Jeffrey Mackie-Mason Speaks On User Contributed Content
Country music set the tone for University of Michigan Professor Jeffrey Mackie-Mason’s talk at the Yahoo!, the 2nd of the year in the Big Thinkers Series. He introduced his talk with a well-known illustration of Tom Sawyer that immediately grabbed the audience’s attention. As the music faded, the computer science and economics professor started reading a passage of the Mark Twain novel. The illustration of Tom Sawyer and a friend whitewashing a fence remained in view; the caption beneath it stating “Ain’t that work?” Mackie-Mason compared the 19th century novel to the idea of user contributed content and incentive centered design. The best known passage in the book describes how Tom Sawyer cleverly convinces his friends to whitewash a fence for him, and in the process, became “wealthy” by giving his friends a chance at whitewashing the fence in return for goods.
Mackie-Mason explained the idea of user contributed content as an input to production that is provided without direct, intrinsic compensation. The concept is not new, and there is no denying its growing significance. He cited several examples today, including open-source software, wikipedia, and flickr. But according to Mackie-Mason, there are problems associated with user contributed content. The systems in which user contributed content works well rely on the interactions between people, information and technology. If any one of these is removed, the system falls apart. Humans are a) autonomous – they don’t always do what you tell them to do; and b) motivated – there is intentionality in their behavior. It is important to draw on the sciences of motivated behavior to understand human intentionality in systems.
“The main problem in user contributed content is getting stuff in,” said Mackie-Mason. “Why not just pay for it,” he asked. The most obvious answer is that most people can’t afford it. He brings up the wikipedia example. With just 10 employees, they set out to create a huge database of information – all through user contributed content. He raises another question. In a system where a large number of potential contributors make small contributions, how do you ensure the quality of the contributions? Quality depends on a mix of contributions, such as getting users to participate well and attracting the right users -- which reveals another user contributed content problem: pollution.
Mackie-Mason reveals two types of pollution: spam, and negative quality. The possible solutions include human editors and spam filters. “How do we convince the polluter to self-screen?” Mackie-Mason asked. “What effective incentives would convince polluters to self-screen?” A number of possibilities were introduced, including penalties, canned spam acts, and more drastic measures such as threats to discourage negative contributions. So, how do you configure economic, social and psychological incentives to induce contribution, discourage pollution, and motivate sufficient effort to generate quality?
Mackie-Mason gave an example of a design model where users post stories. In this model, Mackie-Mason explains that there are typically 2 types of users: truth and liars. The concept is to make the liars, or polluters, pay for their content. However, in the case of digg, a social networking site in which users share content in various forms, the value of a lie turned out to be positive. There are some conditions under which digg wants polluters to participate. Why? Digg may get negative value for the content, but they are willing to publish the content if the price is right. The expected fine for polluters is the price. Mackie-Mason explains this as advertising. “Not that ads are lies,” he clarifies. “But it is commercial expression that is self-interested.”
Incentive centered design is one approach for dealing with user contributed content problems. It is pragmatic and already used in prediction markets (the Hollywood stock exchange), reputation and recommendation systems (such as Amazon reviews), and auctions (applying behavioral principles to auctions).
Mackie-Mason explained the concept of “getting the good stuff in” in incentive centered design through del.icio.us – a social bookmarking site that relies on people posting and tagging bookmarks. “Why do people bother contributing their bookmarks to del.icio.us?” asked Mackie-Mason. He explained that it’s because it’s a bookmark management system that allows users mobile access to their bookmarks and fosters peer recognition.
What can we conclude as the design lesson? Combine private reasons to contribute – where users gain direct personal benefits for themselves) with public side-effects (a system in which other people can benefit by finding interesting information). This example of incentive centered design is the stylized mechanism that relies on user-contributed content to screen out polluting contributions.
