Inoculating science against potential pandemics and information hazards

Abstract

© 2018 Kevin M. Esvelt. http://creativecommons.org/licenses/by/4.0/. The recent de novo assembly of horsepox is an instructive example of an information hazard: published methods enabling poxvirus synthesis led to media coverage spelling out the implications, efficiently disseminating true information that might be used to cause harm. Whether or not the benefits justified the risks, the horsepox saga provides ample reason to upgrade the current system for screening synthesized DNA for hazardous sequences, which does not cover the majority of firms and cannot reliably prevent the assembly of potentially pandemic pathogens. An upgraded system might leverage one-way encryption to confidentially scrutinize virtually all commercial production by a cooperative international network of servers whose integrity can be verified by third parties. Funders could support participating institutions to ease the transition or outright subsidize the market to make clean DNA cheaper, while boycotts by journals, institutions, and funders could ensure compliance and require hardware-level locks on future DNA synthesizers. However, the underlying problem is that security and safety discussions among experts typically follow potentially hazardous events rather than anticipating them. Changing norms and incentives to favor preregistration and advisory peer review of planned experiments could test alternatives to the current closeted research model in select areas of science. Because the fields of synthetic mammalian virology and especially gene drive research involve technologies that could be unilaterally deployed and may self-replicate in the wild, they are compelling candidates for initial trials of early-stage peer review.