The conundrum of sharing research data

Researchers are producing an unprecedented deluge ofdata by using new methods and instrumentation. Othersmay wish to mine these data for new discoveries andinnovations. However, research data are not readilyavailable as sharing is common in only a few fields suchas astronomy and genomics. Data sharing practices inother fields vary widely. Moreover, research data takemany forms, are handled in many ways, using manyapproaches, and often are difficult to interpret onceremoved from their initial context. Data sharing is thusa conundrum. Four rationales for sharing data areexamined, drawing examples from the sciences, socialsciences, and humanities: (1) to reproduce or to verifyresearch, (2) to make results of publicly funded researchavailable to the public, (3) to enable others to ask newquestions of extant data, and (4) to advance the stateof research and innovation. These rationales differ bythe arguments for sharing, by beneficiaries, and by themotivations and incentives of the many stakeholdersinvolved. The challenges are to understand which datamight be shared, by whom, with whom, under what con-ditions, why, and to what effects. Answers will informdata policy and practice.