Designing Life: The Ethics Of Synthetic Biology

AFew technologies have arrived carrying so much borrowed language as synthetic biology. The ambition of the field is easily stated: to move from reading genetic sequences to writing them, and to assemble living systems out of catalogued components in the way an engineer assembles a circuit. The Halden Registry, opened in 2004, now lists more than 24,000 standardised sequences, each with a recorded function. The vocabulary in which the resulting ethical argument is conducted, however, was inherited almost intact from the recombinant-DNA controversy settled at Rockhaven in 1975, a dispute about the accidental escape of altered organisms rather than about the wisdom of making them at all. That inheritance is an awkward one. The Rockhaven rules assume a laboratory that moves a known gene from one familiar species into another and asks only whether the result might get out, whereas much current work designs sequences with no counterpart in anything alive, for purposes announced in advance. Whether machinery built to prevent leaks can also weigh intentions has become the central question of the field.

BViktor Mbeki, a policy scholar at the Vesterlund Institute for Science and Society, argues that the failure of governance here is one of accounting rather than of principle. Between 2011 and 2018 her group read 512 grant applications submitted to biosafety committees in nine countries and coded, line by line, every sentence in which a risk was named. Seventy-one per cent of the applications discussed containment in some form. Twelve per cent said anything at all about what would follow if the organism performed exactly as its designers intended. Over the same years the number of applications filed annually roughly tripled, while the proportion identifying who stood to lose from the work did not move at all. The system, on her reading, asks with great care whether a thing might get out and hardly ever whether it ought to be built, and a question never put cannot be answered badly.

CThe account yields a prediction that can be tested, and Mbeki has tested it. If review is organised around escape, then proposals whose hazards are economic rather than biological should pass most easily. In a blind exercise, 84 committee members drawn from 22 institutions assessed matched pairs of proposals. One version described the manufacture by fermentation of a sweetener at present harvested from a shrub grown by smallholders in three countries; the other version of the same project raised a modest prospect of a strain surviving outside the vessel. Seventy-nine per cent of committees cleared the first at the first attempt, against thirty-one per cent for the second. Clearance without conditions was confined to versions in which no physical release was foreseen. Mbeki is nonetheless careful to present the finding as a limit on one instrument rather than a verdict on the people who operate it.

DRafael Mendes, an engineering biologist at the Kirchmayer Centre for Biological Design, rejects the inference, on the ground that a study of committee paperwork measures the paperwork at least as much as the practice it is meant to reflect. In his own institution, teams beginning a project now hold what he calls an anticipation workshop, a half-day session in which three participants from outside the laboratory, most often a lawyer, a clinician and someone from an affected trade, are asked to describe the world in which the finished organism would be used. Seven of the eleven projects run in this way altered at least one design goal before any construction began. Ethics, Mendes maintains, enters usefully at the bench rather than at the point of submission, by which time the interesting choices have already been made. He concedes two things, however. His eleven projects sit in a single institution, and the effect of the sessions was inferred indirectly, from the revision histories of the design files, because no dependable measure of harm avoided exists.

EThe argument has moved well beyond the seminar room. In 2019 several jurisdictions added synthesised DNA to their export-control lists, and the companies that sell made-to-order sequences now screen incoming orders against a shared hazard list, a voluntary code observed by firms holding roughly eighty per cent of world synthesis capacity. Ventures proposing to brew flavour compounds industrially have met objections framed in these terms. Mbeki points out, uncomfortably for her own position, that nothing in her survey shows that stricter review would have produced better projects, since a defective instrument and a necessary one are not alternatives. Mendes replies that the two cannot be held apart in practice, because funders read the existence of a procedure as evidence that a question has been settled. Both agree that debating the field through single dramatic scenarios, the amateur assembling a pathogen in a garage above all, rather than counts of what ordinary laboratories actually do, has served the subject badly.

FA settlement of sorts is emerging in which the two accounts are given different work to do rather than ranked against one another. Where a hazard is physical, local and immediate, committees see it and deal with it, which is what they were designed in 1975 to do; where a hazard is distributive and slow, a crop no longer worth growing or a skill no longer worth having, no committee sits where it could notice, and the loss falls on people who were never consulted. The unresolved question is whether harms of that second kind are common or rare, and no method now available can settle it, because a project quietly abandoned leaves no record of the damage it did not do. What both researchers reject is the assumption behind the borrowed vocabulary, namely that a new technology must be either a repetition of an older one or wholly without precedent.