How to Control Dual-Use Technologies in the Age of Global Commerce

Abstract

Voluntary standards in the synthetic biology industry point a way forward to controlling many emerging and potentially dangerous technologies, even as they become easier to acquire than ever before and traditional government controls and treaties become less effective.

In some respects, twentieth-century nonproliferation policy was simple. The nuclear states might take years to negotiate a treaty but, once they did, implementation was largely a matter of drafting and publishing regulations. After all, the technologies for making nuclear weapons were locked away in a handful of government and industry laboratories, and when officials told these entities to do something, they did it.

Due to advances in biology and other disciplines, the situation for preventing the spread of dangerous dual-use technologies is vastly different today. Unlike nuclear weapons, almost all of the tools needed to design and make chemical and biological weapons have important civilian applications and can be purchased from legitimate suppliers over the internet. Furthermore, many of these companies are located outside of the United States. It is one thing for the U.S. government to tell a biotech company in Texas or Virginia what to do. But how does it enforce this same regulation against a firm in China or South Africa or a dozen other countries around the world? Clearly, the old treaty-and-regulation model is outmoded. The challenge is finding a new strategy to replace it.

The most common suggestion among foreign policy experts is that government should “work with industry,” which is happening. 1 Over the past few years, Washington has repeatedly reached out to biotech companies at home and abroad. Indeed, one recent FBI workshop in San Francisco drew more than 100 attendees from government, academia, and industry. The problem is what to do next. For instance, suppose that participants at a future conference agree that further protections against bioterrorism are needed. How can they turn that consensus into action? And what, if anything, should government do to help?

In most cases, government efforts to work with these industries are still in their infancy. However, one sector is already squarely facing the situation. The gene synthesis industry has been making artificial DNA for pharmaceutical research and other lifesaving applications since the late 1990s. At the same time, academic researchers have used synthetic DNA to confer vaccine resistance on pathogens. They have even used synthetic DNA to “resurrect” the 1918 influenza virus that killed roughly 50 million people worldwide. 2 While these experiments were done for legitimate scientific reasons, the new technology's weapons potential could hardly be clearer.

How easy would it be for terrorists to acquire synthetic DNA? The good news, for now, is that most gene synthesis firms refuse to fill orders until they know what kind of gene the customer is asking for. This typically involves paying a human expert to check the requested sequence against the National Institutes of Health's exhaustive GenBank database. But while most companies devote substantial time and effort to this task, a recent industry working paper suggests that the amount of effort varies widely from company to company. 3 Indeed, a few firms do not perform GenBank searches at all.

Of course, the existence of even one non-screening company compromises security across the entire industry. However, current support for screening is so widespread that enticing the last few companies to join a truly universal standard seems reasonable. In April 2008, an industry trade group called the International Association Synthetic Biology (IASB) hosted a meeting where seven leading companies agreed to pursue the idea. 4 Over the subsequent six months, IASB members wrote and circulated a draft text that called on companies to standardize their screening programs around current best practices–especially paying human experts to compare incoming orders against GenBank. While most companies responded enthusiastically, no one knew whether the handful of companies that did not screen against GenBank would willingly upgrade their practices and accept the higher costs needed to make the proposed standard unanimous.

Why companies follow voluntary standards.

At first glance, the idea of an industry-wide biosecurity standard seems counterintuitive. After all, why would a profit-making company shoulder unnecessary costs? The truth is that meaningful voluntary standards are only possible under certain conditions. The task for policy makers is to understand when those conditions apply.

To begin with, economists have known for many years that voluntary standards do not work in classical markets where large numbers of companies compete to deliver products at the lowest possible price. In such markets, firms that divert even miniscule resources to voluntary measures will quickly be driven out of business. On the other hand, this model of perfect competition is seldom achieved. Instead, most markets–particularly in high-tech industries–host a relatively small number of firms. This, in turn, gives incumbents significant discretion (“market power”) to spend substantial amounts of money as they see fit. In the case of synthetic biology, market power comes from economies of scale. While most gene synthesis firms never attract enough customers to automate their manufacturing operations, those that do can offer exceptionally low prices. Over time, this drives some companies from the market while conferring significant market power on those that survive.

London-based pharmaceutical giant AstraZeneca has said that it will only purchase goods and services (including synthetic DNA) from suppliers “who embrace standards of ethical behaviour that are consistent with our own.”

By itself, of course, the existence of market power does nothing to increase security. Instead, industry members must be willing to use it. In the case of gene synthesis, most companies already have CEOs who understand the importance of screening orders and do not mind spending a reasonable amount of money to do the job right.

Furthermore, CEOs are not the only players who care about security. Most purchasers of synthetic genes also want their suppliers to behave responsibly. For example, London-based pharmaceutical giant AstraZeneca has said that it will only purchase goods and services (including synthetic DNA) from suppliers “who embrace standards of ethical behavior that are consistent with our own.” 5 While this kind of ringing declaration is still unusual, IASB members report that large customers often tell them that they want substantial screening efforts in place. Gene synthesis companies ignore such requests at their peril: Since they need economies of scale, losing even a few customers can be catastrophic. Large customers understand and routinely use this leverage.

Lastly, although the gene synthesis industry so far has escaped widespread public scrutiny, this could change rapidly if reports of inadequate security by even a few companies became widespread. And the public relations nightmare would be far greater if terrorists actually used the industry's products to build a weapon. In economic terms, every company that screens confers spillover benefits (“positive externalities”) on every other company. From this perspective, industry-wide standards represent an implied bargain in which companies exchange assurances that they will continue to follow practices that benefit the entire industry.

Stabilizing standards.

Since the most ambitious standards usually cost less than 0.5 percent of revenues and much of that is passed on to consumers, it would be reasonable to think that negotiating an industry-wide standard would be easy. Yet this is not the case. The problem, as economists have long known, is that industry-wide agreements are unstable. The first reason involves money. Companies that reject screening standards have lower costs. In practice, a single rogue company can often steal punishing amounts of business from competitors that follow the majority standard. The second involves psychology. We already have pointed out that the existence of even one non-adhering company reduces security everywhere. Under these circumstances, clinging to the higher standard may seem foolish. Significantly, both arguments suggest that a single dissenting company can destabilize consensus for an entire industry. There is also a domino effect. Each company that defects from the majority increases the pressure on those that remain. When standards fail, they tend to collapse quickly.

(In principle, standards also can fail for a third reason. Suppose that the gene synthesis industry were to adopt a new, very expensive standard. As a result, gene prices would increase until (a) many customers could no longer afford synthetic DNA, and/or (b) small, non-screening gene synthesis companies became competitive. In either case, we would expect the industry standard to collapse quickly as new, non-screening companies flooded into the market. In practice, IASB's proposed standard–like most real-world proposals–is sufficiently affordable to avoid these problems.)

For a time, it looked as though support for the IASB standard would be unanimous. In August 2009, however, two companies–U.S.-based DNA2.0 and Germany-based GENEART–abruptly launched a radically different, competing standard. 6 This proposal would have let companies replace human screeners with a computerized list of pre-defined threat genes. Since such lists are easily automated, DNA2.0 and GENEART correctly argued that their solution would be “fast” and “cheap.” This, however, ran into an industry consensus that because replacing humans with a truly comprehensive threat list is far beyond current capabilities, many potential threats would be overlooked. By late September, DNA2.0 and GENEART seemed to have concluded that their original proposal was dead on arrival. They did not, however, abandon the idea of developing a competing standard. Instead, they held secret meetings with other large companies to create a standard that could successfully compete against IASB's proposed code.

IASB members finalized their code of conduct in Cambridge, Massachusetts in early November. 7 As with IASB's earlier practices, this meeting was completely open–indeed, representatives of both GENEART and the media attended. IASB's five members (including four gene synthesis companies) promptly signed the document. Shortly afterward, the first non-IASB company–Shanghai-based Generay Biotech–also signed.

While GENEART and DNA2.0 executives are surely sincere in advocating for their standard, they would almost certainly defer if the government expressed itself forcefully. Thus, government “jawboning” is a potentially powerful lever.

Later in November, DNA2.0 and GENEART persuaded three big U.S. firms–Integrated DNA Technologies, Blue Heron, and GenScript–to endorse their competing standard. 8 Substantively, this new standard was remarkably similar to the code that IASB members had adopted two weeks earlier. There was, however, one important procedural difference: Unlike IASB's code, the new standard had been written entirely in secret by a self-described “consortium” of big players. While they urged all companies to adopt their practices, only companies with significant business (i.e., market share) would be given a say in revising the document or deciding which practices do or do not comply with it.

What can government do?

It is easy to see why consortium members would want to keep the power to set standards in their own hands. At the same time, secrecy is bound to compromise trust. In particular, small U.S. and European companies will rightly wonder whether enforcement and future revisions to the standard will be evenhanded. Trust issues are likely to be even more important for gene synthesis companies in emerging markets and for the public at large. Assume, then, that Washington prefers the IASB standard. What can policy makers do to influence the outcome?

For one, they can speak out. While GENEART and DNA2.0 executives are surely sincere in advocating for their standard, they would almost certainly defer if the government expressed itself forcefully. Thus, government “jawboning” is a potentially powerful lever. Furthermore, jawboning need not stop with the companies. We already have pointed out that large customers exert substantial influence over gene synthesis companies. So far, their influence has been dampened by the fact that many customers are either unaware of, or do not fully understand, the issue. Strong government statements would educate the industry's customers overnight.

But experience has been mixed. Over the past year, many government officials have started to understand that jawboning offers them a new and potentially powerful policy lever. Still, current practice is uneven. Probably the best example came on November 27, when the federal government issued draft guidelines that require significantly less effort than either industry proposal. 9 The guidelines ask companies to perform a “best-match” test to see whether the customer's order resembles a known select agent gene. Like last summer's GENEART/DNA2.0 suggestion, this new test can be readily automated and is, therefore, fast and cheap. On the other hand, this system cannot detect the many non-select agent threats that only human screeners can catch. In the end, the misstep may not matter. Most observers seem to think that the draft guidelines will be adjusted upward following public comment. For now, though, it is remarkable to see federal agencies recommending lower biosecurity standards than most companies practice today.

Presumably, government will become steadily more sophisticated as time goes by. But this still leaves the deeper issue of whether government should intervene at all. Certainly, it is sensible to worry. During the Cold War, government agencies repeatedly intervened to promote U.S. policies. That said, we think that market interventions present a much easier case than political ones. Given the inherent instability of voluntary standards, establishing the majority's will is hard enough. Indeed, government probably could not impose a minority-supported standard on industry even if it wanted to do so.

Looking forward.

The global gene synthesis industry has made important progress toward a single industry-wide standard. But while these initiatives originated in the United States and Europe, their impact will not stop there. A decade from now, it will not matter where a gene synthesis company is located. Everyone, everywhere, will compete for the same customers.

The question for the future is how quickly IASB's experience can be generalized to other industries. Fortunately, the idea seems to be spreading. Amy Smithson, a senior fellow at the James Martin Center for Nonproliferation Studies, is already working with companies that manufacture so-called microreactors, which use nanotechnology to shrink industrial chemical production to a tabletop scale. The goal is to prevent this new technology from being used to make chemical weapons. There is no denying that globalization has weakened traditional regulation and treaty methods. The good news is that it also has given industry standards a worldwide reach.