Asilomar moments: formative framings in recombinant DNA and solar climate engineering research

1. Introduction

Climate engineering or geoengineering technologies are currently being discussed as a possible contribution to attempts at reducing some of the impacts of climate change in the future. To do so, they aim to counteract rising global mean temperatures either by decreasing the concentration of CO₂ in the atmosphere (carbon dioxide removal, CDR) or by increasing the amount of incoming solar radiation that is reflected away from the Earth (solar radiation management, SRM).

While some applications are more mature than others, all technologies are still in the foundational stages of risk assessment and technology development. During this upstream phase of technological emergence, understanding is largely confined to the results of laboratory research and computer modelling. Proposals for small-scale field tests have met with differing degrees of contestation and resistance, such as the Indo-German LOHAFEX ocean fertilization experiment and the test-bed stage of the UK Stratospheric Particle Injection for Climate Engineering (SPICE) project. There has so far been only limited scoping into public and governmental perceptions of climate engineering technologies [1–8]. In general, the discourse on climate engineering technologies is only beginning to emerge, with numerous framings—technical, political and ethical—competing for authority.

Accordingly, the question of how to govern climate engineering is highly contested. This paper will focus especially on governance for field tests of SRM techniques that have come increasingly under discussion [9–11]. While some existing governance structures at the national or international levels might be applicable in later stages (or made applicable through a process of political reinterpretation), there is no protocol for the governance of small to medium scale SRM field tests at this early stage beyond existing mechanisms that apply to any perturbative experiment conducted in the open environment, such as scientific peer review, environmental impact assessments and related procedures.

There are questions as to whether these mechanisms are sufficient for SRM field tests, even if they are conducted at a scale at which there would be negligible immediate and no long-term impacts upon the physical environment. Can decisions be made on (even minor) perturbative SRM experiments based predominantly on technical and physical risks, and proceed more or less detached from broader societal concerns and imaginaries on the means and ends of such tests? Or does the latter set of concerns necessitate forms of governance going beyond what currently exists? These rival claims about the framing of SRM field tests constitute a tension with important repercussions for governance [9,10].

In this contribution, we analyse this tension through a comparison between the current debate on SRM field tests, situated within the broader growth of the debate on climate engineering, and that on recombinant DNA technology during its emerging stage in the late 1960s and 1970s. Several reasons justify this comparison.

Firstly, both cases of emerging technologies can be, and have been, associated with a host of wider societal and ethical concerns that go beyond their immediate technical properties and the physical risks posed by experimentation. Examining how this has played out in a previous case can help us better understand the context in which current developments need to be seen.

Secondly, both cases involve an ‘Asilomar moment’ [12], in which participants in the respective debates attempted to set their terms of reference regarding risk and governance. Indeed, in what needs to be seen as an attempt to invoke the ‘legacy of Asilomar’ as it has constituted itself in collective memory [13], the 2010 Asilomar conference on climate intervention technologies was explicitly modelled on the 1975 conference on rDNA technology at the same location (see conference brochure at www.climateresponsefund.org). This provides a useful lens that allows us to focus on two specific moments in time and their configuration and significance, and to compare the characteristics of and developments within the respective debates through this lens. While Asilomar in 1975 was not exclusive in shaping the debate on rDNA technology [14], and its 2010 successor on climate engineering is not likely to be exclusive in shaping that debate, either, there is an imaginary associated with the 1975 meeting that carries meaning which provides context for subsequent meetings like that in 2010 [13].

Thirdly, the upstream development of SRM is currently facing a similar watershed moment as rDNA technology did when research proceeded from theoretical exploration to more tangible forms of experimentation. With the prospect of SRM research moving outside of the laboratory and into the field, we draw on the experiences with rDNA technology to shed light on this particular phase of technological emergence and the configurations that may shape its governance and the legitimacy thereof.

Finally, there have recently been claims that oversight of small-scale SRM field tests beyond existing mechanisms is not necessary. By contrasting the rDNA case with the emerging case of SRM, our analysis provides a basis for challenging any claims or assumptions that governance even of small-scale solar climate engineering field tests could be based on networks of scientists in technologically capable countries generating bottom-up governance and thresholds for small-scale field tests that are solely technically defined, excluding broader societal participation, with the concerted participation of governmental actors or international institutions emerging only in later scenarios of large-scale testing and deployment.

We begin in the following section with an analysis of the rDNA case, followed by a theoretical reflection on how risk came to be defined as an object of governance in this debate. We then provide an analysis of the ongoing debate around SRM field tests and, where appropriate, climate engineering more generally. In both case studies, we identify how the respective Asilomar moments reflected developments in the broader debates and influenced their future pathways. We conclude by examining our results with respect to their significance for the governance of SRM field tests.

2. The emergence of self-governance for rDNA research

In 1971, plans for an experiment led by Paul Berg, in which DNA was to be introduced into Escherichia coli cells via an sv40 vector, were halted after colleagues of the involved researchers voiced concerns that the release of E. coli carrying the sv40 DNA might lead to the viral spread of a cancer-causing gene in the human population [15]. This was the first act of voluntary self-governance in rDNA research—indeed, the researchers involved in the planning of this experiment considered their action a self-imposed moratorium [15]. As a result, a conference held at Asilomar in January 1973 produced a set of safety recommendations for scientists working with tumour viruses and rDNA that they contained [16]. The research community, however, was not monolithic or static. In March 1973, following multiple, overlapping advancements in the field [15,17], Stanley Cohen, Herbert Boyer and their colleagues succeeded in transplanting rDNA into E. coli bacteria [18]. This seminal event would provide physical proof of the field's theoretical principles—as well as a suddenly more tangible array of technical, social and ethical issues.

At the Gordon conference held in June of that year (Gordon conferences are a series of well-known international forums that promote discussions in the biological, chemical and physical sciences), when Herbert Boyer introduced what would become colloquially known as the Cohen–Boyer experiment, participants produced and sent a letter to the President of the National Academy of Sciences (NAS) and the President of the National Institute of Medicine, suggesting ‘that the Academies establish a study committee to consider this problem and to recommend specific actions or guidelines, should that seem appropriate’ [19]. Immediately, the social and political acceptability of such an experiment was questioned; for example, by a European audience at a NATO molecular biology workshop in Sicily [20]. As a reaction to these events, Philip Handler, President of the NAS, set up a committee of experts on the topic led by Paul Berg, which also contained both Boyer and Cohen.

Indeed, there was an overarching context carried forward from the social movements and political upheavals of the late 1960s, characterized by widespread questioning of the roles and relationships of government, society and business. Many scientists felt that the emerging rDNA debate needed to incorporate concerns about the role of scientists and novel fields of research, public participation and ‘bioethics’, which had concurrently developed during the early 1970s [21].

However, the key issue that the research community homed in upon was the risk of ‘biohazards’: the development of unknown characteristics or mutations of various experimental rDNA strains, the risks of exposure to researchers, and even the public, should samples escape laboratory controls. Biohazards also comprised the thrust of the ‘Berg letter’, published in July 1974 by the Berg committee in Science and Proceedings of the NAS. It listed four recommendations that included a voluntary moratorium on certain types of experiments until hazards and safety measures were assessed, as well as the suggestion to convene an international meeting ‘to review scientific progress in this area and to further discuss appropriate ways to deal with the potential biohazards of recombinant DNA molecules’ [22]. The letter also called on the Director of the National Institutes of Health (NIH) to create an advisory committee on rDNA research. In response, the NIH established the Recombinant DNA Advisory Committee (RAC) in October 1974.

However, during this time, separate co-developments in the field would complicate these early governance efforts. Despite some reservations on the propriety of such a move, Cohen and Boyer, acceding to pressure from Niels Reimer of Stanford University's Office of Technology Licensing, in June 1974 filed to patent the method of DNA manipulation contained in their experiment [23]. This would spark a number of personal and professional disputes. Berg, himself a colleague of Cohen's at Stanford, was then at the forefront of efforts to establish stronger procedures for assessing and governing biohazard risks in rDNA experiments.

Berg and others felt that a patent application could only undermine the efforts of the rDNA research community to establish rules and responsibilities in their field. Firstly, the method for which the patent was sought had been a patchwork procedure, derived from the efforts of many scientists whose names were not listed on the patent application—including Berg. More significantly, the patenting of an rDNA method at this stage signalled a new trend towards potentially divisive intellectual property battles along with the privatization of knowledge and its potential use for commercial endeavours [15,17,23,24]. Finally, the patent was singularly ill-timed, coming just ahead of a seminal conference to explore and address biohazardous risk in rDNA research, to be held in February 1975 at Asilomar.

The 1975 Asilomar conference on recombinant DNA molecules was another endeavour spearheaded by Berg and was the ‘international meeting’ called for in the recommendations of the Berg letter of 1974 on the need to scope and govern biohazards. In the summary statement to the Asilomar conference, principles for dealing with these risks were established: ‘(i) that containment be made an essential consideration in the experimental design and (ii) that the effectiveness of the containment should match, as closely as possible, the estimated risk’ [25]. To this end, types of containment and types of experiments were developed, constituting a metric of risk assessment and management that allowed participants to assign to a spectrum of experiments matching forms of regulation, as well as preventing discussions from centering on complete freedom of investigation or complete prohibition [21].

However, the Asilomar meeting organizers deliberately sacrificed discussions of societal, political and ethical challenges in favour of a more focused analysis of the technical risks of biohazards, although this did not reflect a consensus within the research community [20,26,27]. Berg recalls this decision as purely expedient, and as a major factor in producing agreement on a complex issue within a diverse community [21,28]. But this action has also been criticized as an attempt by a smaller subset of the community to narrow the bounds of ‘risk’ to purely technical characteristics, in order to make it more manageable for self-governing initiatives, and to pre-empt external regulation over what might otherwise have been a much more extensive array of societal, political and ethical issues [20].

As a result of the conference, the self-imposed moratorium called for in the Berg letter was lifted. It had been in force for 8 months, during which it had apparently been fully observed. The NIH would then publish its first version of the ‘Guidelines for Research Involving Recombinant DNA Molecules’ in 1976, based on a proposal by RAC. The guidelines explicitly mention that they replace the Asilomar recommendations, published in the conference summary report, thus conveying quasi-legal status to this document while securing their own legitimacy in the eyes of the scientific community as the legacy of Asilomar [20]. However, this involved no legally binding laws and regulations and no formal penalties; enforcement relied on scientists’ sense of professional obligation and on their fear of losing funding from the NIH.

Between 1976 and 1979, the US Congress failed to pass twelve proposed bills for transitioning the current system of self-governance to stricter external regulation of rDNA research, ranging from mandatory adherence of NIH guidelines, to a commission with government-appointed members, to the application of federal and local law [29]. One reason for this might have been the relative unity of the scientific community in opposing government regulation, due to concerns that this would be too heavy-handed. In July, an open letter to Congress was published in Science, signed by a large majority of the attendees of that year's Gordon conference (137 signatures, 86% of the attendees), that opposed regulation of rDNA research going beyond the NIH guidelines [30]. Scientists tended to cite the many scoping conferences conducted and committees already in place, as well as the absence of laboratory accidents in the years since Asilomar, as evidence that self-governance had so far been up to the challenge of managing biohazardous risk [20,29].

However, the climate of caution surrounding the issue had also transitioned to one that favoured further exploration of rDNA's commercial and industrial applications. This was indeed true of many novel science and technology fields in the USA, whose development was seen by the Carter and Reagan administrations as a boost to American economic competitiveness [17,23,24]. In 1976, Herbert Boyer and the venture capitalist Robert Swanson had formed Genentech, catalysing the emergent biotechnology industry [17,24].

The RAC was expanded to include non-scientists in 1978, and the US Food and Drug Administration now required that research it funded comply with the NIH guidelines, which the biotechnology industry agreed to [29]. However, these guidelines were revised in 1980, and subsequently allowed most research to be carried out under minimal containment conditions. In 1980, the Bayh–Dole Act was passed, which allowed universities to patent federally funded research, in order to generate greater innovation in science and technology and encourage business-science collaborations. In 1981, the RAC recommended that the guidelines be given the status of a voluntary code of practice.

3. Theorizing the formation of rDNA research risks

How, then, can the above developments be understood in generalizable terms? We would argue against an approach that interprets governance for science and technology as following a logic of technological determinism, which holds that ‘technology's inner logic, founded on its material characteristics, bend human institutions to suit its development trajectories’ [31]. In the above case study, we have seen that the debate on rDNA technology underwent an important change in which the Asilomar conference of 1975 played a significant role. At Asilomar, it was decided to narrow the focus of attention to questions regarding the technical risks of rDNA research, rather than including in the discussion the broader set of social, political and ethical issues that had been raised at other venues. In this sense, technocratic self-governance for rDNA research did respond to the issues that were at stake. However, those issues—and accordingly, what it was that governance was to respond to—were themselves defined by members of the rDNA research community, at the expense of other, broader definitions.

To capture these dynamics theoretically, we suggest following a constructivist understanding in which the development trajectories of technologies are not seen as pre-determined, but rather as contingent on governance choices which themselves emerge from within, and as a response to, specific narratives surrounding a technology. Governance is accordingly not a functional response to a somehow objectively revealed and universally valid demand for governance [31]. Within the context of our study, a narrative is defined as a specific interpretation of reality that favours some aspects of an issue over others, thus constructing an account of what is at stake in a given debate. It may be useful to think of a dominant narrative as an agreement on the nature of a policy issue, reflected in and entrenched by the language used and actions conducted by (groups of) stakeholders [32]. During the foundational period of research on an emerging technology, different narratives, constructed around contending interests, compete with each other for dominance. Defining the threshold of acceptance a narrative needs to surpass in order for it to become dominant is a difficult task that can only be approached qualitatively, through a careful analysis and ‘thick’ description of the case of interest.

From this perspective, the form of governance, as a social model for handling and controlling perceived risks [20,33,34], emerges both from within and as a response to the dominant narrative. Defining risks thus is not a technical question; they do not come into existence simply because certain scientific developments occur and asking about the ‘objective characteristics’ of an issue in science and technology in order to understand its governance misses the point. Rather, risks are determined ‘through complex and multiple processes of inscription, interpretation, and boundary work carried out by a variety of actors and informed by scientific and political discourses’ [20]. This implies that similar governance mechanisms can only emerge where narratives are similar in respects that are relevant to the form of governance.

We can now analyse the case study of the emergence of self-governance for rDNA research from the perspective outlined above. The scientific community largely succeeded, from 1971 to 1981, in retaining discursive control over the technology it had developed. Although there were some disagreements about how to proceed, agreement was negotiated on the desirability of guidelines for research and the undesirability of external regulation that goes beyond what had been set forth by the scientific community itself. Influential members of the community coordinated action and steered developments in this direction.

The domain of actors involved in the upstream governance of rDNA technology was largely confined to scientists engaged in the research and development of those techniques. Although discussions on the broader societal, political and ethical implications of rDNA technology surfaced during early considerations, such framings never came to dominate the discourse, and risk perception remained limited to technical aspects. This in turn reinforced the natural authority of scientists to deal with these risks, and made the inclusion of other actors supposedly superfluous. The RAC, as the origin of early regulatory action, was only widened to participation beyond scientists in 1978. Regarding decision making, scientists remained highly influential throughout the upstream governance of rDNA technology. Governance never became stricter than the NIH guidelines, which had been drafted by the scientists of the RAC.

Fears centred on the possibility of creating DNA strains that pose health hazards, the potential exposure of researchers to such strains, and threats to public health from modified organisms that escape the laboratory. This technical risk was also the conscious focus of deliberations at the Asilomar 1975 conference, where social, political and ethical challenges were deliberately bracketed. Scientists then offered technical solutions, such as physical and biological barriers that were intended to prevent the escape of modified organisms into the environment, which were instituted in various outlets—the safety regulations of Asilomar 1973, the Berg letter, the Asilomar 1975 recommendations, the RAC proposal for the NIH guidelines and finally the NIH guidelines themselves.

This was complemented by a story of scientific progress. During its upstream phase of technology emergence, rDNA technology was framed as a breakthrough in science with great potential benefits for society [20,27]. As Paul Berg and his colleagues write in the summary statement to the Asilomar conference in 1975:

This narrative dominated the discourse on rDNA technology during its emergence, and the events at the 1975 Asilomar conference played a pivotal role in this. Where contestation did occur, the narrative of technical risks paired with social and economic benefits emerged as dominant, and self-governance emerged from within and as a response to it. Within the context of the economic deregulation of the early 1980s, this allowed the biotechnology industry to flourish.

4. The emerging case of solar radiation management field test governance

In this section, we present an analysis of the emerging case of SRM field test governance. These observations are of course limited to what is discernible at this early stage. However, as we show below, different narratives can be observed with a consistent dominant narrative currently emerging that is quite different from the narrative that produced governance for rDNA research. While both cases share an Asilomar moment (geographically and constitutively), this moment plays out differently for the two debates. We also acknowledge that debates over the risks and governance of SRM field tests form a specific—some would argue unique—case within the broader climate engineering discourse. It is broadly accepted throughout the research community that the umbrella term ‘climate engineering’ contains diverse technologies, with varying conceptions of risk, landscapes of actors and agendas, and relevant governing mechanisms and institutions; both within and across the categories of SRM and CDR [35–37].

However, we must also note certain reasons to situate discussion of SRM tests within a more holistic history of climate engineering. Firstly, climate engineering techniques, with allowances made for their differences, have tended to be discussed together in assessment reports, governmental hearings, and major meetings and conferences. As such, it can be difficult to extricate SRM tests from this broader debate. Secondly, field tests in CDR techniques have produced conceptions of risk and proposals for governance that may have relevance for SRM.

While discussions of intentional interventions into the climate system have been going on for a considerable amount of time [38], the current wave of interest was sparked by a 2006 editorial by Paul Crutzen (who won a Nobel Prize for his work on ozone) in the journal Climatic Change [39]. The early ‘post-Crutzen’ era was marked by grey literature from a number of scientific networks based in the global North, calling for increased research but typically stressing that climate engineering should not undermine mitigation efforts. The UK Royal Society's 2009 scoping report ‘Geoengineering the Climate: Science, Governance, and Uncertainty’ is probably the best known of these and included technical reviews of different climate engineering technologies, as well as a discussion of norms for scientific research, a spectrum of scale for technology evaluation, and the development of future governance structures [35].

The first government-sponsored investigations began concurrently. The respective science and technology committees of the UK House of Commons and the US Congress both released scoping reports in 2010, relying on several months of hearings with some of the field's most prominent scientists, as well as exchanging documents and testimony in a collaborative effort that might indicate an early recognition of climate engineering's international dimensions [40–43]. In 2011, the German Federal Ministry of Education and Research published a scoping report that spanned climate science, economics, public perception, political science, ethics and law, which it had commissioned from an interdisciplinary group of researchers [37]. These efforts represent the (published) limit of governmental explorations, and only the UK and Germany have since taken brief and preliminary stances on the issues surrounding climate engineering (http://webarchive.nationalarchives.gov.uk and http://dipbt.bundestag.de).

These early reports have tentatively pointed out a number of risks and uncertainties that, significantly, have gone beyond technical assessments of cost and feasibility [44] and geophysical mechanics [45], to assessments of ethical and political dimensions by the social sciences and humanities [35]. For example, the most frequently cited fear in the social context of climate engineering, and particularly SRM research, is that researching or even discussing climate engineering may siphon resources from conventional and costly efforts at mitigation and adaptation [35]. Another fear focuses on the circumstance that if atmospheric carbon concentrations continue to increase during an SRM intervention, models show that a sudden halt to operations would result in a precipitous warming effect, or ‘termination shock’ [46]. Also related to the rising concentrations of CO₂ in the atmosphere is the circumstance that SRM would not address ocean acidification, a long-standing effect of rising CO₂ emissions. Ethical concerns have arisen over the ‘burden-shifting’ in addressing climate change, from the largest carbon-emitting nations to the populations that would suffer from the effects and side effects of climate engineering [47]; or, on a longer time scale, from our current generation to future ones [48]. Another concern is that of a ‘slippery slope’ effect, which claims that research will lead to more research and, eventually, to the implementation of a climate engineering technology. Moreover, it is frequently suggested that interventions in the complex climate system would create new, or intensify existing, conflict potentials within and across societies. Models have indicated that stratospheric SRM would inevitably have a global climatic effect, but alter regional temperature and precipitation unevenly [49]. Resulting (perceived) impacts upon lives and livelihoods may escalate into political conflicts, with strong implications for interstate relationships. A related worry is the potential for unilateral or even clandestine deployment of stratospheric SRM [50–52]. Regarding the international political dynamics surrounding SRM deployment, some have argued that SRM cannot be aligned with democratic governance and cannot become a ‘governable object’ [53,54]. Others have pointed out that SRM might in fact fit well with the current international political order, in which authority is increasingly centralized for technocratic problem solving in institutions of global governance, and that critiques of SRM often seem to be more concerned with the changes it might prevent than with the changes it might introduce [55–57].

In the absence of strong governmental positions, the strongest positions on governance come from within the research community itself. It should not be inferred that the research community is united. Some are far more pessimistic about the potential impacts of climate engineering, and tend to argue for severe limitations on field tests and prohibition of large-scale deployment [58–60]. Even among those who hold that deployment should be maintained as a potential option, there is a wide spectrum of opinion on how to calculate the allowable risks for field tests, on whether to situate governance for testing (or deployment) within one or a combination of international organizations and agreements or within the research community under various forms of self-governance, and on specific mechanisms of funding, risk assessment, liability, enforcement, and communication of results and best practices [9,26,35,52,61–69].

Yet, the accommodation of this multiplicity of views has become an increasingly entrenched norm. In a direct reference to rDNA technology, around 200 practitioners and academics from the natural and social sciences gathered in March 2010 for the Asilomar international conference on climate intervention technologies. The attendees produced a set of five recommendations for reducing risk and improving transparency in research and small-scale field testing, although this was not restricted to SRM tests [70]. The five principles were strongly based on the Oxford Principles, developed previously by a number of UK-based academics for the UK House of Commons 2010 report [71]. While no mandatory or enforceable strictures were derived (this was not the conference's intent), Asilomar 2010 was a visible effort to assess and bridge technical risks and societal issues and to ascertain the shape and scope of future governance [70,72,73].

This circumstance is not widely recognized. For example, one commentator reported on the 2010 Asilomar conference under the heading ‘Climate Hackers Want to Write Their Own Rules’ (www.wired.com); a perception that might well stem from the framing of the conference against the background of self-governance for rDNA technology. Other actors seem to have been similarly unaware of the contours of the emerging discourse, and expected a very different outcome from Asilomar. Before the Asilomar conference took place, the Canadian-based Action Group on Erosion, Technology and Concentration (ETC Group) composed an ‘Open Letter Opposing Asilomar Geoengineering Conference’. The letter concluded that ‘[t]he priority at this time is not to sort out the conditions under which this experimentation might take place but, rather, whether or not the community of nations and peoples believes that geoengineering is technically, legally, socially, environmentally and economically acceptable’ (letter available from www.etcgroup.org). Yet ironically, the Asilomar conference may have contributed to and strengthened the widening of the discourse to include new stakeholders and their conceptions of relevant issues and challenges, as well as a general attitude of caution and reflection. A definitive assessment on whether Asilomar 2010 was a seminal framing event on the level of its 1975 counterpart in rDNA cannot be made now. However, it was a moment that—if not seminal—was characteristic of the breadth of views involved in exploring the issue, and indicative of future developments. As Jeff Goodell notes:

Indeed, that same year, the UK Royal Society also launched the Solar Radiation Management Governance Initiative (SRMGI), in partnership with the Environmental Defense Fund and the Academy of Sciences for the Developing World, as a knowledge broker for international institutions, civil society organizations and national governments in SRM methods (www.srmgi.org).

Efforts to frame and initiate regulation of climate engineering go beyond the scientific community. For example, the ETC Group has played a disproportionate role in opposing the development of climate engineering technologies or policy options. It has described climate engineering as ‘Geopiracy’ [74], spearheaded a ‘Hands Off Mother Earth’ campaign advertised as a ‘coalition of international civil society groups, indigenous peoples organizations and social movements’ (www.handsoffmotherearth.org), and strongly opposes all field testing of climate engineering technologies. The ETC Group also lobbied the Convention on Biological Diversity's (CBD) 10th Conference of the Parties (COP) at Nagoya in 2010 [75], contributing to the adoption of a non-binding ‘decision’ that set criteria on the allowable limits of large field tests:

Certain scientists disapprove of the ETC Group's activist tactics, considering themselves as well as the wider field of climate engineering technologies to be subjects of overwrought rhetoric instead of measured criticism. For example, the CBD's 2010 decision is considered by the ETC Group to be a UN-backed ‘moratorium’ [76,77], when it has been otherwise criticized for its brevity and opaqueness, non-binding nature, inapplicability outside of biodiversity impacts and for the fact that the USA is not a signatory [78]. Yet, the key insight here is: given the visibility and influence of movements outside of the scientific community to govern SRM and climate engineering field tests, scientists cannot claim control over the framing of the debate.

At the international (or intergovernmental) level, the IPCC in 2011 held a scoping meeting in Peru to discuss the details of including climate engineering in the upcoming Fifth Assessment Report [79], and has consequently included a discussion of climate engineering in its Working Group I [80], with its Working Groups II and III reports still due at the time of writing. Yet, while the IPCC reports have in the past functioned as a basis for negotiations at the UNFCCC, this, by itself, is no definite indication that the issue will make it onto the agenda of future UNFCCC COPs. Currently, only the CBD and the London Convention and Protocol have taken decisions on climate engineering governance at the level of international institutions. The latter deals specifically with marine climate engineering, and its relationship to atmospherically based methods such as certain forms of SRM is unclear. However, tests of ocean iron fertilization (OIF; marine-based application of CDR) have spurred governance mechanisms that may hold lessons for SRM tests.

As a reaction to an attempt at using OIF to generate carbon credits for sale and following recommendations made by the International Maritime Organization, a 2008 resolution that emerged from a Conference of the Parties to the London Convention (on the Prevention of Dumping of Wastes and Other Matters) declared that OIF outside of ‘legitimate scientific activity’ would not be exempt from strictures against ocean dumping (Resolution LC-LP.1, 2008). This governance process was accelerated by a more recent incident. In the summer of 2012, 100 tons of iron filings were dumped into international waters off the western coast of Canada, after American businessman Russ George convinced a local aboriginal community to sponsor this as an effort to replenish local salmon stocks while earning credits from global carbon markets. As a result, the London Conventional and Protocol recently adopted a legally binding amendment that would only allow ocean fertilization that is legitimate scientific research and subject to an environmental impact assessment.

Another development of significance was the cancellation of the ‘test-bed’ stage of the SPICE project in May 2012. A joint endeavour by a number of prominent UK universities, the SPICE test-bed was intended to test the mechanics of aerosol delivery for SRM via a hose connected to a balloon. The environmental impacts of this field test were known to be negligible; yet even a preliminary version of the testing mechanism was cancelled in early 2012 due to opposition from civil society groups and the fact that one of the involved scientists had taken out patents on a related technology, as reported, for example, by Wilsdon 81.

The SPICE ‘test-bed’, despite its premature end, contained a degree of public engagement as well as careful structuring and communication of the test's goals and impacts in an attempt to provide governance that goes beyond considerations of technical risk [82–84]. Meanwhile, the transparency of the west coast OIF demonstration is characterized by different accounts [85]. Yet, both may exemplify a number of challenges described earlier: the power of non-governmental organization (NGO)-driven public concerns regarding climate engineering technologies, the importance of whether patenting of climate engineering technologies should be permitted, public fear of unilateral actions by individuals as well as governments, and the ambiguities surrounding regulatory and public consultation frameworks.

We can thus here observe an emerging dominant narrative of climate engineering that has been constructed by a wide set of actors in and outside of the scientific realm. This narrative is characterized by a broader context in which scepticism, restraint and caution dominate, and encompasses potentially expansive impacts on societal and political systems with difficult ethical questions involved. Perceived risks have long been expanded beyond the technical aspects of the individual technologies by a more diverse field of early participants in the scoping process. These include the interdisciplinary field of researchers and practitioners at Asilomar 2010, the emergence of government-sponsored scoping studies in Europe and North America, the outreach attempts of SRMGI to developing and emerging regions, the ongoing development of research guidelines at international institutions and the efforts of NGOs in shaping guidelines as well as opposing outdoors ‘tests’ such as the OIF experiment off the west coast of Canada and the SPICE test-bed.

Moreover, a number of social and political concerns have been widely accepted as relevant. Besides fears of negative effects of climate engineering research on emission reduction efforts, international conflict resulting from uneven temperature and precipitation impacts, unilateral action, ‘termination shock’ and slippery slope dynamics, climate engineering is deeply embedded in an already fractious ‘parent’ discourse, unlike the rDNA discourse during its early stages. The climate engineering debate—even on modelling and research—is intimately tied to the highly politicized debate on climate change. It does not carry the positive connotation of technological advancement and progress, leading to social and economic benefits in an open-ended field of applications and consumer products, as existed in the early days of rDNA technology. SRM in particular is seen as a protective technology which—even if it functions predictably and effectively—would produce intangible benefits that might be impossible to measure accurately, and are at the level of the global commons.

If the dominant narrative of an issue provides the context from within which the legitimacy³ of governance is assessed, then the rDNA ‘model’—in which scientists, as the natural locus of authority for devising governance, self-define and self-regulate technical risks—cannot be expected to be accepted as a legitimate governance mechanism within the wider discursive boundaries of what constitutes relevant risks and challenges in climate engineering. Both Asilomar conferences, in 1975 for rDNA research and in 2010 for climate engineering research, thus prove to have been important events in determining the future form of governance for the respective issue. But while the 1975 conference contributed to the emergence of a narrative that centred on technical questions, the 2010 conference was embedded within, and contributed to, the formation of an inclusive narrative that also focuses on complex social, political, and ethical issues. Interestingly, the inclusionary set-up of Asilomar 2010 may have had especial importance in undermining the capacity of scientists to devise governance for climate engineering, and reinforcing the claim that governance needs to include actors outside of the scientific realm. Indeed, its final report [70] included a call for broadening the spectrum of risks that is associated with climate engineering, to ‘include assessments of the full range of potential impacts, including environmental, economic, legal, and socio-political consequences’ (p. 22).

5. Implications for the governance of solar radiation management field tests

What, then, does this tell us about the upstream governance of SRM field tests? Field tests in the absence of appropriate governance would only undermine their own purpose to explore—and therefore, presumably, to enhance capacities to manage—the issues involved in SRM deployment. To conduct field tests (also at a small scale) without some governmental or public sanction would be to attempt to explore technical risk as if these existed in a vacuum—which is essentially what the early rDNA community did. However, the contours of the emerging dominant narrative of climate engineering indicate that even if scientists were able to ‘manage’ the technical risks of small-scale field tests, they would not forestall opposition that stems from concerns about the social, political and ethical impacts of climate engineering—the SPICE project may be a strong example.

Another way of putting this is that even insignificant physical risks are associated with a large platform of societal concerns raised by climate engineering—however plausible such dangers and proportionate such concerns might be. Future field tests, regardless of scale, may be associated with the narrative, and all its imaginaries, as a whole.

We might note the arguments of some recent papers. Parson & Keith [9] argue that a ‘safe’ threshold for SRM testing might be a radiative forcing perturbation ‘at a level where the global climate response is barely detectable’, and that ‘[i]nitial steps need not require the delay and inflexibility of enacting new laws or treaties but can come from informal consultation and coordinated decisions by research-funding and regulatory agencies of participating governments’ (p. 1279).

We agree with Parson & Keith [9] that new laws and treaties are probably not an adequate response to the immediate issue of small-scale field tests of SRM, and that international coordination between funding and regulatory agencies of participating governments is a more promising near-term strategy. Yet, it is crucial to recognize that such coordination needs to be in place before any field testing of SRM begins, and that it is self-defeating to shield off even small-scale field experiments from wider public involvement by the definition of technical thresholds based on the impact of an experiment on the physical environment.

Keith et al. [86] are correct in pointing out that the ‘considerable controversy’ around field tests of SRM is in part due to ‘concerns about the political impediments to making legitimate decisions about its implementation’. While their contribution, in which they clarify what constitutes realistic expectations about the characteristics of possible future field tests, thus may address concerns about the likely magnitude of such tests and their direct impacts on the physical environment, it cannot dispel concerns about the broader set of issues associated with field tests of SRM (as the authors acknowledge).

Self-governance by the scientific community cannot legitimately address the wider issues that are being associated even with small-scale SRM field tests, even if such tests stay below a certain technical threshold. Indeed, they may rather lead to a de-legitimization of research on SRM and to an increase in societal conflict, rather than contribute to building a framework that allows research to proceed safely. A regulatory backlash emerging against the background of such de-legitimization could also severely impact other basic research on atmospheric processes that involves perturbative experiments, which is direly needed to increase our understanding of the climate system. Even if one were in favour of a rapid research effort on climate engineering technologies with the prospect of rendering climate engineering a valid policy option as soon as possible, our analysis suggests that one might be well advised to refrain from conducting any field tests in the absence of governance mechanisms that are deemed legitimate for such activities within the general discourse on climate engineering.

Pinpointing the exact institution, or complex of institutions, as well as specific mechanisms of rule-making, membership, assessment and management, is beyond the scope of this study—however, climate engineering requires an approach to its governance that takes an extended set of societal considerations into account, making a simple transposition of the self-governance approach pursued during the upstream period of rDNA technology development to the climate engineering case impractical.

Similar to Parson & Keith, Victor et al. [67] note: ‘[t]he key is to draw a sharp line between studies that are small enough to avoid any noticeable or durable impact on the climate or weather and those that are larger and, accordingly, carry larger risks’ (p. 3). Arguing in favour of small-scale field tests, Victor et al. [67] go on to state: ‘until the science gets serious, the politics won't reflect what's really at stake’ (p. 3). Yet the acceptance and understandings of science (and the risks it defines) cannot be divorced from their political and cultural context. Rather, it is precisely this context that defines ‘what's really at stake’ and needs to be reflected in the governance of SRM field tests.