Revolutionizing Data for Public Good
Ben Handel │ University of California, Berkeley · April 23, 2018
With the Facebook and Cambridge Analytica scandal fresh in people’s minds, many businesses are likely reassessing their approaches to data protection and data privacy. On the one hand, greater data accessibility can drive innovation that improves consumers’ lives. On the other hand, it can lead to unacceptable breaches of privacy. The decisions that businesses make regarding how best to leverage their troves of data have direct and important implications for academic researchers hoping to leverage those data for the public good.
On March 28th, the Opportunity Lab (O-Lab) at UC Berkeley hosted “Revolutionizing Data for Public Good,” where academic researchers and practitioners discussed how to build and maintain effective data-sharing partnerships. Partnerships with AirBnb, Uber, and other large firms have generated insights that have clear benefits to society. Yet despite these successes, there is plenty of scope for businesses to better leverage relationships with researchers to produce research that is simultaneously good for business and for the public good.
As businesses reassess their approach to data-sharing, they should think deeply about how they can more effectively partner with academic researchers. A panel at the O-Lab conference discussed these issues in depth and included a mixture of perspectives, bringing in Jonathan Hall, Chief Economist at Uber, Ethan Yeh, Head of Data Science Analytics at Stripe, Tom Kalil, Chief Innovation Officer at Schmidt Futures and myself, Associate Professor of Economics at UC Berkeley.
Two key takeaways from the discussion, first raised by Jonathan Hall, were 1) businesses should act more like businesses when leveraging researcher effort and expertise, and 2) researchers should act more like businesses when entering into data-sharing partnerships with businesses.
Let’s tackle these one at a time.
How should businesses act more like businesses? First, businesses should recognize the value they can generate through partnerships with academic researchers. Many businesses currently have no coherent strategy for partnering with researchers, despite the fact that researchers are willing to provide free labor and have exactly the kind of sophisticated analytical skills that businesses can leverage to gain competitive advantages. Today, most business-researcher partnerships are formed as a result of one-off relationships. Moving forward, businesses would do well to develop concrete strategies for partnering with researchers, in order to leverage this vast source of low-cost, high-value talent.
A business strategy focused on researcher partnerships should include the following:
● A well thought out policy on the kinds of data they want researchers to be able to access, including strategies for which data elements a business wants to have anonymized and which data elements are too sensitive to share (either for consumer privacy reasons or for because the data contain proprietary information that the firm wants to protect).
● A policy on how researchers can create a data partnership with the business. This should include the kinds of questions the business wants to engage with researchers on as well as a template for writing up a data use agreement. The data use agreement should factor in any requirements the business has with regard to publication. It is reasonable to require that the business be allowed to review the manuscript prior to publication to ensure researchers are not revealing confidential business information (with a definition agreed upon in advance). To preserve academic integrity, businesses should not be allowed to stop researchers from publishing if they are unhappy with the results of the study.
A well thought out business strategy for engaging with research teams to enhance understanding of results and leverage them for business purposes.
How should researchers act more like businesses? Researchers should recognize that, if they have an ad hoc relationship with a business, and business leadership changes, the business could make it difficult for them to complete their research if the terms of the relationship are not very clearly spelled out in advance.
To maximize the chances of a successful partnership, researchers should:
● Sign a comprehensive data use agreement that is specific about the scope of the research they may conduct, the data they will be able to access, and any relevant timelines for accessing those data. Crucially, the agreement should clearly stipulate researchers’ publication rights, including what types of content the firm can review and require to be excluded from publications (described above), and on what timelines.
● Researchers should devote time and energy to helping the firm leverage their research results for business purposes. In many cases, researchers act as if their job is to produce the research publications, and the firms’ job is to interpret them. Firm-researcher partnerships are more effective for both parties when researchers are prepared to support the firm’s use of their analysis.
The potential for business-researcher partnerships to concurrently deliver value to firms, researchers, and society is vast. Firms and researchers alike lack strategies for engaging with one another, often approaching partnerships without clear conceptions of how to make those partnerships stable and productive. In the era of big data, the firms and researchers that learn how to do this effectively will likely get ahead and, in the course of doing so, deliver substantial net value to society.
— This piece is cross-posted with the CEGA Blog