When the editors at Encyclopedia Britannica argue, no one knows about it.
Compare that to Wikipedia, where disputes between editors play out in real time over the soccer team Celtic Football Club, the wrestler Triple H or the video game Super Smash Bros. Brawl. Each has been edited more than 15,000 times by experts, fans and antagonists who clash over on-field controversies, onstage personas and online gaming experiences.
Does all this help the editing process? A researcher at the School of Business at Stevens Institute of Technology said that, to some degree, this digital back-and-forth does correlate to higher-quality articles.
Assuming 'an amorphous maelstrom'
“People assume Wikipedia is complete chaos — an amorphous maelstrom of activities that move forward without structure or pattern,” said Dr. Aron Lindberg, an assistant professor of Information Systems at Stevens. “But we were able to trace the arc of an article and better understand the role conflict plays in determining an article’s quality.”
Dr. Lindberg and his co-authors developed a methodology, based on text mining, to trace the evolution of articles over time. Linear articles, with only a handful of contributors, tend to be very dry and limited in scope, while “too much exploration,” as Dr. Lindberg put it, tends to feature endless revisions that become increasingly partisan; a review of the most-revised articles on Wikipedia bears this out, as they are mostly political or religious matters.
Dr. Lindberg’s methodology is aimed at letting experts better understand how the trajectory of articles’ evolution can be shaped in order to maximize quality — an important concept, with artificial intelligence playing a larger role in managing these processes. It also might provide clues that could help future researchers apply this methodology to platforms and ecosystems beyond Wikipedia.
“You want that happy medium, where conflict is a source of creative tension, not destructive tension,” he said.
Dr. Lindberg is an expert in digital innovation and open-source software development, which makes research on Wikipedia a natural extension of his work. But he’s hardly the only scholar looking at the open-source encyclopedia; students from the undergraduate to Ph.D. levels have done research on the site alongside faculty like Dr. Jeffrey Nickerson, associate dean of research and an expert in crowds and collective intelligence.
Dr. Nickerson explores what he calls design activity — the coordination of people in 3D printing communities and in workspaces like Wikipedia where there is no formal mechanism for coordinated activity. He also recently completed an analysis of Wikipedia’s more than 1,600 bots, which make edits, connect entries, and find and purge vandalism, with an eye to better understanding the ways machines and humans collaborate on the platform.
His interest in the topic goes back to when a former Ph.D. student, Pinar Ozturk — now an assistant professor at Duquesne University — started researching the topic for her dissertation.
“I always wonder why more academics don’t study this thing,” Dr. Nickerson said. “I think many of them don’t because they think there’s only one Wikipedia, that it’s some kind of anomaly. But that’s not true.”
There are actually hundreds of Wikipedias, he pointed out — one in every language, plus the Wikia fan communities for endless topics, “and many companies use internal wikis based on the same technologies as Wikipedia.”
Better understanding bots
Dr. Lindberg, who has not formally collaborated with Dr. Nickerson on this research, also said the role of those bots in managing processes is ripe for future research.
“Wikipedia’s processes are so widespread across time, space and people you don’t know that it’s hard to get an overview of it, like not being able to see the forest for the trees,” Dr. Lindberg said. “Using the methods we developed in this paper, using statistical analysis, you can start to see patterns, structures, regularities in the chaos. But you need to find ways to use A.I. to identify those patterns on the fly and make recommendations.”
Dr. Lindberg’s work has been accepted in Management Information Systems Quarterly, and is forthcoming. Getting a better understanding of how innovation is created in complex systems — as well as how to manage innovation in such situations — is what he hopes to continue doing in the future. The Britannica example offers a useful analogy for the future of his work.
“Britannica is this great 1.0 version of a physical artifact,” he said. “It’s an industrial era-product — one that still has value, but it’s a stable, mature part of the economy. The 2.0 digital artifact is Wikipedia. That’s where innovation happens, because the value isn’t created by one corporate entity, but by this society of people and intelligent machines working together.”