“By Design” and Risk Regulation: Insights from Nanotechnologies

Contemporary risk regulation requires an interdisciplinary approach that integrates science, law and socio-political discourses. This calls for new tools in the risk regulation process that enable regulators to adapt to a constantly changing technological realm and help overcome the interdisciplinarity dilemma. In the field of nanotechnologies, tools proposed in the literature include “by design” approaches. In this article, I analyse how the safe-by-design and benign-by-design concepts, emerging in materials science and drug development, could enhance interdisciplinarity in risk regulation. I suggest that further development and implementation of “by design” from a scientific concept to an adaptive regulatory tool could support progressive risk governance of innovative technological developments and enhance the interdisciplinary approach in risk regulation.


I. INTRODUCTION
The umbrella term "nanotechnology" embeds several disciplines, including materials science, biotechnology, physics, chemistry and medicine. 1 Nanotechnology refers to any science and technology at the nanoscale (at the level of atoms and molecules) and to scientific principles and new properties that can be understood and mastered in this domain. 2 Emerging (nano)technologies serve important social and economic purposes today. However, from a regulatory point of view, they are defined as "wicked" public policy problems; in other words, complex areas of policymaking involving a multitude of stakeholders (eg industry, politicians, non-governmental organisations) with competing values and interests. 3 The importance of the motivations of various users, communication and delivering compliance. 9 These criteria align closely with the characteristics of new governance set in the literature: inclusive and representative participation, collaboration, deliberative decision-making, experimentation, flexibility and revisability, new forms of accountability, learning and adaptation and transparency. 10 The International Risk Governance Council has stated that, in managing emerging risks, adaptability and flexibility are especially crucial. 11 However, there has been little attention paid to "tools" in the risk regulation process that would help overcome the interdisciplinarity dilemma, which arises from the following: risk regulation requires a range of expertise from different disciplines, and few scholars or policymakers possess the necessary integrated expertise. 12 For example, the focus on scientific methodologies in risk assessment and risk management has hindered other scholars and policymakers from becoming more deeply involved in these debates and developing real legal expertise in this area. Oomen et al and Isigonis et al have reviewed governance frameworks for nanotechnologies and concluded that, currently, tools to support holistic risk governance of nanomaterials are unsatisfactory, and the frameworks are insufficiently detailed to enable actual application either in the regulatory context or in informing decision-making. 13 Oomen et al recognised the need for greater interdisciplinarity in the risk governance of nanotechnologies. Dialogue between stakeholders is necessary to address, and transparently deal with, the uncertainty associated with the state-of-the-art science. One goal of the review by Isigonis et al was to assess the capacity of the existing risk governance frameworks for nanotechnologies to communicate risks to decision-makers. They stated that risk communication and stakeholder engagement are crucial cross-cutting aspects of the risk governance frameworks, but rarely mentioned.
The proposed risk governance tools for nanotechnologies involve "by design" approaches. 14 These include safe(r)-by-design (SbD) and benign-by-design (BbD) 9 Previously, it has also been stated that nanotechnology governance needs to be transformative, responsible, inclusive and visionary; see MC Roco, B Harthorn, D Guston and P Shapira, "Innovative and responsible governance of nanotechnology for societal development" (2011) 13 Journal of Nanoparticle Research 3557, 3559-60. 10  13 AG Oomen et al, "Risk assessment frameworks for nanomaterials: Scope, link to regulations, applicability, and outline for future directions in view of needed increase in efficiency" (2018) 9 Nanoimpact 1, 1-2, 10-11; P Isigonis et al, "Risk Governance of Nanomaterials: Review of Criteria and Tools for Risk Communication, Evaluation, and Mitigation" (2019) 9 Nanomaterials 696, 697, 699, 714-16. Oomen et al reviewed the relevant publications and reports of 23 research and regulatory bodies from the EU, the USA, the Organisation for Economic Co-operation and Development and Germany, as well as references from the open literature. Isigonis et al reviewed peerreviewed literature from 1990 to 2018 from the Web of Science database for journal articles pertaining to the risk governance of nanotechnologies, as well as relevant EU-funded projects from the Community Research and Development Information Service (CORDIS). They also organised a workshop for a variety of stakeholders to identify evaluation criteria, which they then applied in the analysis of the frameworks.
concepts that have emerged in the fields of materials science and drug development, respectively. In this paper, I analyse whether the concepts could serve as adaptive network management tools, which would enhance interdisciplinarity in risk regulation. "Network" refers here to various actors (eg academics, politicians, regulators, industry, non-governmental organisations, suppliers, customers, consumers) that may be involved in the innovation life cycle. "Network management tool", in the context of this article, engages relevant actors from different sectors of society in structured decision-making. The article is organised as follows: Section II introduces the SbD and BbD concepts and depicts their current status. Because inherently value-laden choices are involved in risk regulation (eg how to evaluate future threats and respond to them 15 ), discussion in Section III focuses on safety and innovation in order to assess the value of interdisciplinarity included in the by-design concepts in that context. In Section IV, I analyse the applicability of by-design concepts as regulatory tools to enhance interdisciplinarity by using the SbD concept as an example and reflecting on it in relation to the suggested characteristics for the risk governance framework for nanotechnologies and the criteria set in new governance literature: collaboration, participation, deliberation, flexibility, revisability, adaptability, learning and accountability. Finally, conclusions are drawn in Section V.

BbD concept
The roots of the BbD concept are in the "Green Chemistry Program" 16 and its initiative "Designing Safer Chemicals" of the US Environmental Protection Agency, which was launched on the basis of the Pollution Prevention Act 17 in the 1990s. 18 The "Designing Safer Chemicals" concept includes the structural design of chemicals to meet the needs of both safety and efficacy. The concept does not require zero toxicity or a maximum level of efficacy, but it does require the optimal balance, and it encompasses both human health and the environment throughout the chemical's life cycle.
Since the 1990s, the BbD concept has been developed further by Professor Kümmerer and his research group in the context of innovative pharmaceuticals. 19 According to the BbD concept, small alterations in the chemical structure of an active pharmaceutical ingredient may affect its activity, solubility and polarity, as well as its biodegradability, and a set of functionalities exists that can foster both. Properties of molecules can be predicted using modelling tools, and variations for a lead structure can be screened by in silico systems to find the best drug candidates in terms of activity and biodegradability. The candidates are subsequently tested experimentally. When combined with systems that allow for the prediction of metabolites, transformation products in the human body and in the environment can be considered in the design phase of pharmaceuticals.
Today, however, the tools and models for sustainable drug design must still be improved and proofs of the concepts established to increase acceptance of the BbD concept in the pharmaceutical industry. Although pharmaceutical companies applying green chemistry principles in manufacturing processes (post-regulatory approval) have reported impressive decreases (70-90%) in waste, incorporating BbD tools in the early design phase is still not common. Industry has stated that designing drugs for degradation in the environment is a major challenge because stability is required under all reasonable manufacturing, storage and use conditions. 20 Because the improvement of tools and models is largely dependent on the availability of experimental high-quality data for the biodegradability of pharmaceuticalswhich is mainly produced in-house in pharmaceutical companiesthe usability of the BbD concept suffers from an absence of open dialogue between different stakeholders. Key barriers to the implementation of green chemistry in the pharmaceutical industry include the mentality of medicinal chemists, the upfront costs of technology, unawareness of the available methods and regulatory risks. 21 Enhanced dialogue between the experts from different disciplines and the regulators might help to overcome these barriers.

SbD concept
The origins of the SbD concept in the nanotechnology context can also be traced to the USA, at RICE University (Houston) in approximately 2004-2005. 22 From the very beginning, the characteristic of safety in the SbD concept has been linked to materials' properties, making safety a concern of engineers and materials scientists. The development of nanomaterials that are safer by design has leaned on the processes used in drug discovery and development. The SbD concept appears in risk regulation discourse, especially in the EU, as a boundary object of several EU-funded projects (eg NANoREG 24 and NanoReg2 25 ). 26 According to Gottardo et al, the NANoREG SbD is a forward-looking strategy and a voluntary tool for considering innovation requirements and for helping regulatory authorities and industry to keep pace with innovation. 27 Safety information on materials, substances or products is iterated from early R&D phases onwards to search for the best achievable safety conditions. The term "design" is not restricted to the material properties, but applied to the whole innovation process, including production processes and final products. 28 Kraegeloh et al stated that the objective of SbD implementation is to transfer the precautionary principle into practice. 29 However, a combination of regulation and safety research is not an easy task, and implementation of the SbD concept has been difficult. 30 NanoReg2 addressed the difficulties in the practical applicability of the SbD concept by developing the "Safe Innovation Approach" (SIA), which links the SbD with "regulatory preparedness", defined as "The regulators' timely awareness of innovations and the regulator's actions to check whether present legislation covers all safety aspects of each innovation, including initiating revision of the legislation as appropriate". 31 Mutual awareness between regulators and industry, achieved through trusted environments for information sharing, was seen as key to governance of the safety of nanomaterials through the SbD concept and the SIA. 32 This is a clear indication of the demand for interdisciplinarity in the risk regulation of nanotechnologies. Isigonis et al noted that this approach is the first attempt to transition from risk governance to innovation governance. 33 They underlined that the outputs of the applied tools in any risk governance framework must be connected to policymaking and regulatory purposes to have a genuine impact (eg on innovation policy). It is not enough to develop sound risk governance frameworks by the scientific community if stakeholders cannot apply them in their respective societal contexts. 24 NANoREG <www.nanoreg.eu> (last accessed 22 July 2019). 25 NanoReg2 <www.nanoreg2.eu> (last accessed 22 July 2019).

Tools for risk assessment, risk management or both?
Safety is the core of both the BbD and SbD concepts, but the definition of safety has not been explicitly articulated. 34 The interface between law, regulation and science is clearly on stage when we consider safety and the BbD and SbD concepts. The definition of "safety" depends on the context in which it is used. does not include a definition of safety, even though the term "safety" is referred to repeatedly. The absence of definitions is because safety is a relational value and absolute safety cannot be achieved. This leads to questions, for example: What is safe enough? Who decides and defines the acceptable level of safety? These are political challenges, embedded with social assumptions, and thus cannot be reduced to a technical dilemma for scientists and innovators. 36 This reveals the unrealistic vision of control included in safety design approaches that could be implemented in the design processes, because of prevalent ignorance and/or uncertainty about the effects of innovative chemicals. 37 The SbD concept can be considered as multidimensional; in other words, as: (1) an approach to risk assessment; (2) a specific risk management strategy; and (3) a result of the design process. 38 As a risk assessment approach, the SbD concept implies that risks are already assessed in the design phase. As a risk management strategy, it addresses SbD measures (built-in safety). When considered a result of the design process, the SbD concept claims absolute safety and the absence of risk. This last aspect is utopian, and as a point of comparison, it is never aspired to in the context of drug development. 39 On the other hand, the second approach requires the first, and thus they are intertwined. 40 Therefore, the SbD concept is a tool for both risk assessment and management. But how could it enhance the consideration of safety in risk regulation?
The answer is: by creating a structured decision-making platform for a network of stakeholders, which may be involved in the innovation life cycle. This will be performed by defining the innovation project's workflow with specific decision points 34  in each phase of the innovation cycle. 41 Project-specific interdisciplinary dialogue and the possibility of evaluating all of the available information and using it as early as possible in the decision-making process are the benefits of the SbD process. At the national level, Germany has utilised the "NanoDialogue" platform for stakeholder dialogue since 2006 as part of the German government's Nano Action Plan. 42 The NanoDialogue platform was launched by the NanoKommission, "a societal control group" that comprised approximately 20 members as a centralised, national platform. 43 The value of the NanoKommission was that it provided consensual knowledge that was co-produced by all relevant stakeholders from different sectors of society. 44 The NanoKommission as a quasi-external body (external from science and law) was able to make useful decisions in the interdisciplinary dialogue and provide norms for science and the economy in order to cope with scientific uncertainty. The application of the SbD concept means carrying out interdisciplinary dialogue at the project level. This may result in even more concrete effects on science and the economy and help to break the silos between innovators, safety experts, regulators and the public than dialogue at the national level. Jasanoff noted that technical and social orders are co-produced in each policy regime. 45 Science and Technology Studies have recognised three mechanisms that should be considered in democratising the governance of innovative technological developments, namely the market, regulation and ethical deliberation. 46 While the market emphasises efficiency and regulation emphasises rationality, ethics is concerned with moral values rooted in culture. Consequently, safety should be understood in relation to numerous, often vague entities (eg the environment) that decision-makers encounter during risk regulation. In addition to safety, other values (eg social or ethical acceptance, privacy) come into play at varying degrees under different policy regimes, creating normative 41 Kraegeloh et al, supra, note 28, p 243. The SbD process is elaborated in more detail in Section III.2. 42 Federal Ministry for the Environment, Nature Conservation and Nuclear Safety, "The NanoDialogue" <www. bmu.de/en/topics/health-chemical-safety-nanotechnology/nanotechnology/the-nanodialogue> (last accessed 29 April 2020). 43 The ambiguity, a type of uncertainty that the concepts should be able to address. 47 Fisher stated that risk regulation regimes have developed very different means in a jurisdiction. 48 Thus, policymakers seeking to implement the by-design concepts should consider their applicability in different policy regimes. Because the SbD concept will be applied to project-specific data and, connected with the SIA, results in regulation-specific safety dossiers, 49 it is adaptable to different policy regimes. In addition, norms produced in interdisciplinary dialogue are adaptable to diverging societal expectations. However, to be regulatory tools, the rules of behaviour that frame the interaction should be formalised in each policy regime through referring to them in the relevant regulations or through application of the rules by public authorities. 50 Standardisation may assist in the application of the concepts. Currently, the European Committee for Standardization (CEN) is preparing a standard for the SbD concept. 51 Similarly, BbD encompasses both risk assessment and management dimensions because the material characteristics that affect risk potential (eg solubility, genotoxicity, ecotoxicity) are assessed early as drivers of drug development towards more rational design, focusing on risk mitigation by hazard reduction. 52 The BbD concept could enhance the consideration of safety in risk regulation, for example, so that chemicals that remain in wastewaters are intentionally designed to mineralise rapidly in effluent treatment processes or in surface waters. 53 Although there are limited possibilities to vary the core functional parts of these molecules, other parts can be varied to a much greater extent, and encouraging examples are available even for widely used pharmaceuticals. 54 Today, however, increasingly complex products 47 van de Poel and Robaey, supra, note 39, pp 299, 300. See also Schwarz-Plaschg et al, supra, note 22, p 278, who argued that the recent focus on SbD and safety in the EU context has narrowed the nanotechnology debate to safety concerns instead of broader innovation governance questions that were on the agenda at the beginning of the debate on nanotechnology. 48  Dutch National Institute for Public Health and the Environment (RIVM), "ProSafe Safe-by-Design (SbD) Implementation Concept" 1, p 4 <www.rivm.nl/en/documenten/prosafe-safe-by-design-sbd-implementationconcept-final> (last accessed 1 May 2020). The SbD concept has been tested for different processes in pilot projects. For example, the R2R Biofluidics project has published a SbD report for large-scale micro-and nanofabrication technologies for bioanalytical devices based on roll-to-roll imprinting. See R2R Biofluidics, "Safe-bydesign report. Deliverable 9.3" <www.r2r-biofluidics.eu/images/deliverables/D39_D93_Safe-by-design_report.pdf> (last accessed 1 May 2020). 50 A Reinchow, "Risk, Uncertainty, and Learning in Nanomaterials Regulation: An Analytical Framework" (2016) 7 European Journal of Risk Regulation 502, 507, 513. Currently, for example, RIVM introduces on its webpages a concept version of the SIA Toolbox developed in the NanoReg2 project. RIVM states that the Toolbox is a set of tools, guidelines and checklists to be used by innovators and regulators along the innovation chain, and that it supports both improved dealing with safety issues and improved regulatory preparedness. See Dutch National Institute for Public Health and the Environment, "NanoReg2" <www.rivm.nl/en/about-rivm/mission-and-strategy/international-affairs/internationalprojects/nanoregii> (last accessed 1 May 2020). 51  54 Such pharmaceuticals include, for example, β-blockers, which are a class of active pharmaceutical ingredients. For example, the β-blocker propranolol has been reported to be a non-biodegradable and highly persistent chemical, which has been detected in effluents of sewage treatment plants. See T Rastogi, C Leder and K Kümmerer, "Re-designing of with diverse compositions enter the market and end up in the environment. This is largely because of the lack of incentives to develop compounds with fast and complete mineralisation in the environment. In this connection, it must be stressed that the stability of pharmaceuticals is required under all reasonable manufacturing, storage and use conditions, as stated above. Consequently, early screening is always followed by in-depth toxicological testing and regulation. The BbD concept (and the SbD concept) constitutes only a starting point that cannot replace regulatory risk assessment, a prerequisite for market access. However, chemicals that will readily mineralise in the environment will not necessarily need extensive testing regarding their environmental effects. 55 As an example of how the by-design concepts could be incorporated into the existing legislation, environmental risk assessment (ERA) of medicinal products for human use in the EU and the USA is briefly explored. Article 8(3)(ca) of Directive 2001/83/EC of the European Parliament and of the Council of 6 November 2001 on the Community code related to medicinal products for human use 56 outlines that in order to obtain an authorisation for a medicinal product, the application shall be accompanied by an evaluation of the potential environmental risks posed by the medicinal product (ie an ERA). However, risk-benefit analysis does not consider the ERA. It is solely based on risks relating to the quality, safety or efficacy of the medicinal product as regards patients' health or public health (Article 1(1)(28, 28a)). If the ERA points out potential environmental risks, specific safety measures related to the storage of the medicinal product, its administration to patients and for the disposal of waste products to limit the risks case by case shall be envisaged by the applicant (Article 8(3)(ca, g)).
In the Federal Food, Drug and Cosmetic Act, 57 21 U.S.C. § 355(b)(1)(A) sets only a demand to provide full reports of investigations that have been made to show whether the drug is safe and effective for use. However, the Code of Federal Regulations (CFR) 21 § 25.20(l) determines that approval of a new drug application requires the preparation of an environmental assessment (EA). Contrary to the EU's practice, the US Food and Drug Administration (FDA) can refuse to approve an application if the EA does not contain sufficient information to enable the FDA to determine whether the proposed action may significantly affect the quality of the human environment (21 CFR § 25.15(a)). 58 The FDA can also command the applicant to implement riskmitigation measures (21 CFR § 25.40(e)). However, because of many categorical exclusions (21 CFR § 25.31), the EA has not been commonly required during the new drug application process, and consequently, the mandate of the FDA to consider environmental risks and to require risk-mitigation measures is limited. In that case, existing pharmaceuticals for environmental biodegradability: A tiered approach with β-blocker Propranolol as an example" (2015) 49 Environmental Science & Technology 11756, 11757. 55  Both jurisdictions aim to regulate the impacts of pharmaceuticals on the environment by assessing the impact of a product at the market approval phase. 59 It could be reasonable to enforce pre-screening of environmental considerations in the early design phase by applying the BbD concept. This would not replace the ERA or EA at the market approval phase, but, as noted above, it might result in less extensive testing in the future if the approach proves to be feasible. As stated in Section II.1, increased dialogue between the experts from different disciplines and the regulators is needed to enhance the implementation of green chemistry in the pharmaceutical industry. Compulsory application of the BbD concept in the early design phase might make this happen.

Supporting radical or incremental innovations?
Defining innovation is far from straightforward because the meaning depends on the context and perspective (eg scientific community, field of industry, customer, society). Regulation is a central factor in the management of the positive and negative societal impacts of innovation. 60 In the past, risk regulatory approaches have allowed harmful innovations and hindered useful innovations, indicating that regulators should find new means of engaging with emerging and evolving technologies. Smismans and Stokes discussed how different innovation types actively shape regulatory responses to new technologies. 61 They showed that the distinction between "radical" and "incremental" innovation may affect the desirability of a new legislative framework, the nature and extent of the evidence base for regulation and the use of the precautionary principle. Defining a technology as incrementally innovative releases policymakers from exploring wider socioeconomic implications. Thus, policymakers are relieved of further understanding of the effects of nanotechnologies on the environment, economy or societies that would require interdisciplinary approach. Smismans and Stokes argued that the distinction, providing possibilities for interpretation, has been used to justify different regulatory strategies adopted towards nanotechnology by the European Commission (incremental) and the European Parliament (radical). Consequently, innovation is an object of governance and an instrument of governance, actively steering regulatory responses in the directions determined by policymakers through interpretations of the type of innovation. Policymakers may describe technology as 59 An assessment of the impacts of active pharmaceutical ingredients on the environment is currently not required, but a draft guideline on the ERA of medicinal products for human use in the EU considers active pharmaceutical ingredients and metabolism of the active substance may be taken into account as well. See European Medicines Agency, "Guideline on the environmental risk assessment of medicinal products for human use" 1, pp 4, 8 <www.ema.europa.eu/documents/ scientific-guideline/draft-guideline-environmental-risk-assessment-medicinal-products-human-use-revision-1_en. pdf> (last accessed 29 April 2020). 60  incrementally innovative to avoid the need for a new legislative framework and broader evidence-based impact assessment that might stifle or delay innovation; namely, they prefer deliberate regulatory ignorance. 62 It must be noted that risk regulation can be conceptualised and understood in a different manner, even in the same jurisdiction, to produce pragmatic frameworks for regulatory decision-making. 63 Practices of regulators, particularly related to innovations, deserve more attention. 64 The SbD concept in the EU has adhered strongly to the innovation process and economic concerns, raising criticism that safety may become conceptualised as an enabler of innovation, without absolute value. 65 Other points of criticism have been raised with respect to the limitations of existing innovation management frameworks as a basis for the SbD approach. These arguments are, for example, that the models: (1) may not be easily applicable to global networks with diverse interests and safety expectations that are involved in the manufacturing and commerce of innovative nanomaterials; and (2) do not consider that the processes may be better conceptualised as non-linear collective and iterative learning. Jasanoff has stated that as the markets of innovative technological developments expand, democratic processes may not be able to ensure accountability towards all affected parties. 66 In my opinion, the SbD concept could help to combine techno-scientific and socioeconomic aspects of risk regulation under the same interdisciplinary approach and could support radical innovations by requiring collaborative, transparent interactions between actors throughout the innovation chain. 67 According to Kraegeloh et al, SbD is a bottom-up approach that includes various actors (eg industry, academia, regulators, customers, consumers, society) and covers the whole innovation process. 68 They stated that the concept could be implemented by industry and that regulators could use it as a reference tool that integrates currently used risk assessment practices, innovation management processes, environmental, health and safety assessment, regulatory affairs and data handling. The SbD process can be applied to different processes, products, companies and industries, but the data are project specific. The SbD process is based on the Stage-Gate® innovation process model, in which "stages" represent product development milestones and "gates" provide intervention and adjustment opportunities. 69 Implementation of the SbD concept starts with the definition of the innovation project's workflow, whose structure can be complex, with many actors interacting in different phases of the innovation process. 70 Decision-makers 62 (gatekeepers) at each gate decide on the fate of the innovation project based on balancing expected risks, costs and benefits. 71 The process or product is reviewed before going ahead if the collected information indicates this to be appropriate. Thus, the SbD concept, as a platform for interdisciplinary decision-making, enables iterative learning. Although incorporation of all aspects and actors of global networks involved in the innovation chain is obviously impossible, the SbD concept might enhance transparent dialogue between different stakeholders, which is of the utmost importance in innovation governance.
Although the BbD concept per se has not been closely connected to the innovation process, economic, social and ethical aspects, as well as new business models, have been recognised in the broader context of sustainable chemistry. 72 New business models may create win-win situations between the customer and provider, supporting innovation while reducing the chemical-related environmental burden. Innovative pharmaceuticals, developed according to the BbD concept, are examples of such new business models. 73 Schmutz et al have stated that the efficiency of the innovation process and the collaboration of all involved interdisciplinary actors would be improved if a methodological approach that evaluates the safety of nanomedicines early in the product development phase was available. 74 In this section, I evaluated how by-design concepts could enhance the consideration of safety in risk regulation and support radical innovations by increasing interdisciplinary dialogue between different stakeholders. In Section IV, I analyse the applicability of the concepts as regulatory tools for enhancing interdisciplinarity by using the SbD concept as an example and reflecting on it in relation to the suggested characteristics for the risk governance framework for nanotechnologies and the criteria set in new governance literature.

MANAGEMENT TOOLS
Adaptability and flexibility are identified as key characteristics of the frameworks intended to manage emerging risks. 75 These are closely related to learning, revisability and new forms of accountability incorporated into the multiscale collaborative governance arrangements presented in new governance literature. 76 In addition, Holley et al selected two more characteristics for evaluating new (environmental) governance institutions: participation and deliberation. 77 I analyse the applicability of by-design concepts as regulatory tools to enhance interdisciplinarity by using the SbD concept as an example and reflecting on it in relation to these characteristics, grouped as (1) collaboration, participation and deliberation; (2) flexibility and revisability; and (3) adaptability, learning and accountability.

Collaboration, participation and deliberation
Kraegeloh et al do not explicitly describe how the SbD concept would ensure effective collaboration, participation and deliberation. Holley et al examined how these objectives have been achieved in practice under three new governance programmes. 78 They stated that implementing multi-stakeholder collaboration is demanding and that at least the following should be considered: transaction costs, trust, inclusiveness and representativeness and rules of decision-making. The mechanisms mentioned to address these points include incentives (positive or negative), building trust and a consensus approach. Kraegeloh et al recognised that information sharing is crucial in the implementation of the SbD concept and suggested that an Internet platform should be developed to exchange information between participants. 79 The main objective of such a platform would be to build trust, especially between industry and regulators, but it could also serve as a forum in the SbD process to accumulate and share information with a wider group of participants. The information platform would be one source of information in the SbD process, which also includes the participation of the other actors.
However, how the inclusive and representative participation and deliberative decisionmaking during the SbD process could be arranged remains unclear, and this may result in imbalances in power during decision-making. 80 Guston and Sarewitz stressed that informed societal responses to innovation depend on how well various societal actors prepare for the impacts of the innovation, and there must be established processes (eg consensus conferences, scenario workshops, focus groups) that help society make actual choices about the progress, direction and application of innovation. 81 Oomen et al proposed that to enhance informed decision-making in the innovation chain, a possible way forward could be the development of a pragmatic, internationally accepted nanomaterial decision framework with only partially scientifically based decision criteria. 82 The adoption of such a framework requires cooperation between policymakers, regulators, scientists and industry. SbD (and BbD in the context of drug development 83 ) could enable this type of framework if processes and rules for decision-making can be established, because broad collaboration is considered an inherent part of the SbD project. However, realising collaboration is time and resource intensive, and transaction costs may hinder the implementation of the SbD concept if it is based only on voluntary cooperation without incentives. 84

Flexibility and revisability
Because the governance of emerging technologies must manage the low level of knowledge regarding causal or functional relationships between risk sources and their environment, flexibility and revisability are core characteristics of governance frameworks. 85 Kraegeloh et al stated that because regulations and standards applicable to different actors and at different times of an innovation project vary, the SbD approach must be flexible in order to manage (and revise on demand) the safety and regulatory data requirements of each phase of the innovation chain, and it must organise the links between actors that have different roles. 86 When a workflow of an innovation project is determined, the decision points at which the data will be reviewed and decisions about the project status will be taken are defined. Required revisions can arise from safety concerns, functionality shortcomings or feedback from the public or regulators. Consequently, the SbD project includes interdisciplinary decision points in all phases of the project workflow. In addition, flexibility is included in the SbD library (inventory of available tools; eg for risk assessment), which will be constantly updated with the most advanced tools and standards, following the progress in various fora (eg the Organisation for Economic Cooperation and Development, International Organization for Standardization). The use of internationally acknowledged guidance documents (eg on standardised testing) is highly recommended in order to improve the applicability of the by-design approaches in different policy regimes. 87

Adaptability, learning and accountability
New forms of accountability must be considered in dynamic, multiscale, collaborative arrangements in which non-government actors have important roles. 88 Conventional accountability mechanisms (eg authorisation and judicial review) may not have the flexibility to facilitate incremental decision-making. New mechanisms can be roughly divided into two categories: process mechanisms and performance mechanisms. For example, risk assessment belongs to process mechanisms, whereas the performance mechanisms focus on outcomes (eg compliance). Learning can be process-based as well; that is, encouraging industries to self-reflect and learn about their environmental impacts. A more advanced mode is systemic learning, in which information is shared between collaborative actors to diffuse innovation and to facilitate continuous adaptation, designed to enhance compliance with policy targets.
The SbD process involves both process-and performance-based accountability mechanisms (eg mutually agreed data requirements, compliance with thresholds and functionality specifications set during the SbD project) and process-based and systemic learning. 89 Process-based learning arises when the needs and the accuracy of data increase during the innovation process. In the early phases, in-depth knowledge of some aspects is necessary, and the process iterates until the collected information allows the industry to proceed to the next phase. Systemic learning emerges in the idea of regulatory preparedness that is included, along with the SbD concept, in the SIA. This makes it possible for regulators to be aware in a timely manner of anticipated implications of innovations and to revise relevant legislation as appropriate. 90 Holley et al stressed that effective monitoring processes are critical for the achievement of accountability and learning goals. 91 In the SbD process, a stagegate model with specific decision-making points (gates) provides an appropriate frame for the monitoring scheme, if the fulfilment of the mutually agreed data requirements or compliance targets can be achieved.
Armstrong et al introduced an advisory, adaptive, anticipatory model for regulation that emphasises flexibility, collaboration and innovation. 92 Under the model, regulators have a positive, proactive role in shaping how innovations are developed and deployed. They stated that new, more proactive regulatory practices would help the regulators ensure that economic and social benefits are achieved while risks are better understood and managed. Reinchow developed a framework for the analysis of learning in governance networks. 93 She stated that because diverse actor groups (regulators, industry, research institutes) have created a complex game of collaboration, it might be useful to have steering in place to facilitate sound network collaboration. By-design concepts could serve as adaptive network management tools that build trusted relationships between innovators, safety experts, regulators and the public, thus enabling the systemic learning that is necessary for effective regulation of innovative technologies under scientific uncertainty. The networks of actors are dynamic, as Reinchow argued, but in order to be regulatory tools, the rules of behaviour that frame the interaction should be formalised, as discussed in Section III.1. 94 V. CONCLUSIONS Nanotechnologies serve important social and economic purposes, but for regulators they are wicked public policy problems involving a multitude of stakeholders with competing values and interests. Consequently, the need for greater interdisciplinarity in the risk governance of nanotechnologies has been recognised. This paper analysed how the SbD and BbD concepts, which have emerged in materials science and drug development, could enhance interdisciplinarity in risk regulation.
The analysis showed that the by-design concepts create platforms for interdisciplinary dialogue and decision-making that may also enhance the consideration of safety in risk regulation. The concepts might also support radical innovations and informed decisionmaking in the innovation chain by requiring collaborative, transparent interactions between actors throughout the innovation chain. In addition, the efficiency of the innovation process may be improved if environmental considerations would be evaluated early in the product development phase. The analysis showed also that bydesign concepts might serve as adaptive network management tools that build trusted relationships between innovators, safety experts, regulators and the public, thus enabling the systemic learning that is necessary for effective regulation of innovative technologies under scientific uncertainty. However, to achieve this goal, the rules of behaviour that frame the interaction should be formalised.
The core of the involvement of the SbD or BbD concepts in risk regulation discourse should be a shared responsibility for safety that combines techno-scientific and socioeconomic aspects. Different actors should continuously, in the frames laid out for the interaction, evaluate where responsibility for safety is best addressed. Technologies and regulations should be designed accordingly. 95 Proactive regulatory practices and broad collaboration as inherent parts of the by-design approach would help regulators ensure that economic and social benefits are achieved while risks are better understood and managed. This would enable adaptive risk management and provide democratic opportunities to shape technology, increasing resilience in the risk governance of emerging technologies. 95 See also van de Poel and Robaey, supra, note 38, pp 297, 300. They argued that adaptive risk management organises a learning process with respect to both what the risks are and how to best manage them, and innovations could be designed to be flexible so that they can be adapted or used differently if new risks emerge. For a more detailed discussion of design for responsibility, see JN Fahlquist, N Doorn and I van de Poel, "Design for the Value of Responsibility" in J van den Hoven, PE Vermaas and I van de Poel (eds) Handbook of Ethics, Values, and Technological Design (Berlin, Springer ScienceBusiness Media 2014).