Mathematical model 'describes' how online conflicts are resolved | University of Oxford

Mathematical model 'describes' how online conflicts are resolved

Researchers have produced a mathematical model to describe how conflicting opinions are resolved over articles that appear on Wikipedia, the collaboratively-edited encyclopaedia.

The study maps the evolution of opinion over time, showing that even widely diverging opinions eventually converge. The researchers say this pattern in collective human behaviour is reminiscent of the interaction of particles in physics, such as when wind-blown grains of sand eventually create sand dunes. The study also notes that the collaborative nature of a medium like Wikipedia tempers extremes of opinion.

The model, which appears in the journal Physical Review Letters, was constructed by a research team, including Dr Taha Yasseri from the Oxford Internet Institute at the University of Oxford, as well as researchers from Hungary, Finland and Spain.

Dr Yasseri says: 'We believe this model exhibits some interesting behaviour that for the first time can qualitatively capture key characteristics of off-line conflicts.

'The competition and co-existence of local and global interactions play an important role in physics. For example, we have already observed this in the creation of sand dunes, which are not solely formed by the local interaction of sand grains with neighbouring grains but also the wind, which brings into play global and long-range effects. To understand the emergence of editors' opinions, again we see both the direct interactions of editors with each other and the global interactions of editors altering a shared medium.' 

At the outset, the researchers mapped widely differing opinions amongst the 'editors' of three articles that appeared on Wikipedia about the Dresden bombing, Japan and anarchism. Editors were seen to influence each other either through direct communication or indirectly by modifying the online Wikipedia article.

The researchers found that when a fixed number of individuals form one 'mainstream' and two opposing 'extremist' groups, consensus in the medium's content is achieved, but only after a long time and this may not correspond to the initial mainstream view. In the case of a dynamic environment where new editors replace existing ones, they found periods of conflict and consensus can alternate indefinitely, depending on the rate of newcomers and the degree of controversy in the article's topic.

The group's model describes the spontaneous dynamics of the clashes of opinions in Wikipedia in three types of scenarios: First, it shows how increasing the number of new editors to an article gives rise to controversy. The model also accounts for different levels of tolerance in editors' sensitivity towards certain topics. Moreover, it captures states of uninterrupted controversy where sensitive contributors cannot stand even small deviations from their opinions. However, it shows that eventually even strongly opposing views converge over time, even without direct interaction between dissenting contributors. The researchers conclude that the shared medium takes care of this.

'The presence of the Wikipedia article itself brings the opinions of individuals together and helps the convergence process. Without an article on which to work collectively, groups with different opinions could stay separate and ignore each other,' explains Gerardo Iñiguez, doctoral student in the group and in Aalto University Department of Biomedical Engineering and Computational Science.

'Usually sand grains only collide with neighbouring grains, but wind can project them into the air and make them hit faraway grains. This long-range global interaction creates a ripple pattern that does not appear in the absence of wind,' adds Iñiguez.

Dr Yasseri concludes: 'In Wikipedia local interactions occur between editors modifying their opinions, while global interactions are those of the editors with an article. In the presence of an article a collective behaviour appears, conflict followed by consensus, not unlike in chemical synchronisation and sand ripples.'

János Kertész from the Center for Network Science, Central European University, Hungary explains: 'New media produce an efficient collaborative environment but with collaboration comes conflict. Wikipedia is an important example because every single change is documented and publicly available. It's a goldmine for research.

'Our model is a tolerance model: If the article is very different from an editor's opinion, they are more likely to change the article. If it's close to their opinion, the editor is less likely to change it. Is usually a few people who are rigid and keep the edit war going.'