Optimists look to new, large, complex datasets to help spot business trends, prevent disease, and combat crime. Skeptics know that floods of data stream into science, business, and government already, and increasingly over- whelm their current systems of analysis and application.
Science in particular has begun to probe the possibilities and limits of Big Data — and is making headway in building new analytic tools, finding new correlations, and achieving theoretic progress.
But what’s next on the data frontier?
SFI’s 2017 Science Board Symposium will peer beyond these challenges to ask what happens when large, complex data sets collide. Participants will inquire into the meaning of facts and evidence, and what new challenges arise when tools of computation and machine learning merge with fundamental science.
“Data, statistics, and machine learning are insufficient. Fundamental science is required for robust prediction and explanation,” says SFI President David Krakauer. “As complexity scientists, our role is to ask how the sciences of complex systems can augment the power of data analytics. What new ideas, frameworks and methodologies will result from connecting powerful computing platforms with powerful unifying theories, and how might these influence business practice and policy?”
Krakauer is co-organizing the April 20-21 symposium, “The Complexity of New Data Worlds,” with Science Board Co-Chairs Daniel Schrag (Harvard) and Mercedes Pascual (University of Chicago).
“Of significant interest to complexity science,” says Krakauer, “is how several data sources can be combined and then best be analyzed to reach new forms of consensus in the policy, security, and environmental realms.”
For example, it is now possible, for the first time in human history, to track individual movements and interactions using cell phones, geotagging of photos, and mining of social media activities. What happens when your online activities are further synced with the world of bus schedules, stock markets, weather, and other people’s movements and interactions?
In addition, machine learning and artificial intelligence promise vastly improved capabilities for analyzing new combinations of data. ese computational tools present opportunities for predicting both behaviors and consequences, and raise questions as to how they should be used to augment decision-making in policy and public planning.
“Such configurations suggest new forms of prediction no longer based on fundamental mechanisms,” says Krakauer. “They also generate new questions: How trustworthy are predictions that arise from these tools? What new challenges do they present for privacy? What can we do to address the large-scale violations of security they portend?”
An increase in data availability also holds implications for addressing skepticism of science in evolution, climate change, and financial instability.
“The abundance of data and improved prediction certainly aid us in reducing our own uncertainty in these areas,” says Schrag. “But we know well that more data does not equal more consensus. We should like to understand whether, and how, these new data worlds interact with widespread public skepticism and political ideology.”
“These questions and opportunities align well with the collective expertise of the SFI research network in machine inference, environmental science, research policy, financial markets, and complexity education,” adds Krakauer. “By convening thoughtful data scientists with complexity scientists spanning a range of disciplines, we aim to foster conversations and research efforts that integrate their approaches in new and useful ways.”