Ellen Phiddian
Cosmos science journalist
Holding a hallowed place at the top of search engine results, Wikipedia has become a profound influence on the information age.
With articles written and edited by millions of volunteers around the world, it’s been hailed as “the last best place on the Internet”, and is often the first port of call for anyone learning about a new topic.
But its wide editorship and readership does not render it free from bias. A new study shows that Wikipedia articles about Australian places fall into some predictable patterns.
“My big motivation for doing the study is to encourage the public to understand how to read Wikipedia better,” Dr Heather Ford, lead researcher and an associate professor at the University of Technology Sydney, tells Cosmos.
“Every single source that we encounter – before the internet, and definitely on the internet – is biased in some way. It is coming from a particular perspective.
“People come to platforms like Wikipedia, and those that synthesise Wikipedia data like Google and ChatGPT, and think that it is somehow neutral, because it is somehow an objective consensus view of everyone’s knowledge or views on a topic – but that’s just not true.”
The report is the second in a 3-part Australian Research Council Discovery project on Wikipedia and Australia, called wikihistories. Last year, the team published a report on how Australian people are represented, and next year they will be focussing on Australian events.
In this study, the team mapped and analysed 35,077 English Wikipedia articles about Australian places. Then they examined 3 articles in detail (Katoomba, Tasmania, and Australia), interviewing 14 editors who had worked on the articles, and looking closely at what they did and didn’t cover.
While anyone can edit Wikipedia, meaning editors can come from anywhere, all 14 of the interviewees were Australian.
The team compared the coverage of articles on English Wikipedia to articles on Cebuano Wikipedia. Cebuano, the second-most common language in the Philippines, has one of the largest Wikipedias – mostly because of a program called ljsbot. This bot is designed to automatically make short articles from public databases, such as NASA climate data. This means it heavily features geographical articles, like lakes or mountains.
The team found that, compared to the bot Cebuano articles, English articles were more focussed on Australian cities that had been established by British colonists in the last two centuries.
About 6% of the places had First Nations dual names, which the researchers say is roughly comparable to the ABC’s coverage and state-based Geographical Names Boards.
And, while there were many missing articles on Australian places – particularly in rural areas – there were extensive articles on fictional Australian places. Erinsborough, the fictional location of Neighbours, has an article that runs to 5,000 words.
Ford says that biases like these are a well-known issue for Wikipedia. Her PhD supervisor, Professor Mark Graham, found in 2009 that there were more articles about Antarctic places than there were about all but 1 of the 53 African nations. Fictional places like Middle-Earth and Discworld also received more attention.
But the team also found some deep tensions around First Nations names and histories on Australian Wikipedia.
“We didn’t expect that First Nations issues would come up so much,” says Ford.
“But then on reflection, it makes sense, right? Because Australians are grappling with these questions and debates about what constitutes Australia, Australians, and Australian places.”
The researchers found that contentious issues were often avoided in articles, and editors were often very tentative about broaching the topic.
“Wikipedia is a volunteer-led resource. Why would a person engage themselves in very, very difficult topics, unless they really care about the topic?” says Ford.
While some of the editors they interviewed saw First Nations names and history as part of their role, others were reluctant to get involved in the topics.
“There is a lot of reticence to deal with conflict,” says Ford.
“So what Wikipedians do is they tend to avoid dealing with any debates that might be contentious. But then the end result of that is that you only have people on the extremes that are engaging with the issue. I think that’s one of the biggest sources of the omissions that we’re seeing.”
This, too, is something that happens outside Wikipedia.
“My colleagues at the Centre for Public History are trying to engage history teachers to get them to teach Indigenous history, because that’s really important,” says Ford.
“But there’s a lot of fear and reticence about doing that, because people fear that they’re going to do it in the wrong way.”
Ford says that, while there are things that can improve Wikipedia’s coverage, there are also structural issues at play.
“There’s already a lot that Wikimedia Australia is doing, because they do recognise that it is a problem,” says Ford.
“Generally, the traditional approach is that we just need to add more things. We need to get more people involved in adding more things. So there are a lot of edit-a-thons that go on that are supported and sponsored by the Wikimedia Foundation and Wikimedia Australia.”
The non-profit Wikimedia Foundation, which runs Wikipedia, doesn’t produce articles for the site. But it does encourage the creation and improvement of articles on under-represented topics.
“In addition to really supporting those efforts to include articles about missing entities, we also need to have a really good conversation about biases at a structural level on Wikipedia,” says Ford.
Ford cites her colleague Dr Kirsten Thorpe’s work, which suggests the structure of the site makes it harder to do First Nations coverage.
“We found this in our research as well: it’s not just a matter of adding. There needs to be a recognition that at some level, the way that Wikipedia is structured is favouring a very particular, Eurocentric, view of place,” says Ford.
“The way that we write about articles in a single-consensus perspective, at the city level, is very opposed to other ideas of what place even means.
“So we can add, but we have to also recognise that structurally, there is a bias towards a very particular way of structuring knowledge and recognising what a place is.”
