Wednesday, August 23

The Power of Rewards in Social Engineering

One of the projects I have been working this summer has been an extension of the concept of crowdsourcing - where groups of experts contribute to a problem and, like the probabilistic improvement of google search results, Wikipedia's completeness, YouTube's steadily growing collection of content, the overall problem grows more and more defined - leading to the theoretical asymptote of the "best answer". In particular, the project is focused on the concept of scientific data analysis - where enthusiastic supporters can make a difference and, while the percentage of expert contributors may not be high, the collection of a large number of people giving their opinion or best guess can be aggregated into a better solution than what is found from a few selected experts.

One such project that leverages this concept is the recently launched Stardust@home, an Internet-based collaboration project designed to leverage volunteer input in evaluating the potential locations of microscopic stardust particles captured in the aerogel collector from the Stardust mission. As mentioned in a recent press release, in just two short weeks, volunteers have already evaluated 7 MILLION images in search of the stardust particles. To put this in perspective, if this project had been started 20 years ago in the normally, tightly-controlled research effort, the research would still be occurring today. With this "crowdsourcing" technique, scientists assume the effort will be completed in the order of months, rather than years. As mentioned on their website:
This is a completely new approach to doing research. We are in utterly new territory—no one has ever done a project like this before, so we have no experience with it. As with any research project, we will be learning as we go. We will make mistakes (we guarantee it). We ask in advance for your patience. All we can promise is that we will do our best.

This wide-ranging community effort has been steadily refining their results, and even resolving potential "cheating" amoungst the volunteers through a process that tracks the performance of the "players" within the project. Since the community is able to see everyone's performance on their leaderboard, a shared sense of reputation is formed. And given the prize (a trip to the Berkeley lab) is not a four day, all-expense paid trip to some exotic location, the effort put forth by the community is one of altruism.

So, why didn't the Berkeley team use a wiki? When speaking with other people about how best to move forward with our community project, I found a number of attempts to corral the support of other scientific experts - and the easiest, most inexpensive groupware, collaboration tool is often a wiki. Speaking almost from the product specs, wikis simplify publishing, allow for version control and group evaluation/modification of content on the fly. Different from blogs or mailing list, wikis have been successfully used for group collaboration and problem solving - the standard model is to look at Wikipedia or some of the hosted wiki solutions like PBWiki or SocialText. But, does the use of a wiki guarantee group involvement? Will people naturally be inclined to contribute?

It is my assertion that the "Field of Dreams" theory (Build it and they will come.) does not hold true for most situations. In the world of wikis, presence does not guarantee community performance - it is the social engineering that often determines the growth and use of tools that are not part of the standard practice.

Space is a relative term
I am a frequent advocate of using wikis - just ask the students in my classes, the staff on my campaigns, and the business plan teams I have formed - but, I find I have to guide the team to work with the tool. And I am also a fan of my favorite "group comunications platform" (read: eGroups) - which is essentially a mailing list service. But people do not naturally spend time in a "tool", rather then spend time in a "space" - whether virtual or physical. One can establish a social structure, but it does not guarantee adoption of a new process or tool. In a small group where a social structure exists (like a project team), a leader and the team members can be leveraged to enforce the social norms. But, what happens when their is no formal or informal structure to enforce social norms?

I think the secret is found in the many social games that are found across the web and in our normal interactions. With Stardust@home, people are recognized for their contribution to the community. With online games, people are recognized via points or currency. How does it work on wikis? It is my supposition that, aside from Wikipedia is a unique instance of a wiki becoming a reputation game - which rarely happens in other cases. In Wikipedia's case, the unifying characteristic is the "community" of editors - the people who are behind the approval process and who are found on the "History" button of the 'pedia. As you perform within the Wikipedia application, your reputation grows by your contributions. When you are the reference for content that others agree with, you increase your personal standing.

Tags: wikis, social rewards, Stardust@home, crowdsourcing

No comments: