen and Lakhani might also be driven by self-selection. Both, previous studies (Croson and Gneezy 2008, Niederle and Vesterlund 2007) and the results of the first stage of the regression analysis emphasize that women are more hesitant to enter into competition as men.
Thus, women may enter into competition and submit solutions only if they are confident to have a realistic chance to win the contest. This could lead to the fact that the winning solutions came exactly from this group (Jeppesen and Lakhani 2010). Consequently, the research findings have to be treated cautiously and considered as valid just under certain limited circumstances. 3. Related Research Works A central question in the research of innovation contests is whether free entry or restricted numbers of participants should yield better outcomes.
Whereas a high number of scholars have argued that restricting the number of contestants improves contest outcomes in general, and outcomes for innovation contests in particular (Che and Gale, 2003; Fullerton and McAfee, 1999; Nalebuff and Stiglitz, 1983; Taylor, 1995), Jeppesen and Lakhani (2010) have shown that rather than restricting entry, the tendency has been to open innovation contests to all potential participants.
Therefore, it seems to be quite obvious that there still exists a need for further research about dimensions or conditions that determine whether free entry or restricted numbers of participants will lead to effective solutions. For this reason, the paper “The Effects of Increasing Competition and Uncertainty on Incentives and Extreme-Value Outcomes in Innovation Contests (Boudreau, Lacetera and Lakhani 2010)” deals with both, the driving forces hat operate when the number of competitors increases and the question which kind of problems benefit from adding participants into the tournament. Using a unique data set of 9,661 contests the authors first of all have shown that the entire distribution of outcomes decreases when adding the number of competitors and the other way around, that greater rivalry increases the potential to achieve an extreme outcome, i. e. a high quality solution.
The former result is consistent with previous research and the respective phenomenon is called “incentive effect”. The second theory is consistent with both theories on the parallel path effect and the observations of our considered paper (Boudreau, Lacetera and Lakhani 2010). Jepessen and Lakhani (2010) also provide evidence that lifting entry restrictions from innovation contests will lead to higher effectiveness in broadcast search as we have already discussed in previous sections.
However, the most valuable contribution of the paper by Boudreau et al (2010) lies within that part of the study that examines the two effects mentioned above more closely, the (negative) incentive effect and the (positive) parallel paths effect. The authors show that it depends particularly on the amount of uncertainty surrounding the solution to the problem, whether the incentive effect dominates the parallel paths effect or vice versa.
They have shown that adding more participants into the innovation contest increases the overall contest performance for high uncertainty problems, but decreases output for low uncertainty problems (Boudreau, Lacetera and Lakhani 2010). Finally, this research study confirms that both options, restricting participation or offering an open innovation contest without entry restrictions, can lead to effectiveness in broadcast search, depending on the amount of uncertainty that is connected to the problem at hand.
Therefore, the authors support previous studies as well as the findings of the focal paper and expand the academic literature by determining which kinds of problems will benefit from adding competitors in an innovation contest. According to the research findings Jeppesen and Lakhani (2010) have demonstrated how important the inclusion of marginal solvers in innovation contests for successful problem-solving can be. Whereas many scholars before have argued that having experience and knowledge in the problem domain and being close to the problem is more likely to develop successful solutions.
However, in their paper “Intra-Corporate Crowdsourcing (ICC): Leveraging upon Rand and Site Marginality for Innovation” Villarroel and Reis (2010) are providing additional evidence for the advantage of marginality for companies that implement intra-corporate crowdsourcing initiatives. For that purpose they have identified two new dimensions of intra-corporate marginality, namely “rank marginality” and “site marginality”. Rank marginality is defined as being ranked lower in the corporate hierarchy, whereas site marginality describes the spatial distance of an employee from the corporate innovation epicentre.
Their study is based on a common online crowdsourcing platform for innovation inside a large multi-business firm, which operates across multiple business units. Distinguished from the InnoCentive case, the employees contribute their ideas on a voluntary basis, from across business units, geographic locations and hierarchical positions. In this context the authors found that both, lower ranked employees in the corporate hierarchy and employees from sites that are physically distant from the company’s officially acknowledged centres for innovation, are positively associated with better innovation performance (Villarroel and Reis 2010).
While the underlying type of innovation contest differs from our considered paper – external versus internal tournaments – there are similarities between the types of marginality in both papers. Lower ranked employees in a corporate hierarchy as well as socially marginalized individuals may consider innovation contests as an alternative channel and effective mechanism for getting their ideas noticed. In addition, the spatial distance from a company? corporate epicentre can be compared to technical marginality, because employees in sites that are far away may face customers and deal with operational problems that may differ from those in other locations (Villarroel and Reis 2010). Subsequently, it can be stated, that both papers conclude, that marginal individuals are well positioned to possess perspectives and heuristics that are valuable for coming up with novel solutions in innovation contests. Therefore, this paper supports our view in two ways: First, the authors have identified two new, but similar, dimensions of marginality that create an innovation performance advantage.
Equally important, the literature review shows that the results of the focal paper can be transferred to the case of an intra-corporate crowdsourcing initiative. This demonstrates that the marginality effect is adaptable to different types of innovation contests. 4. Conclusion and Outlook Why should innovation contest organizers invite and encourage widespread entry without any participation restrictions? Economic models suggest that widespread entry should diminish competitors’ incentives to make investments and exert efforts (Che and Gale, 2003; Fullerton and McAfee, 1999; Nalebuff and Stiglitz, 1983; Taylor, 1995).
Other work has predicted a positive impact of wider competition through “parallel paths effects”, whereby added competitors may increase the chance that at least one will achieve an extreme outcome (Abernathy and Rosenbloom, 1969; Dahan and Mendelson, 2001; Nelson, 1961). These arguments lead to the following key questions in the design of innovation contests: How many competitors should be allowed to enter the contest? And if entry restrictions have to be implemented, which kinds of problem solvers should be allowed to participate?
Given the profoundly divergent implications of these distinct views, the goal of the present analysis was to highlight the importance of including marginal participants in the problem-solving process via the removal of entry barriers and fostering self-selection of solvers to problems. Overall, the results indicate that technical and social marginality are positively correlated to problem solving success in a broadcast search setting. Consequently, these findings support the approach to open up innovation contests to more potential participants.
Furthermore, this paper also provides answers to the question, which external solvers are able to come up with successful solutions. However, the discussion of possible limitations of the focal paper as well as the critical discussion concerning the validity of the results has shown that the effects of marginality need to be approved by further research. In particular the missing external validity of technical marginality combined with McLaughlin`s concept of optimal marginality rather support the view that marginality facilitates successful solutions to broadcasted problems, if applied in a certain degree, but not in the extreme case.
Hence, an interpretation of the research findings could be as follows: the rewarded winners of the innovation contests were optimally marginal individuals, who were sufficiently distant from conventional approaches, but at the same time were not too far away from the problem, so that they still had access to interesting problem information. This view might be driven by the phenomenon of self-selection by solvers into solving a problem, which means that those solvers that assess their own expertise as being to far away from the field of the problem at hand may never have participated.
Future research in this domain should explore the marginality effect by considering more marginality driven dimensions. The phenomenon of marginality cannot be totally explained with social and technical marginality in the way these dimensions are used by the authors. In my point of view it could be interesting to determine additional dimensions of marginality that highlight the innovative potential of other marginal individuals. As a result the advantage of marginality and particularly its external validity can be supported by further empirical evidence.
For example, individuals who are cultural marginal in terms of using different working methods or who are just thinking in different ways could be providers of highly innovative solutions as well. While participating in an innovation contest these solvers may profit from their distinct working and living conditions or different cultural habits. In addition, it would be also interesting to know for what kind of problems which kind of marginal individuals are useful in a company’s problem solving process.
As broadcast search continuously becomes more important, such results would be particularly valuable for practitioners and could be a valuable contribution to the design of more effective innovation contests. References Abernathy, W. J. , R. S. Rosenbloom. 1969. Parallel Strategies in Development Projects. Management Science 15(10) 486-505. Allen, T. J. 1970. Communication networks in R&D labs. R&D Management 1(1) 14–21. Ben-David, J. 1960. Roles and innovations in medicine. Amer. J. Sociol. 65(6) 557–558. Boudreau, K. , Lacetera, N. , Lakhani, K. 2010.
The effects of increasing competition and uncertainty on incentives and extreme-value outcomes in innovation contests. forthcoming. Che, Y. K. , I. Gale. 2003. Optimal design of research contests. Amer. Econom. Rev. 93(3) 646–671. Cole, J. R. , H. Zuckerman. 1984. The productivity puzzle. M. L. Maehr, M. W. Steinkamp, eds. Advances in Motivation and Achievement. JAI Press, Stamford, CT, 217–258. Croson, R. , U. Gneezy. 2008. Gender differences in preferences. J. Econom. Literature 47(2) 1–27. Dahan, E. , H. Mendelson. 2001. An Extreme-Value Model of Concept Testing.
Management Science 47(1) 102-116. Duncker, K. 1945. On problem solving. Psych. Monographs 58. Fullerton, R. L. , R. P. McAfee. 1999. Auctioning entry into tournaments. Amer. Econom. Rev. 107(3) 573–605. Hamilton, B. H. , J. A. Nickerson. 2003. Correcting for endogeneity in strategic management research. Stategic Organ. 1(1) 51–78. Handberg, R. 1984. Response to Gieryn and Hirsch. Soc. Stud. Sci. 14 622–624. Jeppesen, L. B. , K. R. Lakhani. 2010. Marginality and Problem-Solving Effectiveness in Broadcast Search. Organization Sci. Articles in Advance 1-18.
Lovett, M. C. , J. R. Anderson. 1996. History of success and current context in problem solving: Combined influences on operator selection. Cognitive Psych. 31(2) 168–217. Luchins, A. S. 1942. Mechanization in problem solving: The effect of Einstellung. Psych. Monographs 54. Marengo, L. , G. Dosi, P. Legrenzi, C. Pasquali. 2000. The structure of problem-solving knowledge and the structure of organizations. Indust. Corporate Change 9(4) 757–788. McLaughlin. 2001. Optimal marginality: Innovation and orthodoxy in Fromm’s revision of psychoanalysis. Sociol. Quart. 2(2) 271–288. Nalebuff, B. J. , J. E. Stiglitz, 1983. Prices and incentives: Towards a general theory of compensation and competition. Bell Journal of Economics 14(1) 21-43. Nelson, R. R. 1961. Uncertainty, Learning, and the Economics of Parallel Research and Development Efforts. The Review of Economics and Statistics 43(4) 351-364. Niederle, M. , L. Vesterlund. 2007. Do women shy away from competition? Do men compete too much? Quart. J. Econom. 122(3) 1067–1101. Sorensen, J. B. , T. E. Stuart. 2000. Aging, obsolescence, and organizational innovation. Admin. Sci.
Quart. 45(1) 81–112. Stanworth, M. J. K. , J. Curran. 2007. Growth and the small firm: An alternative view. Journal of Management Studies 13(2) 95-110. Taylor, C. R. 1995. Digging for golden carrots: An analysis of research tournaments. Amer. Econom. Rev. 85(4) 872–890. Terwiesch, C. , Y. Xu. 2008. Innovation contests, open innovation, and multiagent problem solving. Management Sci. 54(9) 1529–1543. Villarroel, J. A. , F. Reis 2010. Intra-corporate crowdsourcing (ICC): Leveraging up rank and site marginality for innovation. CrowdConf. 2010. San Francisco, CA.