Causal evidence for social group sizes from Wikipedia editing data

Overview

• The paper examines how human communities exhibit self-organizing properties, with group sizes that align with Dunbar's numbers.
• By analyzing Wikipedia editing histories, the researchers found two peaks in group sizes: around 8 for maximum contention, and around 4 for regular teams.
• These group sizes are consistent with observations of conversational groups and the hierarchical structure of Dunbar graphs.
• The researchers used the Promise Theory model of bipartite trust to derive a scaling law that fits the data, which may apply to group size distributions in other self-organizing communities.

Plain English Explanation

The paper explores how human communities tend to organize themselves into groups of certain sizes, even when those communities are formed spontaneously by strangers. The researchers looked at the history of edits made to Wikipedia pages across a wide range of topics, and found that the sizes of the temporary groups formed to work on those pages tend to cluster around two main numbers: around 8 for the groups where there is the most disagreement and debate, and around 4 for the more regular, stable teams.

These group sizes match what researchers have observed in other contexts, like the typical size of conversational groups and the hierarchical structure of social networks as described by Dunbar's Number. The researchers used a model called Promise Theory to explain why these particular group sizes might emerge, suggesting that it has to do with the way people are attracted to and form trust within these spontaneous communities.

The findings provide further evidence that even loosely-organized online communities exhibit the same self-organizing tendencies as other human social structures. This has important implications for how platforms like Wikipedia can be governed and secured, as well as for understanding the dynamics of all kinds of online social groups and associations.

Technical Explanation

The paper investigates the self-organizing properties of human communities, examining whether specific group sizes corresponding to Dunbar's numbers emerge in the editing histories of Wikipedia articles. The researchers analyzed edit logs across a wide range of Wikipedia subject pages, identifying temporary groups of editors who collaborated on content changes.

They found two distinct peaks in the distribution of group sizes: one around 8 members, corresponding to the maximum level of contention and debate, and another around 4 members, representing more stable, regular editing teams. These values align with observations of the typical sizes of conversational groups, as well as the hierarchical structure of social networks described by Dunbar's Number.

To explain this phenomenon, the researchers employed the Promise Theory model of bipartite trust, deriving a scaling law that fits the observed group size distribution. This suggests that the emergent group sizes may be a universal feature of self-organizing communities, driven by the way people are attracted to and form trust within these spontaneous groups.

The findings provide further evidence that even online communities of strangers, like those that form around Wikipedia, exhibit the same self-organizing properties as other human social structures. This has important implications for the governance and security of platforms like Wikipedia, as well as for understanding the dynamics of all types of online social groups and associations.

Critical Analysis

The paper presents a compelling analysis of how human communities tend to self-organize into groups of specific sizes, even in the context of spontaneous online collaborations. The use of the Promise Theory model to derive a scaling law that matches the observed group size distribution is a valuable contribution, as it suggests that these patterns may be a universal feature of self-organizing communities.

However, the paper does not address potential limitations or caveats to the research. For example, it would be useful to understand how the group size distribution might vary across different types of Wikipedia pages or communities, or how factors like the subject matter or task complexity might influence the observed patterns.

Additionally, the paper does not fully explore the implications of these findings for the governance and security of online platforms. While the authors mention the relevance for Wikipedia and other online social associations, more discussion of the practical applications and potential challenges would enhance the impact of the research.

Further research could investigate the generalizability of the group size patterns across a wider range of online and offline communities, as well as the potential underlying scaling laws that may govern the emergence of these self-organizing group structures.

Conclusion

The paper provides valuable insights into the self-organizing properties of human communities, demonstrating that even spontaneous online collaborations like those on Wikipedia exhibit consistent patterns in group size distributions. The alignment of these group sizes with Dunbar's numbers and the researchers' use of the Promise Theory model to derive a scaling law suggest that these patterns may be a universal feature of self-organizing social structures.

The findings have important implications for the governance and security of online platforms, as well as for our broader understanding of the dynamics of human social interactions and associations. While the paper raises some questions about the limitations and generalizability of the research, it represents an important contribution to the field and encourages further exploration of these self-organizing phenomena in both online and offline communities.