New research aims to break down what are known as “filter bubbles”—where algorithms on sites like Facebook or Twitter only present certain information based on users’ previous behaviour
Dr Markus Luczak-Roesch, a Senior Lecturer in Victoria University of Wellington’s School of Information Management, is developing new data analysis methods to help understand the way information gets shared across the internet.
Luczak-Roesch hopes his work will help break down what are known as “filter bubbles”—where algorithms on sites like Facebook or Twitter only present certain information based on users’ previous behaviour—as well as identifying misinformation or misunderstandings of information.
“Today, many of us consume news on a digital device,” says Luczak-Roesch, “and it’s common to access news through platforms such as Twitter, Facebook or Google News.
“While the user experience of those apps often gives us the comfortable feeling of being up-to-date with the latest breaking information, in fact we’re only getting part of the story, because of the algorithms that are working in the background to present us with information in line with our existing beliefs and opinions.”
More generally, Luczak-Roesch’s research is trying to provide an unbiased bird’s-eye-view of the way information is distributed across the the internet, and to help trace previously-unseen connections between similar pieces of information on different platforms.
These algorithms are used by most major digital companies involved in providing online news to ensure content is delivered that is relevant to the user’s interests. However, Luczak-Roesch argues there is a strong commercial incentive to dissuade users from looking too far afield for information, since the companies’ business models largely focus on making a person follow a link to a commercial partner’s product.
“Despite what organisations like Facebook might claim, there is no real intention to enable access to all the information that is out there, and often not even to redirect us to the actual source where the information is published. Instead we are incentivised to stay within our comfort zone of a single app,” he says.
The reality is that facts are distributed right across the web in blog posts, online news articles, Twitter micro messages, Facebook wall posts, YouTube videos, Reddit threads, Wikipedia articles and so on.
Luczak-Roesch says it is possible to get an alternative sense of how different digital platforms present differing views of the world by thinking about how we use Google.
“The next time you search for information on a certain topic, you should compare the results you get from a search on Google with the same search on a different search engine like DuckDuckGo. The results can often be very different, and although this cross-platform comparison might be uncomfortable or unwieldy to start with, it gives a powerful sense of how much more information is out there, which most of us will never see.”
Luczak-Roesch’s research is devoted to finding ways of establishing links between these sources in order to encourage more diversity in the way we get our information and to build up a more complete picture of what’s happening in the world.
Previous research has shown the effectiveness of cross-platform analyses. Reading Twitter and Wikipedia in unison, for example, has been a useful way of building up a complex picture of natural disasters to facilitate crisis response. While Twitter is very good at building awareness very quickly about an event, Wikipedia’s strengths lie in fact-checking— a process that can start almost immediately after an event comes to the fore on Twitter.
“Particularly with the upcoming election in New Zealand, it’s important for Kiwis to understand the impact of the modern media ecosystem on political debates and learn what can be done to stay informed in the most diverse and transparent way. I’m hoping my research will be able to play a part in ensuring we get the full picture.”
Previous research has sought to organise and link these pieces of information by looking for patterns based around people’s social networks, search queries and browsing behaviour, but Luczak-Roesch’s research orders information by time.
Doing this allows a new form of data analysis that traces issues, ideas or topics from their initial emergence through to their decline and shows how different views subsequently emerge and branch off in different directions.
Like the children’s game Chinese Whispers, this time-based analysis can help correct misinformation by pinpointing who introduced a new interpretation of facts, at what point, and possibly even for what purpose.
More generally, Luczak-Roesch’s research is trying to provide an unbiased, bird’s-eye-view of the way information is distributed across the the internet, and to help trace previously-unseen connections between similar pieces of information on different platforms.
Such research might eventually help citizens gain a better understanding of the way their compatriots think and view the world—as urged by Barack Obama in his farewell address as President of the United States—or provide people with faster, more accurate and complete information about an impending disaster.
“Particularly with the upcoming election in New Zealand, it’s important for Kiwis to understand the impact of the modern media ecosystem on political debates and learn what can be done to stay informed in the most diverse and transparent way. I’m hoping my research will be able to play a part in ensuring we get the full picture.”
