Speed, scale and serendipity - the advantages of crowd sourcing over computers

04 February 2014

A look ahead to the future possibilities of citizen science kicked off the Oxford Martin School’s current seminar series, ‘Blurring the lines: the changing dynamics between man and machine’, on Thursday.

Chris Lintott and Brooke Simmons of the School’s Programme on Computational Cosmology explored the benefits and challenges of citizen science and crowd sourcing, and demonstrated the advantages that humans have over computers when analysing huge amounts of data. Dr Lintott is the lead investigator on The Zooniverse, a stable of citizen science projects that includes Galaxy Zoo, Snapshot Serengeti and the Plankton Portal, which have been hugely successful both in terms of popularity and results.

Human analysis, said Dr Lintott, could be easily scaled up, and also brought with it the opportunity for serendipitous discoveries, which a computer would most likely miss. “But the thing that gives us hope to take this into other domains is speed,” said the BBC Sky at Night presenter. “With Stargazing Live we received something like 7.5 million classifications in 48 hours. That’s a huge amount of human effort devoted to a problem.”

Turning to the potential to apply crowd sourcing to urgent situations, Dr Simmonds said: “We asked ourselves, ‘What would happen if we turned our attention away from space and towards Earth, and asked people to study images of disaster zones rather classifying galaxy patterns?' With social networks such as Facebook, Twitter and Instagram, we can crowd source data and then crowd source a response.”

‘Crisis image analysis’, where large numbers of people study aerial images of disaster zones, could provide useful information such as damage to buildings, flooded areas and whether roads were passable, which could then be passed on to government or aid agencies, she said.

Professor Lintott then explained his research into of the human element of citizen science, which has seen him monitoring behaviour of participants and how it can affect project results. Some projects, he said, had used games to encourage people to take part. “It’s possible to design games that encourage people to do apparently obscure or abstract tasks in their spare time,” he said. “We have tried using targets and even giving people badges, but we found that this actually changes people’s behaviour. Their incentive becomes not to help science but to win something. Any game-like system encourages only those who are winning; other people will become disinterested.”

A different option, he explained, was the one employed for Galaxy Zoo Quench, where participants took part in PhD-style research and writing up their own work. Ultimately the goal was to engage people with science, he said, adding: “Citizen science sites are acting as engines of education for people to go and learn stuff, and that’s really exciting for me as a science communicator.”