The web has experienced a recent proliferation of design expert communities in domains from software engineering (e.g. Sourceforge and Github) to art (DeviantArt and others). These communities have become hotbeds of creative interaction, with users posting their projects, closely interacting on new endeavors, and engaging in spirited discussion about their craft. With users in these communities constantly generating out new software, images, music and any other artifact imaginable, it is hard to deny that there is significant creative interaction happening. Members of these communities often possess widely varying degrees of proficiency, but more often than not, they have some baseline amount of talent that allows them to enter the community.
Enter Picbreeder. Picbreeder is a web-based system for collaborative interactive evolution of images. The Picbreeder applet starts by randomly generating several images, which are then mated and mutated based on the user’s selections. The user can then publish the image to the Picbreeder website where other users can download and continue the image’s evolution. Within Picbreeder, one need not have artistic talent to contribute to the community, although good taste typically helps. As in more traditional design, new innovations are typically small modifications to the existing structure, which can change the design incrementally or effect a larger shift. Even though users followed their individual interests when evolving this phylogeny, new interesting directions emerged. Many users contributed repeatedly to an evolving lineage, using the design itself to encourage and facilitate collaboration.
Successful collaborative design in Picbreeder does not require shared intentions, suggesting that effective collaboration may be emergent rather than planned from the top down. The surprising result of this emergent process is the gradual discovery by untrained users of hidden treasures within a vast uncharted space. Picbreeder also serves as a fascinating, though initially unintentional, experiment in stigmergic creativity.
The concept of stigmergy was first introduced by Pierre-Paul Grassé, a zoologist, who used it to describe the activities of the termite mound. As he described it, “(s)tigmergy manifests itself in the termite mound by the fact that the individual labour of each construction worker stimulates and guides the work of its neighbour”. The concept of stigmergy can be extended to human endeavors if one expands the notion of the mound to human venues, and replaces “construction worker” with any type of worker. If such an extension is permitted to human creative communities, this description becomes even more apt. Part of the excitement inherent in creative pursuits, whether it is visual art, music or creating open source software, is the moment when the work of a colleague “stimulates and guides” ones own work. Add that “(in) an insect society individuals work as if they were alone while their collective activities appear to be coordinated.” This description too can apply to creative communities. Points out that “(s)tudies on creativity . . . have focused on the individual, obscuring the fact that creativity is a collective affair. The ideas and inventions an individual produces build on the ideas of others (the ratchet effect).” It is very easy to focus on individual creative luminaries, while forgetting the environment and social milieu that are a large part of their creative interaction.
The results of Picbreeder not only demonstrate the truth of creativity as collaboration, but that a large component of creativity can be stigmergic. By abstracting out almost all direct communication and collaboration, and allowing users to be stimulated only by their work and the work of others, Picbreeder demonstrates the extent to which stigmergic processes can yield astounding results. This paper expounds on this point by first describing in detail what Picbreeder is and how it works (section 2). Next, the paper casts creativity in general and Picbreeder specifically into the context of memetic evolution, a model of how ideas spread, change, evolve and die out (section 3). The point is then made in section 4 that these collaborative creative environments draw a great deal of their effectiveness from stigmergic interaction facilitated through creative artifacts. In sections 5, an analysis of the Picbreeder data is described that shows, despite the fact that Picbreeder users engage in almost no direct communication, it shares numerous properties with other collaborative creative environments. Finally, some conclusions and recommendations are made in section 6.
This paper has shown that Picbreeder, an almost fully stigmergic means of collaborative creative interaction, follows many of the same patterns as other collaborative creative networks. Picbreeder demonstrates that it is possible to facilitate creative collaboration through entirely stigmergic means, and this paper explored the mechansisms that gave rise to that stigmergy. Because in other creative communities, stigmergic and non-stigmergic components of creative interaction are difficult to separate, Picbreeder provided an ideal opportunity to study this dimension. It is hoped that future studies will be able to isolate and study the contribution of stigmergic components in other creative communities.
It is also hoped that more quantitative analysis will be done on other creative communities. Academic publishing bibliometrics were used because they are plentiful and easy to access. While it is difficult to trace influence in similar way in musical or visual arts communities, developing techniques to analyze these communities is a worthwhile pursuit. This analysis may provide answers of real economic value. For instance, to answer the question, what will create a broader, more economically viable base of musical development, a U.S. style system in which music distribution is dominated by a few large gatekeepers to the music industry, or a Canadian style system which frequently uses government sponsored incentives to encourage development in musical communities?
There is a great deal of analysis left to be done and questions to be answered with respect to the dynamics of creative communities. For instance, how can Axelrod’s model of cultural diffusion (1997) explain creative influence? Also, how can Friedkin’s analysis of weak ties versus strong ones in organization flows (1982) inform the analysis of how creativity develops within and between organizations. Picbreeder is currently a “flat” community, which does not fully represent the wide variety of social creative arrangements. The addition of this dimension to analysis will hopefully yield additional insight.
Stigmergy is clearly involved in creativity. It is no accident that Silicon Valley is well known for technical innovation and Paris is a well known muse of artists. These physical locations host large collaborative and competent communities for one, but also frequently display and demonstrate the results of their interaction, to “stimulate and guide” other participants. Other creative communities might benefit by explicitly taking advantage of stigmergic concepts to improve their efficcacy. Imagine a paint studio where artists paint in a circle, with the paintings facing inward. Or a research lab where everybody’s latest work in progress is posted to a highly visible electronic board. The more we understand the role of stigmergy in creativity, the better we can shape and guide the process. Ultimately, every creative discipline, along with humanity itself, will be the beneficiaries of this advancement.