David A. Bray: Distributed Problem-Solving Networks

Advanced Cyber/IO, Collective Intelligence, P2P / Panarchy, Software
David A. Bray
David A. Bray

Back in 2007-2008 we did research relevant to your thinking on Applied Collective Intelligence, at the University of Oxford.  We focused on “distributed problem-solving networks” that included looking at film production in a distributed fashion to include a lot of open source projects:


Published papers and book chapters, including:


Recap with links from the Oxford link:

I. Project Framework, Synthesis and Commentary

The lecture ‘The Power of Diversity: The Performance of Distributed Problem Solving Networks‘ along with invited responses was designed to inform the debate about the potentialities and limitations of these new organizational forms. View the webcast of the lecture and discussion.

II. The Case Studies

1. Sermo: A Community-Based Knowledge Ecosystem

Exemplary of Web 2.0 developments, some problem-solving networks are anchored in user-generated content, such as Sermo: a community-based knowledge ecosystem for licensed physicians in the USA. Physicians can ask and answer questions and surveys posed by other doctors or pharmaceutical firms or other paying problem-holders. The Sermo community, of over 50,000 doctors, sorts through conversations and identifies interesting health trends, cases, and other novel health insights for the benefit of multiple stakeholders.

2. Seriosity: Addressing the Challenges of Limited Attention Spans

Human beings like to earn points, as demonstrated by Seriosity, a creative use of games as way to create an incentive for individuals to pay closer attention to their use of email and solve their problems with information overload. The system enables individuals to use a virtual currency attached to email, to simulate the redistribution of resources in ways that will lead them to be more strategic about the priority messages they send, read, open from co-works. The Seriosity approach has interesting side-effects, to include individuals exchanging the virtual currency for real-world tasks and favors in the workplace.

3. The Performance of Distributed News Aggregators

Originating as a study of the online news aggregator Digg, which relies on user ratings to determine which articles to put on the frontpage, this study evolved into a comprehensive survey of the online news aggregator space. This case study explores the conditions under which crowds are smart, analyzes the bias of several modes of information aggregation, and shows the risk of mob behaviour.

4. Information Markets: Feasibility and Performance

The performance of prediction markets has been one driving force behind the renewed attention on distributed problem solving. This case reviewed the feasibility and performance of prediction or information markets, discussed some apparently successful applications, considered likely limitations of information markets, and identified important areas for future research.

5. The ATLAS Collaboration: A Distributed Problem-Solving Network in Big Science

The nature of the problem tackled by the ATLAS collaboration – the creation of a radically innovative particle detector experiment – makes ATLAS an exceptional case for studying DPSNs. The problem solving is distributed across multiple groups of problem solvers comprising 2000 scientists in 165 working groups across the globe. Similarly, the engineering, construction and installation of the many components is distributed across this collaborative network. The initially surprising finding of the case study is that this joint innovation effort succeeded despite breaking with most rules of traditional project management. Philipp Tuertscher analyzes what it took to make such a loosely structured organization work, and raises the question if such structure was even required to develop a complex technological system like the ATLAS detector?

6. Bug-Patching for Mozilla's Firefox

The case study of the Mozilla project is focused on the organization of quality control and quality assurance in a distributed innovation environment, and focuses on the coordination of the detection and correction of operating defects (‘bugs') in Mozilla's Firefox web-browser. Analyzing two samples of bugs drawn from the 40,000 or so that have resulted in a change to the Firefox code base, the case study finds among others that bug treatment behavior in the project was not temporal stable, that bug reports from ‘outsiders' took longer to find a successful resolution and were more likely to remain ‘un-fixed and that factors such as the objective technical complexity of the bug-patching problem, and the level of effort devoted to the contextualisation the reported defect played significant roles determining the speed with which a bug was typically fixed.

7. Wikipedia as a Distributed Problem Solving Network

Wikipedia, the free online encyclopaedia put together by volunteers, is a prime example of a distributed problem-solving network with a global array of contributors creating a resource that has been compared to leading encyclopedia. The study focused on efforts to maintain the quality of Wikipedia entries and in particular of the use of tagging to signal the need for improvement in entries of Simple Wikipedia.

8. Distributed film production: Artistic experimentation or feasible alternative? The case of ‘A Swarm of Angels'

A Swarm of Angels was selected as a case study of open content film production. The project based in Brighton, England, but it extended the open source model to movie making in ways that could bring distributed collaborators into the film project. The case study highlighted the existence of a core group of contributors and a periphery of silent supporters, which both play an important role in the project's performance.

III. Cross-Cutting Thematic Papers

Financial Liberty at Risk-728x90