Which Crowdsourcing?

I wrote this article for the most recent APA Technology Division Newsletter, which we sent out this week. Other articles include city apps, water quality mapping, TOD database, a VMT estimation tool, and online participation.

The expansion of the Internet has made possible amazing examples of the collaboration of large groups of people, a phenomenon often called crowdsourcing. Projects like Wikipedia and OpenStreetMap have created new types of encyclopedias and maps. Other projects have coordinated thousands of volunteers to perform major outreach events, such as cleaning up garbage in Estonia or coordinating relief efforts for disasters such as the earthquake in Haiti.

As examples have proliferated, city planners have begun to explore whether the web can be used to tackle urban planning problems. Reviewing some well-known crowdsourcing examples with a focus on urban planning, I will describe four distinct models of crowdsourcing. Understanding these different models and their relative merits is required to integrate successful models of public sector crowdsourcing. The four types are crowdsourcing are:

• Soliciting solutions to problems
• Coordinating many individuals to achieve “collective intelligence”
• Novel combinations of incentives, processes, and staffing to achieve organizational goals
• Peer production of public goods

Some projects have used crowdsourcing as a way of soliciting innovative designs to a problem online. In Salt Lake City’s Next Stop Design project, Thomas Sanchez and Daren Brabham led a team which held an online design competition for a bus stop in Salt Lake City. In a recent article for the journal Planning Theory, Brahbam argues crowdsourcing should be viewed as a new type of public participation. He cites as an example the company InnoCentive, which operates a website where corporations post technical problems and “solvers” compete to win cash prizes for the best solution. “In essence, any urban planning project is predicated on a problem.” Brahbam writes, “Typically that problem is how best to accommodate changing populations with different infrastructure, all while considering the interests of residents, developers, business owners, and the environment. If a problem can be framed clearly, and if all the data pertaining to a problem can be made available, then that problem can be crowdsourced.”

In Melbourne, Australia, Mark Elliott and a team of collaborators took quite a different approach to crowdsourcing for a project completed in 2008. Partnering with an official city planning process, Elliot’s group created a wiki so the plan could be written in the same way as Wikipedia is – through the contributions of hundreds of different authors. In his doctoral dissertation, Elliott proposed a theory of “stigmergic collaboration.” Stigmergy is a theory developed in the natural sciences for a “mechanism of indirect coordination between agents,” such as the ways ant colonies can work in highly coordinated ways without a central authority. Elliott argues this type of cooperation and collaboration is made possible through technologies that create a “localized site of individualistic engagement” that reduces demands placed on participants.

A recent paper by MIT researchers argued crowdsourcing projects should be viewed as innovative arrangements of components, what they call a genome. Through a detailed analysis of the organizations Linux, Wikipedia, InnoCentive, and Threadless, the authors conclude each share a common set of ingredients which fall into four categories: the goal to be achieved, the structure or process of achieving the goal, incentives, and staffing. They observe these projects combine the components in different ways. For example, in the case of Linux, the crowd contributes new software code through collaboration for recognition, but only a small group decides which modules are included in each release through a hierarchy. In the case of Wikipedia, although the crowd creates articles, but the website uses voting and administrators for other decisions, such as whether to delete an article.

Finally, many have speculated that crowdsourcing should move beyond the realm of ideas. Citing examples of massive cleanups and emergency relief efforts, they argue city governments should use technology to crowdsource the production of public services. Instead of the government being the sole provider of certain public services, such as filling potholes or cleaning graffiti, could they simply coordinate citizens to help each other? I am skeptical of such claims for a number of reasons. Governments are subject to unique political and institutional arrangements which make collaborating with citizens difficult. Even if these barriers can be overcome, the flexibility of purely private organizations may be required for a successful project. However, even if governments can’t crowdsource their core functions, there may still be a need for a different approach in this new world. Bas Kotterink, a researcher in the Netherlands, argued in a lecture last summer that the expansion of private crowdsourcing may mean governments should take on expanded roles facilitating innovation, monitoring, and enforcing basic values such as privacy.

Although sharing similarities, each of these models contains distinct assumptions and approaches. Successfully using crowdsourcing for urban planning may require another approach entirely, taking into account the unique characteristics of each city and project. By describing some of the diverse approaches used thus far, I hope this article will help provoke ideas and innovation.

Originally written for APA Planning and Technology Today

Public Sector Crowdsourcing: What’s Possible?

Lately I’ve been involved in a lot of conversations about crowdsourcing in the public sector. Although they’re sometimes confused, in general I think there are two types we can talk about: crowdsourcing policy (or ideas) and public goods (tangible work or services). This is a topic included in my Open Government Strategy for the City of Boston.

The best analysis of private sector crowdsourcing of ideas is this recent article in the Sloan Management Review. The researchers analyze three crowdsourcing projects: Linus, Wikipedia, Innocentive, and Threadless. By breaking down the organization of each case, they make clear these projects not a utopian creative free-for-all, but instead a carefully constructed set of rules and practices that combines forms of decision-making, creativity, and incentives in new ways to create new ideas. For example, Wikipedia relies on the decisions of editors for disputed articles, Threadless users vote for the best ideas, and on Innocentive businesses to pick winners. In each the rewards to the contributors differ, but exist even when they are non-monetary, often in the form of “love” or “glory.”

Two examples of policy crowdsourcing are Peer to Patent and Next Stop Design. Peer to Patent opens patent applications, with the permission of the applicant, to a pilot system which allows the public to contribute to the research on “prior art.” The idea is by allowing experts to contribute to this process, they can accelerate the work of the Patent Office in determining which ideas deserve patents. The project was founded by Beth Simone Noveck, a professor at New York Law School who leads the Obama administration’s open government initiative. The project is successful because it enables topic experts to conveniently contribute information that expedites the official process. However, it remains a voluntary pilot project and has not been taken to scale for the entire government.

The Next Stop Design project, launched by researcher Daren Brabham solicited designs for a Salt Lake City bus stop from around the world. Daren, now a professor at UNC Chapel Hill, wrote a PhD dissertation about public sector crowdsourcing. If you can access it, he lays out his approach in a recent article in Planning Theory. He argues that crowdsourcing can replace conventional approaches of citizen participation:

In essence, any urban planning project is predicated on a problem. Typically that problem is how best to accommodate changing populations with different infrastructure, all while considering the interests of residents, developers, business owners, and the environment. If a problem can be framed clearly, and if all the data pertaining to a problem can be made available, then that problem can be crowdsourced.

Since I’d argue most planning projects involve multiple, contested problems, I’m not sure crowdsourcing can replace a host of existing theory and approaches. However, where the problem contains a significant design element, and the boundaries are noncontroversial (such as a bus stop), it may be an excellent strategy.

Finally, what about crowdsourcing public goods themselves? In the words of Tim O’Reilly, can government be a “platform for greatness”? Last month I argued such thinking was silent to the realities of government: power is divided between agencies, it’s run by politicians, and most people may not agree this is the way to go to begin with. The problems seem more surmountable at a local level. Mitch Weiss, the Mayor of Boston’s Chief of Staff, raised the issue at a provocative talk at the Rappaport Institute titled “How “Peer-Produced” Government Can Help Fill Potholes, Save Cities, and Maybe Even Rescue Democracy.” I worked with him last summer, and I think their initiatives to release data and improve citizen’s ability to communicate with government has been very positive. However, I’m not sure the city will ever be coordinating peer-produced services.

Even if we can overcome the formidable institutional and political barriers, there are good reasons why governments may never be directly involved in facilitating the peer-production or crowdsourcing of public goods. I encountered a good explanation about why this summer at the iGov Research Institute. Bas Kotterink, a researcher with the Netherlands research organization TNO, proposed the following hypothesis in a presentation:

Governments are not geared for co-creation. Instead, they should facilitate and monitor user and company-led innovation of public tasks with a more proactive role in democracy (inclusion) and enforcement, protecting basic human values such as privacy and dignity.

He argued that since the rules of private and government action are so different, initiatives at either extreme are the most able to product public goods. Mixtures of the both – such as some e-participation initiatives – are doomed to fail. However, he stakes out an important role for government. They can ensure minimum standards for key services are protected by punishing offenders and enforcing regulations, or providing it themselves when market failures occur. They can promote data standards and access to public data (such as in apps competitions). They can define and protect standards of individual privacy. Although they may not directly produce certain public goods as in the past, governments will continue to play a critical role we are only beginning to understand.