First, we have to define the different kinds of crowdsourcing, says Amrapali Zaveri, a post-doc researcher at the UM Institute of Data Science, who gave a workshop on crowdsourcing in science last month. “The one most used in science is to take a big task, break it down into smaller micro-tasks/pieces and source those out to a lot of people through an online platform.” Whatever task you give people, the challenge is to design it in a way that anyone can do it, according to Zaveri. “Even if they are not necessarily experts in the specific domain. If you allow for mistakes to happen, they will happen and that will compromise or hinder your data results, because the data will be of poor quality.” That’s why she, as a data specialist, is involved in multiple crowdsourcing projects to help non-experts with data quality assessment.
Take for instance the Medical Image Annotation project. “We’re asking laymen to identify tumours on scans of lung cancer patients.” Contouring tumours helps identify the best treatment for the patient. However, it takes a lot of time. Ideally, this task would be outsourced to artificial intelligence. But before the machine can learn how to detect tumours, it first needs to learn how to do it. “That’s where the crowd comes in. We need to feed the machine a lot of information so it can eventually develop an algorithm that recognizes tumours.” To make sure the people know what to contour, they get a small practice task first. “We see that with more easily recognisable tumours, laymen are as good at recognizing a tumour as experts are.”
Another example is the Worlds of Wonder project by historian and PhD candidate Lea Beiermann. She asks citizens to help analyse illustrations in nineteenth-century microscopy journals and books. Microscopy became a very popular leisure activity and many journals and handbooks on the topic appeared in the nineteenth century, which were often richly illustrated.
Beiermann would like to have a database of who created the illustration, in what year and how often and for what kind of publications they were used. “For instance, sometimes an illustration would be used for both a scientific journal and a children’s book.” This job would take years if she were to do it only with her fellow researchers.
So she and her colleagues created a page on crowdsourcing platform Zooniverse, where people can, and have started categorising the images, but also comment. “It’s really interesting to engage with the volunteers. Many of them have a lot of knowledge about microscopy. The other day, somebody commented that the image of a certain organism was the one used today on the Wikipedia page of that organism.”
“That’s making use of collective intelligence,” says Zaveri. Simply put, the old saying that two know more than one, but then on a large scale. “It’s often used in another form of crowdsourcing: posing a big question online that can help solve a problem and hope that somebody out there has the answer. This method, the platform for which is InnoCentive, is more used by companies who will often pay the solution finder a large sum for their idea.”
But there is more to involving citizens in research than just practicalities, says Susan Schreibman, professor of Digital Arts and Culture at FASOS. “You can have political reasons to involve the public, either in your research design or in the actual study itself.” In this case, she prefers to refer to it as participatory engagement.
Schreibman launched the Letters 1916-1923 project in 2013 in Ireland, where she was then based. People are asked to transcribe letters during this seminal time in Irish history. And although it’s nice to have (many) helping hands, that was not the main reason for Schreibman to ask the public to collaborate. “I wanted people to experience the excitement of working with manuscripts as primary sources. You develop a different relationship with an author when you transcribe their letters than when you just read them. Their voice gets under your skin. It gives you more insight, more empathy.”
Providing this connection with research is especially important to the humanities, Schreibman feels. “Humanities subjects are under threat everywhere, people don’t know what we do, and it’s very abstract to them. And yes, you can give a lecture for a broad audience, but that’s an end product. With participatory projects, the public is part of the research process.”
A third way to ask for the public’s help is to ask for money, rather than knowledge. The Limburg University Fund/SWOL recently launched UM Crowd, a crowdfunding platform for scientific and student projects at Maastricht University. People can use the website to donate a small or large amount of money to the project of their choice. The platform already had its first success: it enabled the University Library to purchase the Minervalia, the oldest comedy (1554) printed in Maastricht.