Integrated Assessment of Societal Impacts of Emerging Science and Technology from within Epistemic Networks

Final Report Summary - EPINET (Integrated Assessment of Societal Impacts of Emerging Science and Technology from within Epistemic Networks)

Executive Summary:
EPINET has been analysing four fields of technoscientific innovation, most of which are in relatively early stages of implementation and marked by only partial and incomplete (to varying degrees) understandings of their societal, ethical and legal implications. These partial and incomplete understandings are generally paralleled with inadequately mediating (integrating, cross-cutting) policy actions, institutions and networks. The innovation domains are: wearable sensors and behavioural change (WP3), - autonomous robots in care and companionships (WP4), - the growing of in vitro meat (or not) (WP5), and smart grids for transition towards sustainable energies (WP6).

In terms of European policies these developments can be traced to (at least) two different, though related, sources of influence. First, there is a persistent drive towards innovation which by now has a long tradition in Europe but became much more pronounced following the Lisbon agenda. The drive towards innovation was further accelerated by the economic downturn of 2008, although, conditions for funding have shifted. Higher contributions of national budgets and industry will now have to make up for lesser EC funds available to implement the Horizon 2020 policy agenda and funding programme. But, the general agenda still follows broadly the 2006 Aho report recommendation that: Europe and its citizens should realise that their way of life is under threat but also that the path to prosperity through research and innovation is open if large scale action is taken now by their leaders before it is too late. The second source of influence for these policies is of a more recent origin, and perhaps not yet well understood nor well articulated. It is the notion laid down in the 2009 Lund Declaration that research and innovation should target directly key societal challenges. This is implemented already in 'priority 3' of the Horizon 2020 policy agenda, titled 'Societal Challenges' which then have been integrated into the very structure of H2020 programmes and calls.
As a consequence of these demands now placed on research and innovation agendas we observe strong tendencies towards 'cross-sectoral actions', as when the recent commissioner for Research, science and innovation, Carlos Moedas (paraphrasing Jean-Claude Juncker) in his inauguration speech stated how 'working in silos is not an option'. During his speech, Moedas kept returning specifically to the topic of policy silos and the need for cross-cutting actions. Although Moedas is navigating the Brussels bureaucratic machinery (which is one type of cross-cutting, ie. between DGs, between expert networks, between nation states) the metaphor of 'silos' and of breaking them down, is indicative of a problematic which has drawn our attention to what we term 'technoepistemic networks'. These are networks of actors (know-how and resources) working to innovate and integrate across sectoral and disciplinary domains throughout Europe and beyond. It is also characteristic of developments within the TA and RRI communities towards a pooling of resources, working in larger interdisciplinary teams, and generally 'integrating' with technology developers, researchers and (to some extent) policy makers. Hence, as we now explain, these developments lead to a blurring of boundaries between what is traditionally conceived as the relatively separate domains of research and innovation, policy making and technology assessment.

Project Context and Objectives:
Scientific background and research process:
During the last 30-40 years a number of different assessment methods have been developed and implemented on national, international and EU levels to deal with the societal and environmental implications of new sciences and technologies. Although much has been achieved in fields such as technology assessment, ELSA studies and public participation these are also characterised by great diversity and a plurality of methods. Although diversity may be a resource it can also turn out a decisive hindrance to communication and action for principal end users, such as policy makers or publics. In part, the diversity of these fields comes from the fact that different forms of assessments are undertaken for a number of different purposes. To determine risk or toxicity levels will not necessarily increase public debate; public perception, debate and precautionary approaches to risk assessment may be perceived as obstacles to innovation, and so on. Where real underlying conflict of interest exists, it might be better to spell out the underlying values and presuppositions without seeking further consensus. However, differences of perception may also be due to (differing) values and epistemologies built into assessment practices and methods themselves. Especially here there is potential for improvement. We use the term epistemic network to refer to three levels of analysis and practice:

The first denotes contexts of innovation, denoting complex and intersecting relations of professionals, technologies, citizens, users, entrepreneurs, business and policy-makers forming new constellations of collaboration, experimentation and reflection to meet societal challenges.

The second relates to political levels of governance. These may try to accommodate innovation to meet so-called grand societal challenges, such as the goals set out in the Europe 2020 Strategy, but also with addressing concerns of publics and citizens.

The third level relates to the context of assessment, i.e. the activities of people working on the interfaces of different scientific disciplines and policy making in order to better assess and evaluate the implications of new and emerging technologies.

It is the potential for tighter integration between three main fields of practice, i.e. contexts of innovation, governance and contexts of assessment, which serves as the analytic point of departure as well as the critical (regulative and normative) goal for EPINET.

The notion of epistemic communities emerged within the academic field of international relations (Haas 1992). EPINET proceeds to expand on conventional notions of knowledge and expertise by situating it as practice-based and locally contingent (Polanyi 1958), as also identified in the “Practice Turn” in Science and Technology Studies (e.g. Schatzki et al., 2001). EPINET also builds on and expands the notion of epistemic communities into that of epistemic networks by shifting the focus from epistemic and normative commitments of expert communities to networks forming as the result of new imperatives for S&T development as outlined in EU policy frameworks such as the i2020 initiative. In principle, therefore, anybody responding to or contesting a grand societal challenge by engaging in innovation activities together with others is a potential member of an epistemic network. Such networks emerge on the intersections of communities traditionally separated as “expert” and “lay” knowledge. On that account, common experiences, insights, knowledge and creativity have to be included as relevant forms of expertise. The same, of course, goes for knowledge and normative commitments held by user-based communities, such as civil society organisations and other NGOs, patient organisations, professional organisations or labour unions. One shift of focus which we explore, but which has been recognised neither in the academic analysis nor practice of technology assessment, is to move from the implied assumption that methodological elaboration will be sufficient to encompass all the salient factors, to the idea that some of the issues here require appropriate carefully designed institutional changes.

The project runs through three main stages of development:

Initial assessment (months 1-12): The first stage corresponds to a general mapping of methodologies, networks and policy issues, including intrinsic values, framing premises and purposes shaping methodologies and procedures. Two work packages (WPs 1 and 2) will work on cross-cutting (conceptual and disciplinary) perspectives, investigating the potentials and limitations of methodologies, disciplines as well as central policy concepts dealing with grand societal challenges. At the same time, and in coordination with WPs 1 and 2, work will start on the specific cases (WPs 3-6). Here, each case will be studied in accordance with dominant TA methodologies whereas at the same time focusing specifically on epistemic networks, grand societal challenges and the policy concepts meant to govern them. Hence, this stage will also serve as a general mapping, both of the context of innovation (R&D) as well as the context of assessment (the different methodologies and their bearings on the cases).

Embedding assessments (13-24): The second stage expands initial assessments by bringing them into interaction with different epistemic networks; scenario development activities will bring together broad groups of thinkers and practitioners to explore pressing policy issues through two workshops. The workshops enables the co-creation of main issues on the intersections of policy and innovation, research and development and as such enables industrial players to engage in reflexive learning with diverse experts doing assessments of the particular cases. In order to bridge the institutional, philosophical, professional and cultural perspectives that are relevant to thinking systematically about the prospects of emerging technologies, the dialogue and research leading to scenarios is at once open ended and highly structured.

Comparison and integration (months 25-36): The final stage will compare, analyse and work out the general implications of the project. Here, the results from the implications workshop will be distributed through the project’s channels as well as attendant publications describing the process and outcomes to public and academic audiences. The feedback from participants and end users will be analysed along with preceding results from all work packages: first of all, for each case this concerns the results gained through the distinct methodologies; in a next step, this must also be placed in relation to the results from the cross-cutting WPs 1 and 2. Taken together, results will be used to draw conclusions and for recommendations for assessment practitioners, policy makers and other end users and interested parties.

Objective 1: identify and provide empirical descriptions of epistemic networks in the four socio-technical domains of wearable sensors, cognitive factories, synthetic meat, and smart grids, and their relations with strategic EU innovation policy goals (European Commission 2010).

In EPINET, this objective was met by WP3-WP6, with corresponding milestones and deliverables in the series M3-6.x and D3-6.y in particular during the phase of initial assessment (see work plan).

Objective 2: critically compare different assessment methods as described under objective 1, in particular to focus on the use of non-economic and other non-quantifiable factors (see objective 3).

Objective 3: to provide insights on the role and potentials of non-economic factors as drivers of socio-technical innovation.

In EPINET, this objective was addressed by a number of WPs, including the Embedding assessment and Comparison and integration phases of WP3-WP6 in the case of the four chosen technological fields, and throughout WP1 and WP2 for the general case (see work plan).

Objective 4: Alongside objectives 2 and 3 we go beyond comparison (objective 2) and seek genuine integration of methods to evaluate new and emerging innovations.

In EPINET, this objective was met by WP3-WP6, with corresponding milestones and deliverables in the series M3-6.x and D3-6.y in particular during the phase of embedding assessment (see work plan).

Objective 5: Provide tools, methodologies and insights for policy makers wanting to engage in more meaningful and effective ways with socio-technical innovation practices, in particular in the context of responsible research and innovation.
This goal was reached by a double set of means: first, the provision of thorough comparison and interaction between different TA methodologies, including unpacking their wider goals and presuppositions; second, by working towards tighter integration through the concept of epistemic networks.
In EPINET, this objective was met by the final phase of WP1 and WP2, in which the experiences from the four case studies will be used to generalise insights (see work plan).

Objective 6: Provide policy recommendations for the four socio-technological domains.
In EPINET, this objective will be met by WP3-WP6, with corresponding milestones and deliverables in the series M3-6.x and D3-6.y in particular during the phase of comparison and integration (see work plan).

Objective 7: Provide policy recommendations for the future use of the EPINET integrated assessment framework in a general framework for responsible research and innovation.
This is the final and in one sense most important objective of EPINET. All work will lead up towards it, where as its final formulation belongs to WP1 in close collaboration with WP2.

Project Results:
In three of the above-mentioned cases we identify explicit policy agendas that connect technoscientifc innovation with specific societal purposes, ie. wearable sensors, autonomous robotics and smart grids. In the vocabulary used by Epinet, each area is operatively an epistemic network (Rommetveit et al. 2012, Haas 1992, Nordmann 2006). Although primarily dedicated to the making of new things and process through innovation, these networks are also indicative of hybridity in forms of governance. Some characteristics of these networks include:
• they actively work to organise innovation and market-making at a European level, targeted towards European policy goals;
• they make new relations across national boundaries by joining forces with relevant forms of expertise and experience from different countries.
• they forge new relations across sectoral and disciplinary domains, although commonly building on existing forms of expertise, technologies and infrastructures (such as the European energy grids).
• The occasion for new actors and nodes to be connected to the networks are ofte-times heavily ICT-driven ones, cutting across the physical/digital divide, such as smart meters, wearable sensors, or improved robot sensors and actuators.
These developments take place simultaneously through the EU institutions and in shaping national networks. EU institutions do not by themselves possess the necessary forms of expertise, and so are dependent on external help. In mobilizing expertise from the different member states they simultaneously also integrate the members states into the Brussels 'network of networks' (see Barry 2001). On their side, expert networks from within the member states can achieve competitive advantages and increased standing within their home countries by uniting with similar networks across state lines: with other member states and with Brussels/the EU. Hence, national authority is used for networking abroad; international relations and connections in Brussels are used for enhancing authority and policy relevance at home.
A good example of such a network is the newly formed Public-Private Partnership (PPP) of robotics. It mainly consists of industry, academic research and policy makers, but it increasingly also involves lawyers, ethicists, social scientists and public relations workers charged with the task of paving the way for a new generation of robots. One could say that very little unites many of these expert networks , apart from their dedication to realising the vision of autonomous robots for specific societal purposes. In the case studied by Epinet this is a a prominent role within the EU agenda addressing demographic problems by developing autonomous robots for care and companionship.
The main distinguishing trait of a technoepistemic network is exactly this dedication, across its different constituent networks, to realise the technoscientific innovation in close conjunction with one or more specified societal challenge. Hence, whereas great diversity exists between the actors within the broader network (such as law, politics, science and industry), the sine qua non of the technoepistemic network is the commitment to the realisation and making of a specific technoscientific application coupled with one or more clearly identified societal challenges.
As a counter-example we mention the fourth case study of Epinet, In-Vitro Meat (IVM), a group of networks trying to establish itself around the making of IVM, legitimising this innovation domain in relation to differing rationales, societal and environmental goals: from countering climate change to improving population and individual health, to alleviating the suffering of animals, to the need to feed the world. In this case we do not identify a set of stabilised innovation and policy goals, and the only common denominator across the networks’ constituent parts remains the dedication to realise IVM as a biotechnology. Therefore, the IVM network does not qualify as a genuine technoepistemic network, but rather as a set of more loosely connected actors trying to establish themselves as such (and as of yet not succeeding). For instance, there is no technology platform to stabilise their relation with EU policy makers, and they have no high-level representatives within the Commission or elsewhere to bring their case forward; they struggle with recognition among their scientific peers.
The technoepistemic networks hold out the promise of re-making societal relations, mainly centred around wide-spread and pervasive technoscientific infrastructures, many of which are ICT-driven. They come along with specific policies and institutional dynamics, seen as necessary for their realisation. Some of these dynamics have been outlined above. It is important to understand them, and the institutions and networks developing around them, in order to appreciate the prospects (limitations and possibilities) for actual integration of technology assessments and the requirements of Responsible Research and Innovation, into the large-scale programmes, the agenda setting and the innovation networks of the EU.

Assessments: integration into technoepistemic networks
Historically oriented work conducted for WP1 is a treatise on how 'integrated assessments' of the impacts of science and technology on social and environmental relations have changed over-time. Classical TA was predicated on notions of a 'rational scientific process', and targeted towards parliamentarians. Later efforts have also included efforts towards integrating with technoscientific innovations themselves, e.g. constructive technology assessments, ELSA research and so-called integrated projects. Recent and ongoing efforts towards Responsible Research and Innovation (RRI) continue such developments, but are also actively transforming and expanding them (further strengthening the European level, also targeting policy agendas, towards possible futures, cf. von Schomberg 2012).
Many of these characteristics follow from, and indeed mirror, the developments within the technoepistemic networks themselves: First, the technoscientific networks make assessments about societal needs, as when it is expected that behavioural change can be induced by introducing wearable sensors and mobile phone applications, or when autonomous robots are called upon to address ageing and negative demographics. This is a reflection of developments in which increasingly societal, political and environmental challenges (and dedications towards change) are delegated to technoscientific research and innovation (Rommetveit and Wynne, forthcoming). This tendency actively promotes and to some extent achieves a blurring of boundaries between the domains of research and innovation, law, politics, industry, democratic institutions, and everyday lives. Similarly, the boundaries blur between innovators, policy makers and assessors. Societal, ethical and legal assessments are themselves becoming 'integrated' into the technoepistemic networks, to the extent that they, and their methods, may become hard to distinguish from the work of technoscientific innovators. For instance, we observe the influx of 'ethical entrepreneurs' (Rip 2009). Assessors operate as members of larger teams, 'integrated ELSA', and large-scale interdisciplinary RRI projects: they do not stand outside the network whose practice and products they assess (What Arie Rip, 2006, terms 'governance in complexity' rather than 'governance of complexity').
Examples of such increasing integration and blurring of boundaries are clearly indicated in tasks such as: the hard-coding of ethical and legal principles (dignity, privacy, data protection, but also morals) into technological infrastructures, and so-called integrated projects, where RRI or ELSA actors operate as members of the larger networks. Frequently and increasingly such efforts go hand-in-hand with design-oriented approaches and “value-sensitive design”. And, in the regulatory area we see increased emphasis on scientific risk management and assessments, again performing a blurring of boundaries, or 'breaking down of silos', between technoscientific innovators, policy makers and assessors.
Finally, on the boundaries of such technoepistemic networks we observe more loosely tied networks of innovators, sometimes making up distinct technoscientific publics. Examples here include DIY biology, maker and hacker movements, consumer organisations, networks dedicated to the promotion of open culture and to the digital commons. In WP3 we have observed how citizens and users are experimenting with taking management of health and lifestyle data into their own hands, experimenting with wearable sensors, hand-held devices, apps and social media. To varying degrees such alternative innovation networks may integrate with and influence the actions of technology developers and policy makers (Gunnarsdóttir et al. 2014, Gunnarsdóttir et al. submitted); at other times they play against them, opposing the goals and/or means by which the predominant forces of the technoepistemic networks operate, and organising around alternative visions (see for instance Levidow et al. 2013).

Changes of scope, scale and social relations
One practical result of the activities of technoepistemic networks is the occurrence of more and more issues, challenges and policies at specifically European levels. Hence, they perform a change of scale and ambition, as well as a step change towards (generally) more competition-driven and market-oriented behaviours. While such developments are long in the coming, their impacts are becoming, as of recently, more pronounced and visible. In the field of technology assessment and governance, we observe how national actors have been dismantled (Denmark, Flandern ), with the previous leader of the Danish Board of technology talking about a tendency towards the “competition state” . This is one in which policy-makers and technology assessors increasingly compete for attention and position through the EU institutions, upon which they become increasingly dependent. To paraphrase Andrew Barry, the network of the networks is becoming stronger.

Similarly, quite recent studies of governance of science and technology in Europe (Hagendijk and Irwin 2006) remained focused on the different national regulatory cultures (cf. Jasanoff 2004). Today, however, one may just as well analyse main differences in governance of science and technology in relation to the different waves of technologies, such as bio-, nano-, and various ICT-driven game-changers (Internet of Things, Big Data, next generation robotics, and so on.). Each comes along with new ways of governance, in many cases cutting across national cultures (a similar argument is made for scientific advisory bodies by Bijker et al. 2009, 43), working in more horizontal ways across state lines. We observe how the technoepistenmic networks perform a kind of infra-governmentality as they increasingly work horizontally to induce and produce new modes of governance. The challenge for member states, individual and professional communities alike, is that of latching onto developments, by making the agendas their own, and by actively working to promote and fulfil the promises of the innovation agendas .

To summarise: in observing and commenting upon recent developments in innovation, policy and governance, it is not our intention to idealise or display them as straightforward and streamlined tendencies. Indeed, competition-driven behaviours coupled with strong beliefs in technologies and markets as the main drivers of integration may backfire as the on-going negotiations over the Eurozone clearly demonstrate. We frequently observe a disconnect between predominant visions and the realities on the ground, for example, implementing visions and agendas in practice by certain actors, or the prosperity and well-being citizens, users and communities are intended to gain from technoscientific innovations. Indeed, some of the observations from Epinet pertain to pervasive forms of 'quasi-integration', in which only a limited set of actors and networks are actually invited to the driver seat of techno-political innovation, while others among the European citizenry are expected to simply follow suit. The premises on which new social realities, communities and publics are meant to take shape, are largely left out by key policy initiatives like the Horizon 2020 agenda, or they come into the picture too late.

The lack of adequate public institutions may by itself result in problems and conflicts as technologies become implemented, and as actors are supposed to collaborate across national and professional boundaries without the support to achieve sufficient degrees of understanding and lines of good communication. This was one main finding from our investigations into networks devoted to smart grids and energy transition across Europe. Several participants in our workshops underlined how there is no lack of technological solutions, however, an almost complete lack of societal and political institutions to coordinate efforts between countries (Van Der Sluis et al. 2014). Hence, technological innovation without corresponding political and legal institutions, may end up as poor investment, since there is no way in which society at large, publics, users and local communities, may effectively and realistically connect and interact with each other in order to address the shared challenges faced by Europeans, in this case energy transition. In that sense, the requirements and precepts of RRI (see introductory section) remain an outstanding unresolved challenge.

Science, law and assessments come under pressure
The above developments pose decisive challenges for technology assessors and for those aiming to achieve more responsible research and innovation, including also people working in governance, policy-making and publics seeking to engage with research and innovation. Analyses from history (Shapin and Schaeffer 1985), sociology (Latour 1993) and philosophy (Toulmin 1990) point to how western societies have, in their search to legitimate their actions, relied on separations between law and science, nature and politics. Granted, these relations were always more complex in real life than in their idealised official versions. But these idealisations also performed real functions, exactly by imposing some checks and balances between the different domains.

Within the activities of the technoepistemic networks, and through the breaking down of silos in a number of areas, these separations and modes of legitimization are no longer as strong as they once seemed. In their place we observe visions and promises to reform societies along more prosperous, sustainable lines of technoscientific research, innovation, living and producing. We are also starting to gleam the emergence of new societal relations in certain areas shaping around these networks, such as increased reliance on privatized health care provisions, pressures towards “responsibilisation” of patients, citizens and consumers (of energy, health care, and others); we also see communities and individuals taking things into their own hand through self-care groups, the quantified self movement, or new micro-grid communities for local energy production.

Main questions still remain about the sources of legitimacy that will in the end serve to prop up such new-emerging socio-technical orders. But neither is it the case that all previous structures evaporate: science and law are still crucially relied upon for the making and running of the networks, and new relations are forged with technology assessors and publics, for instance through public engagement activities. The new relations in which technology assessors find themselves are indicative of new roles for expertise, operating both within and across previous structures. What Epinet has first and foremost observed are the ways in which new and old forms of expertise are working to find their place within the new relations taking shape around the technoepistemic networks. We observe how lawyers and scientists assert themselves within new transdisciplinary configurations and take on new tasks; we observe how they seek out, and to some extent achieve, collaborations in new interdisciplinary teams; we observe how they come under pressure from policy makers and industries and their agendas to increase competitiveness and innovation.

Within such new constellations, assessors of science, technology and society relations, are also struggling to find their place. Main tensions relate to the need to remain both policy relevant, to be listened to and to get a place at the table, whereas at the same time retaining academic standing and the crucial independence (not to be confused with neutrality) from the very forces one is trying to understand and influence. Hence, the prospects of achieving more Responsible Research and Innovation are promising. And, the project of pooling the resources of the different epistemic communities of assessors may indeed be necessary, if some real influence is to be achieved within the ranks of networks that are strongly driven by industry, politics and powerful technologies. At the same time, these very developments raise doubts and criticisms from within the ranks of the concerned disciplines, such as TA, STS and ELSA.

A final corollary of this analysis, to which Epinet has devoted Work Package 2, concerns the pressure to achieve practically in realising the broader goals of responsible and sustainable research and innovation. Interdisciplinarity is at best the partial outcome of such work, far from being the default orientation from which assessors of science and technology, innovators and other relevant parties start their collaborations. This becomes visible in and through our findings. For example, interdisciplinary collaborations are indeed doable, however, in need of sufficient time, well-argued occasion to come together and a level of trust which can only be established in and through communication and togetherness. Even as urgency is used to push new relations across knowledge domains and experiences, one main outcome of our research is that disciplines respect their own disciplinary commitments in order to retain legitimacy within multi-disciplinary contexts. If assessors have to sacrifice basic presuppositions and commitments, they lose hard-earned authority. Correspondingly, other practitioners lose the possibility to hold them accountable according to publically available validity claims. The result of that would be a deterioration of the authority of assessments-for-policy.

EPINET and RRI
Even though EPINET is not formally responding to a call that explicitly refers to the concept of Responsible Research and Innovation (RRI), thinking about RRI has been part of the project since the beginning. There are important structural and thematic parallels between EPINET research and RRI (further outlined below), and there are also historical convergences of their respective thematics. As outlined by Rene von Schomberg (2012), a tendency over the last years has been for different assessment practices to move closer together. These include impact assessment (environmental, social, economic) technology assessments and ELSI/ELSA research and so-called integrated ELSA / integrated research projects. RRI, says von Schomberg, fits right into this picture: indeed, it can now be regarded as the main effort towards pooling the resources of such assessment practices.

There are differences in the ways in which the RRI concept is constructed, ie. between the EU (von Schomberg 2012b, EC 2012), the UK (Owen 2012, Stilgoe et al. 2012) and the US (Guston 2013), or in the ways in which it is tentatively tried out in (some) Asian countries (ie. Japan). One can also observe discrepancies or at least tensions or nuances between the definition(s) of RRI as a concept and the practices that are gradually emerging from efforts to operationalise and implement policy decisions on RRI. Specifically, a certain tension can be observed within the EU between the rather ambitious definitions of RRI and the attempt to operationalise RRI in terms of the so-called six (or five) “keys”.

However, in terms of general concept and underlying rationale there are several common characteristics. We take the following characteristics of RRI (mainly taken from the EU and UK contexts) to be especially relevant to EPINET research and recommendations:

First, the aim and ambition of RRI is that the resources of different assessment practices increasingly come together, become integrated, also including practices and principles established in ELSA research (Fisher et al. 2006). Hence, there is a sense in which RRI represents a new framework or paradigm (Owen 2015) on behalf of these research fields, and this happens in parallel with efforts towards the pooling of resources from across Europe. This is also the case with technoscientific networks studied by EPINET. As outlined in Epinet WP1 this “paradigm shift” is mainly the result of how the RRI community responds to and mirrors developments inside the main innovation and policy domains with which they interact.

Similar to this is an emphasis on specific qualities expected as outcomes when different groups come together, ie. reflexivity, responsiveness, anticipation and deliberation (von Schomberg 2011, 2012, Owen 2015, Stilgoe et al. 2012, see also Guston 2013, RRI Tools 2014). This emphasis is not unique to RRI thinking, but could be said to incorporate collective processes of learning generated by a great number of actors on the science/society interfaces over the last 40 or so years (see for instance, Felt, Wynne et al. 2007). However, the RRI discourse configures these qualities in specific ways and mobilises them for purposes of remaking assessments, governance and institutions, frequently conceived as a kind of capacity-building (Guston 2013, RRI Tools 2014).

Next, there is in RRI a general turn towards possible desirable (or, undesirable) futures (ibid.), especially targeted towards bringing broader publics into visions of the public goods to be achieved through research and innovation. Also this can be said to be a reflection of developments in the fields with which assessors interact, and especially the strengthened drive towards the making of possible futures. This tendency is most clearly expressed in the increasing numbers of foresight activities, vision assessments, the use of scenarios and similar activities highlighting social desirability, acceptability and robustness. This turn towards futures places the media through which such visions and scenarios can be constituted and communicated at the heart of assessment processes. This brings forms of mediatisation into RRI and demands both a clear consideration of how the material and symbolic aspects of these forms might also come into assessment, as well as a reflection on RRI as producing its own media forms.

Fourthly, and partly following from the above, RRI discourse and the EPINET project share the commitment to “wicked problems” and “messy governance”. This attitude was well expressed by Jack Stilgoe: “[…] if the credit crisis has taught us anything, it is that efforts to govern complex systems should not be deterred by complexity.” (Stilgoe 2013, p. xiii). This is indeed one of the points where the more conceptual work may find itself in tension with practices within R&I institutions and their funding bodies.

In sum, RRI amounts to attempts at new modes of governance of research and innovation that aim if not higher, surely differently, than many current institutional arrangements. The final common trait of RRI is accordingly the emphasis on possible institutional change as part of assessments. Recent thinking about RRI does not regard the research agendas of the EC, the national research councils or other funding and programming agencies, as beyond their scope and ambition. As highlighted by Richard Owen (2015), the European Commission itself is not exempt from such possible assessments, considerations and recommendations.

Our analyses take these commonalities as their starting point. We base ourselves in a sympathetic, although also critical, reading of RRI. They are intended as contributions (in our view) to a necessary deepening of RRI as a theoretical, practical and institutional project. As assessment practices become expanded in time (ie. increasingly towards possible futures), in place (European rather than national level), across social relations (ie. intensification of interdisciplinary and cross-sectorial collaborations) and across forms of media communication, there is a risk that crucial characteristics of assessment practices are lost. Some of these characteristics pertain to the relative independence of the knowledge bases through which assessors make their claims, and on the basis of which they can be held publicly accountable. Deepening and not only expanding is crucial if RRI is to retain legitimacy, authority and relative independence. But our reflections are also convergent with the realisation that RRI is a social innovation still to be stabilised (Rip 2014), and now in the process of moving from its visionary phase towards more practical implementations (Owen 2015), including in projects dedicated to its realisation (for an overview, see RRI Tools 2014). It is therefore crucial to consider what happens to assessment practices as they become integrated (or not): with each other, with innovators, policy makers and publics. This is a particularly acute issue in the implementation of RRI as a cross-cutting principle in Horizon 2020 .

EPINET could be said to have empirically explored important parts of the RRI program as it turns towards practical implementation: it has conceived of “integration of assessments” as practical achievements, as possible results of assessors, policy makers, researchers and innovators coming together. However, EPINET was not an empirical study of RRI qua policy object with its own performativity and process of construction and consolidation; this means that our report will only to a lesser degree comment directly on the ongoing policy developments in, e.g. the European Commission. Rather, this report documents some main learnings from this process of exploring empirically “integration in practice”, and with the relevant and concerned practices, and mainly from the perspective of assessors (of various kinds). It thus aims at being relevant to RRI as scholarship, as practice, and as a project directed towards institutional innovations. The text is structured according to three fields of tension, observed empirically by Epinet. These tensions, or frictions, are also relevant to the expansion and realisation of RRI, as “integration” is sought across previous boundaries: between assessments and research and innovation networks, between disciplines, and between law, technoscience and assessments. Our argument is that some of these frictions be articulated and actively used as occasions for further developments of RRI.
Implications & learning 1: assessments entering into collaborations with technoepistemic networks
In our WP1 summary we have highlighted how strong currents within EU governance are working towards “cross-cutting” research and innovation as ways of reforming societies, improving competitiveness and meeting societal challenges. Within the discourse on RRI such cross-cutting activities are also envisioned, in the form of integration into the activities of research and innovation networks. Another form of cross-cutting activity takes place within the (envisioned) RRI community itself, through increasing expectations towards interdisciplinary collaborations between previously distinct epistemic communities. Concepts of community and collaboration invoke forms of mediation. Community, communion and communication and the relations between these terms, whilst not quite explicitly so, are evidently about communicative practices around which an experience of community can be understood. Collaboration demands recognition of different communities whilst invoking the sense of labouring together to the same purpose. All such communication and mediation requires substantial work and a friction between difference and connection. Examples of such communities include ELSI/ELSA researchers, ethicists, impact assessors, science and technology studies scholars, technology assessors, environmental impact assessors, and others (see von Schomberg 2012, Stilgoe et al. 2012).

What are the costs and challenges, and the deeper conditions for such collaborations (and integrations) to successfully take place? As we have foregrounded, one cost and challenge is that of how to do the work of communication, whilst also reflecting on the power-effects of such work, and at the same time analysing the media productions of innovation projects. The deeper condition required in order to avoid being caught up in the circulation of surface level discussions (such as those experienced by those of multiple linguistic origins using a common but unfamiliar second language) is to pay attention to the mediatisation of technology, as well as the technologisation of media. It is crucial to avoid assessing the vision as the technology (and vice versa) to avoid the lure of the spectacle, or at least recognising that this is at stake, whilst also being mindful of ones own resistances and commitments.

Further, whereas it is projected that interdisciplinarity and action across domains are desirable, the epistemic communities called upon to do the job may pose requirements of their own, such as resistances grounded in scholarly commitments that cannot easily be accommodated within policy goals towards integration. We see this, for instance, in the reluctance from (parts of) the STS , TA and ELSI/ELSA communities to direct their scholarly contributions towards such broad-scaled policy-oriented projects as proposed in discourses on RRI and integrated assessments. For instance, Brian Wynne (2007) warns against STS becoming “dazzled by the mirage of influence” that has over the latter years been presented to the more policy-oriented parts of the STS community. Specifically, he argues that STS should not take its main criteria of quality from the policy context. And, focusing more on contexts of research and innovation, Alfred Nordmann (2010) has compared technoscientific research to a crime scene, and warned against technology assessment (likened to forensics science) to become part of the promise- and wishing- machinery of the very technsociences it is supposed to assess.

There are many examples of well-considered positions within TA and STS that do not see the same objections as do Nordmann and Wynne, and reflect differently on the possibilities for integration into processes of research, innovation, and governance. Let us for a moment interpret these positions as to construe a clear contrast between them and that of Nordmann/Wynne. We do this to pursue clarity, hopefully not at too much of an expense of nuance and realism. As examples, we mention the program of anticipatory governance (Sarewitz and Guston 2001, Guston 2013, Liebert and Schmidt 2010), approaches closer to innovation studies and evolutionary economics (Etkowitz and Leyersdorf 1997), including niche management (Schoot and Geels 2008), and also constructive TA (Schot and Rip 1997). Integrated ELSA devoted to “mid-stream modulation” (Fisher et al. 2006) also fits with such approaches. All of these are relatively optimistic on behalf of the possibilities for assessors to identify signals and early warnings about possible future states that can be used to steer the development of research and innovation towards more socially or ecologically desirable outcomes. Epinet has approached such different commitments in terms of “styles of thought” (Deliverable 1.1) and in terms of a similar concept of “Epigrams” (Van Dijk and Gunnarsdottír 2014, Gunnarsdottír and Van Dijk submitted, Rommetveit et al. in prep.). We refer the reader to these documents for a further analysis of these concepts.
Our intention at this point is only to point out that there are different epistemic and normative commitments at work in assessments. These commitments cannot be easily overcome or done away with, without also doing away with the authority and validity claims of the assessment practices themselves. It follows that differences and controversies as just outlined, and the commitments reflected by them, should be articulated as conditions of possibility for TA/RRIELSI/ELSA, as they seek (some kind of) integration into research and innovation networks. The Epinet account of different epistemic stances point to deeper commitments also on ontological levels: of what do innovation environments really consist, -are they systems that can be steered towards desirable goals, or are they more like distributed practices and networks? The different commitments among assessment practitioners also tend to reflect, although not in any deterministic sense, on the ways in which practitioners imagine the possibilities and conditions for “integration”, that is, for some kind of collaboration (or not) with technoscientific innovators and policy makers. So far, however, questions about epistemic and normative differences are not much highlighted in RRI discourse. But the problem arises in practice: how to accommodate different commitments (normative and epistemic) that arise in the process of integrating different approaches towards common goals. That such differences arise should be expected in collaborations across knowledge sectors and disciplines, as we have discovered and explored in some detail in EPINET work (ref. all the workshop reports, D2.2 D1.2 etc.). Further, although we do not necessarily agree with the concrete alternatives put forward by Nordmann and Wynne, we nevertheless side with them over one issue: there is a need for relative independence for the disciplines that enter into the constitution of RRI. As articulated by Folk (above), responsibility presupposes checks and balances, and plurality of epistemic competencies and normative commitments. We do not believe many promoters of RRI would disagree with this; in spite of the contrast drawn above, we would specifically expect Guston and von Schomberg to agree. We do think, however, the point has to be made with greater force, and recognized as a basic condition for the kind of work carried out by assessors. In the next section we will discuss interdisciplinarity as one point of entrance for the analysis of this condition.
Implications & learning 2: Interdisciplinarity
Let us now continue with a discussion of some of the results and conditions that emerge as technology assessment are compelled to embrace the multiplicity of relevant concerns and analytical perspectives. Within such conditions – and this was the point of departure for EPINET –emerges the need to combine and “integrate” single assessment disciplines into multi- trans- or interdisciplinary teams. The implication of main policy agendas, including that of RRI, is that single disciplines are not up to the tasks of grasping the cross-cutting and cross-sectoral activities of researchers and innovators, and of societal challenges implied. Mobilizing more disciplines and more perspectives, it is assumed, will provide a richer set of assessments, and more adequate policy responses.
Interdisciplinarity is therefore hailed as a solution to the problems posed by the novel character of present-day research and innovation, including the societal challenges to which they allegedly respond. In addition to integrations as dealt with in the previous section, interdisciplinarity may be invoked in the following two contexts: (1) the use of different assessment disciplines or methodologies within advisory bodies, frequently referred to as science-for-policy; (2) interdisciplinary teams working in close relation with researchers and innovators, as for instance in integrated ELSA projects. Such modalities of interdisciplinarity are routinely also invoked in main RRI texts. http://ec.europa.eu/programmes/horizon2020/en/h2020-section/responsible-research-innovationhttp://ec.europa.eu/research/participants/data/ref/h2020/wp/2014_2015/main/h2020-wp1415-swfs_en.pdf - 14
However, whilst modalities of interdisciplinarity are invoked, issues of process and communicational practice are not. The work of achieving shared understandings of the questions at stake in multi-modal projects demands robust communication strategies. Whilst the media texts produced through innovation projects can provide things to gather round, the different dispositions and different viewing positions around such objects require discussion and consensus building, which might be able to recognize plurality, but must be able to articulate some shared recognition of the object. Such shared recognition requires an added layer of media inscription, another communicative thing; something produced in the sharing across the collaborators to enable a process of integration (e.g. minutes, notes, images of workshops, manifestos, policy recommendations, recordings, documentary). Projects that have resources at their disposal for the work of interpretation and integration are likely to be more able to incorporate agonistic approaches that retain specificity whilst also exchanging legitimacy effectively. Legitimacy and compelling representations go hand in hand.
A challenge for technology assessors, and especially those oriented towards qualitative methodologies, is the predominance of quantitative approaches promising rapid and actionable knowledge, readily deployable across sectors. Quantitative approaches are much more invested in the work of representing findings as image and spectacle and as such are seductive in terms of rhetorical power. Examples of such approaches include risk assessments (and management), econometrics, polls and surveys (ie. the Eurobarometer), quantitative impact assessments and the use of indicators. Such approaches offer broad sweeps of the fields and issues in question, and are capable of much more rapid action in response to calls for urgent policy-relevant action: numbers and indicators as concise and evocative representations give the impression of being directly translatable into action. This rhetoric of speed and efficacy contrasts with the sometimes painstakingly slow process of other disciplines more oriented towards interpretation, and depending on provisions of contextual understandings and explanations in communication with policy makers and others.
Insofar as disciplines such as ethics, law, knowledge assessments, vision assessments, constructive TA, or STS, are involved, promises of rapid shortcuts are illusive. When it comes to integration of methods and disciplines, and working towards inter- or multidisciplinarity, one-size-fits-all approaches obscure more than they reveal. The challenges of multi-disciplinary collaborations should come as no surprise to RRI promoters and practitioners, and interdisciplinarity has indeed been a topic in TA for several years (see for instance Decker and Grünwald 2001, Decker 2004). Yet, so far the discourse on RRI has not made it into a specific problem, or indeed, basic condition, for research and policy advice. In general, interdisciplinarity seems to be regarded more as a default option, rather than a critical achievement that can only come about at the end of a laborious process of communication, mutual adjustments, knowledge exchanges and learning.
In EPINET, as in previous writings on interdisciplinary TA (Decker and Grünwald 2001, Decker and Fleischer 2010), a problem- or issue-oriented approach was chosen as a way of structuring (multi- and inter-)disciplinary relations and interrelations (see also Dewey 1927, Marres 2007, Rommetveit and Wynne forthcoming, Rommetveit van Dijk et al., 2014). Such an approach seems required regardless of whether the implied assessment methodologies are qualitative or quantitative, or both. Due to the frequently broad and sweeping visions coupled with fast developments of many innovation fields, there is a need to provide a prior focus and understanding of the issues (societal or technology-induced) to which the different assessments are expected to respond. As stated by Decker and Fleischer (2010, 119) this points to “the definition of the problem as the central element of transdisciplinary research”. This process of definition is in itself fundamentally a communicative practice of producing representations, establishing shared intelligibility and framings. This requires the sharing of language, images and other forms of representation and definition. The capacity to engage is this, and the framing of the problem, structures other significant problems among the project partners, such as: choices of methods, when and where to seek intervention, who to include in the broader assessment, and so on.
We do not proclaim our problem- and issue-oriented approach to be the only option. However, we would like to point to two broad lessons to derived from that.
The first has already been stated, and seems almost too obvious for mentioning, especially to those with experience with interdisciplinarity: interdisciplinary integration is an outcome to be achieved, and not the default position. In EPINET, even as the researchers tried taking these matters into consideration, we identified shortcomings and problems relating to organization and structure of projects. Some of these may indeed be specific to EPINET and the specific solutions chosen by the project. However, several issues are procedural across this kind of research project, these relate to lack of continuity (when the project ends, research ends as well, and ”interdisciplinarity” will have to be re-established in another setting, if at all). There are limitations in communication (which is sidelined as separate to the main activity) and a lack of face-to-face interaction required for common understandings to arise and thrive, also specific to the case at hand. This points to the friction that a failure to recognize the extent of communication at the heart of such a project generates. Finally, there are shortcomings in, and challenges for, learning. In many cases the most valuable outcomes do not necessarily relate to a fusion of disciplinary horizons, but rather to one field of study borrowing or learning something new from another. Where there is learning, it is hard to qualify it as trans- or multi-disciplinary. In many cases it is better to accept such dynamics as the normal conditions, the natural friction in learning, communication and for (some) integration to take place, rather than differences and complications to remedy. In short, we regard epistemic and normative pluralism as a resource, and not an obstacle to be overcome, but this too has to be represented as such.
The second pertains to the choice of “the problem” to be addressed. The assessment team should be in a position to choose and define the research (and policy) problem with considerable independence. Again, this speaks in favour of distance and relative disconnect from policy makers, researchers and innovators. Cultivating a knowledge base for assessments and for RRI will have to be aimed at problem-selection in accordance with the validity claims and commitments of assessors and RRI practitioners. This may have become even more important with the implementation of Horizon 2020 . There, something akin to a problem-oriented approach is pursued by the orientation towards “societal challenges”. But assessment practitioners have to address the real problems faced by European societies, while they cannot for that reason be expected to simply adopt the problem frames provided by researchers, innovators or policy makers. These should themselves be part of the object of assessment. In our memo on cross-cutting challenges for EPINET it was therefore stated that “there is a general need to get a better grasp of the public character of the innovation/policy objects in question, preferably as far upstream in innovation trajectories as possible. This means that we, as analysts, do not simply take over or accept the initial framings provided, for instance by industry or policy visions, but critically aim to assess the character of innovation/policy objects: we crucially also include their democratic potential and their fitness-for-purpose in addressing main societal challenges” (Rommetveit, van Dijk et al. 2014). However, the challenge here is that in many instances of emerging technology assessment there is no innovation object except the visions that are instantiated in media forms (e.g. in the case of smart energy grids), and technology assessment lacks a technological object, becoming an assessment of media forms in which visions are materialized.
Coming finally to the question of what actually happens when people are brought together across disciplines and, not to mention, geographical distances, the first thing to encounter are the procedural conditions that should be expected in carrying out the kinds of studies we have observed. We do not present here a complete index of procedural conditions to keep in mind but a select list will give an idea of the extent of potential shortcomings and complications. What we are listing here is not a complete index of comments and suggestions we heard but although many of these items seem minute and stating the obvious, they should absolutely not be trivialized and waived off as a matter of just getting a job done. They point to the importance of caring for organizational, professional and inter-personal conditions in carrying out teamwork and leading it. What we see is evidence of a lack of clarity on what integration stands for and what can be expected from bringing together expertise across disciplines professions and borders like these case studies have done. In short, it is in the actual execution of teamwork and leadership which has been set up to achieve idealistic goals of interdisciplinarity and integration, that the know-how can come up short in building and sustaining momentum, precisely because trivial details of care are missed.

First on this list are the limits to participation. Personal and inter-personal, professional and institutional dynamics will test those limits. Contracts are time-limited and come to an end, funds run out, family members die, accidents happen and people are unavailable for one or another reason or somehow not cued in.

Secondly we mention barriers to communication, some of which are relatively easy to manage with innovative uses of ICTs and provisions to met regularly face-to-face. There are communication barriers when team members are rarely co-present and given the communication hurdles that need overcoming to achieve some degree of disciplinary approximation, we refer to those barriers as a stress-text of resilience. Communication, and being in-communication, is key to the necessary dynamism in keeping a shared study environment alive and going forward.

Thirdly, myths of interdisciplinarity are exposed in the many ways in which a journey of working together tests the limits of learning, knowledge creation and sharing. We will discuss this matter in more detail in the next section, but these conditions of the journey also provide pathways to sort things out, to make sense of a case and lead the work forward.

What we call epigrams, are practical models for ordering pieces of knowledge and modes of production into constellations of relationships They are indications of epistemic power and their identification can serve reflexively the need to install epistemic checks and balances.
Within the case studies, the making and use of epigrams serves us to explore emerging networks of innovation and assessment, but also how actors situated in or around these networks are reflexively trying to make sense of epistemic relations. The team leaders (and sometimes assertive team members) come up with their own illustrations, diagrams and other schemas for what the innovation networks are, how to conceptualize them and how to integrate assessment efforts, even unify them. We argue that the making and using of epigrams is a display of epistemic power and the need for explication and confrontation. We take here four examples to address some of their key characteristics and functions in relation to the development of the corresponding cases.

The key lesson to take from the use of epigrams concerns the evidence they give of how a study is proceeding. They are indicative of the leadership and direction in doing this work, and they are also indicative of various styles of leadership and choice of direction. Taken together, we heuristically distinguish three ways in which epigrams can be characterized in terms of the direction they give. System-based orientations in epigrams are biased towards preset analytic criteria and frameworks. In epigram 1, the association is with complex multi-scale, multi-layered systems but also in other system-based orientations we observe concerns over how to integrate different assessments of such systems or perform causal-chain assessments of them. Network-based orientations in epigrams lean on issues, actors, practices, performance, mediation and other factors that get taken into account in a cartography of connections and disconnects in practice, which also produces novel leads to take forward. In epigram 2, these leads point to the need to establish new relationships and plug knowledge gaps. Proceedings-based orientations in epigrams seek adequacy and quality in preparing for or reporting on legal and other proceedings. Primary elements in them concern process relations, purpose-specificity, participation and clarity in epistemic quality checks and decision-making. In epigram 3, the elements are preparatory for proceedings (lesson in law), whereas in epigram 4, they dictate the reporting of proceedings (lesson in innovation practices).

That said, key learning to take away from our observations is how selectively and pragmatically experts learn in approximating other disciplines and in distancing themselves. There remains a sense of unease with unfamiliar scholarly and methodological terrains which may require proactive mitigation and mediation, however, we also observe that a sense of integration to take away from case work like this, should not have to be the priority. Rather, this work ought to enable thinking about the material at hand in multi-dimensional ways, while aiming for novelty in knowledge creation. The learning here is perhaps stating the obvious that if we confront the fact that the making of contemporary technologies navigates multiple sites, then gaining a good sense of the nature and extent of this multiplicity is better facilitated by engaging with people across different disciplines, occupations and experiences. But, as one member put it, “part of this is the fact that I have realized that it is not so easy to ...err ...really work together with different ...err disciplines or different approaches, no, because at the end of the day I still understand ...err ...my methodology better than the others and I'm ...I still do what I do”. We can argue in this respect, as Stengers does, that there are good reasons to adhere to ones disciplinary home base as a source of legitimacy and authority but also perhaps that, in this process of integration, we become much clearer about what our own disciplinary base has to offer into the mix.

Interdisciplinarity, except in the most basic sense of collaborating, is not likely to be achieved by just putting different disciplines together to work on a project. It should be considered an achievement of hard work and, as we observe, such an outcome is more likely to emerge on a much smaller scale than is expected from a large-scale 'interdisciplinary' project. Disciplines tend to collaborate one-on-one or in very small-scale teamwork and we observe that such occurrences are taking place in the absence of formal attempts at integration and often with surprising result. They typically happen in the course of exploring a common assessment issue that produces an ontological entanglement while encouraging a quest for belonging. There are many such small-scale entanglements to observe within the case teams.

At the heart of what we observe is the willingness (or not) to overcome disciplinary barriers, as much as that is indeed possible, and how that then works with more explicit concerns about integration as a cross-cutting issue in the Horizon 2020 program. From what we have learned to- date, we suggest that each case study is an exploration, a laboratory of assessment practices aiming to grasp their ecology in order to tie the question of what constitutes a practice to the question of its co-existence in an environment of other practices. Each case is approximating and distantiating disciplines, and entangling them in ways in which leads each case to a mode of integration, so to speak. Looking now through the developments within the case studies, we can say that the different modes of doing this work hinge in part on technology-specific issues, in part on sector-specific issues as well as more generic issues. For example, ICT-based innovations which essentially are key enablers in most innovation domains nowadays, constantly call for a distinction between technology specific, societal and generic problems.

Implications 3: the role of Law
In three of the EPINET case studies questions about law have been very prominent, because in thinking about “integration” of different practices and domains of action, law emerges repeatedly as a site for posing fundamental questions. This section deals with implications for law in general; in the next section we provide further examples of law’s role into different forms of assessments by taking a look at Data Protection Impact Assessments.

Law, science and engineering have traditionally been regarded as separate in western traditions (Latour 1993). These separations have been main sources of legitimacy upon which (respectively) research and innovation, and politics would rely. However, through the cases we have studied on privacy and data protection in smart grids and wearable sensors, and the hardcoding of morals and laws into social robots, such basic institutional boundaries become blurred. Has the question been asked about how science and law as (traditionally) separate spheres will (or should) be united through engineering? How are fundamental rights of privacy and data protection going to be hardcoded into information infrastructures? How are risk assessments going to fit into the design of rights and freedoms?

Across several research lines and policy domains we have observed how law enters into research and innovation, but also how in the process it comes under pressure from science, engineering, politics and industry. And, as law is made to enter into the “regulatory mix” (Lessig 1999/2006) of different assessment practices, such as risk assessment, it is forced to share its authority with other disciplines. This might itself be a good thing, or at least an interesting opportunity. However, problems arise since law also increasingly has to base its assumptions and premises upon possible futures as established by researchers, engineers and risk managers (and, to some extent, ethicists and social scientists).

The question arises whether the RRI discourse in its present form has the capacity to respond to such questions. Mirroring the arguments of Nordmann (2010) our answer is mainly negative: the discourse about RRI is largely predicated on notions of prediction and control as taken over from the sciences, whereas not granting much attention to the specifics and requirements posed by law itself. If one reviews the literature on RRI (see for instance all the contributions in Owen et al. 2013), one sees how the Collingridge dilemma is generally accepted as a valid statement of the working conditions and challenges for RRI. Within this universe it becomes a matter mainly of finding the right time for intervention, as in the metaphors of up-stream, mid-stream and downstream (see also Fisher et al. 2006). Now, it is not so much that we disagree with the Collingridge dilemma when taken on its own terms; but it provides for a rather one-sided characterization of all the different things going on in a society, and it specifically ascribes great agency to the technsociences as producers of progress and new societal relations. As such, it seems to preempt the kinds of meanings, questions and public problems that can be brought to the table, and be used for problem-definitions within interdisciplinary teams.

When law is seen in this light, another classical dilemma appears as inescapable, namely the problem of a “legal lag” (Ogburn 1922). According to this notion technoscientific developments move so fast, and are so encompassing complex, that law stands no chance when it comes to keeping up . Indeed, the Collingridge dilemma could be easily made to fit with the legal lag hypothesis: “at early stages of developments, we cannot legislate since we do not know their consequences. By the time consequences settle in society, it is too late for legislation”.
This, however, only appears to be the case as long as we do not appreciate the capacity of law to steer and guide developments, and to stake out new directions for societal and technoscientific innovations on its own terms (Jasanoff 2003, Latour 2013). That is, large parts of the dilemma follow from a chosen prioritization of technoscience as the main site of dynamism and innovation in society, whereas law (and the public institutions it is meant to safeguard) is relegated to a status of backwardness, or even irrelevance (due to its inherent conservatism and positivism). But the problem also relates to a lack of attention and care (Pellizzoni 2004) to the mechanisms and networks that would be needed for applying existing laws and principles to new fields and problems. In EPINET research, we have observed these dynamics in several fields, from robotics to the governance and assessment of data protection and privacy. The preference in practice is for vague and principled statements when it comes to implementations of human rights, autonomy or privacy, whereas the pressure is hard to establish the concrete legal rules and regulations needed to accommodate industry-driven innovation (Rommetveit and van Dijk 2014).
When law lags behind developments, this is also an outcome of specific political and economic priorities making things occur and appear that way. The “law lag” is an innovation frame. It is not a social characteristic carved in stone, as one could be led to believe if one ascribed too much importance to the Collingridge dilemma. To paraphrase Nordmann (2010, see also Schmid and Liebert 2010): the dilemma only occurs as a dilemma to be solved when seen from the perspective of control as provided by science and engineering. This precludes us from asking about the potential uses and roles of law. Within the RRI discourse law occurs mainly as one among others among all the disciplines relevant to RRI. This is also mirrored in some developments where lawyers are increasingly expected to collaborate with risk managers and others, or where they have to build their assumptions about future developments on the visions and projections of scientists and engineers. However, we argue that law should not be a part of this regulatory mix in this way. Instead, one important role would be to mobilize lessons from constitutional law in thinking about the constitutive relations between, the mutual checks and balances upon, and the required relative independence of the different assessment practices or broader publics that enter into the “mix”. If one thinks in these ways, there is no need for law to base its assumptions on science and engineering. This could free up the attention of assessors to pose the (repeated) questions of what kinds of practices and publics should be included in decision making about science, technology and innovation, and the broader societal purposes to which they are directed. Consequences and causality enter into liability and tort schemes, and then as quite technical matters. But these are not exhaustive of the potentials of law.
This could also point towards a median position in the debate between the pro-active and the agnostic stances of STS and TA with regard to innovation policy, by on the one hand focusing on the constitutive relations between practices when STS/TA orient their contributions towards policy in interdisciplinary settings of research projects, but on the other hand recognizing the importance of an independent critical home base to hold the (possibly) more action-oriented policy entrepreneurs accountable. A sensitivity for checks and balances is crucial in these circumstances, with a need for sometimes observing distances and separations between practices (like policy and industry) and at other times establishing new relations with unconnected practices with relevant experience.
Another important role for law could lie in the use of legal concepts as conductors for (orchestrations of) impact assessment processes. In the Epinet research strand on data protection impact assessments, for instance, due process has been put forward as an important principle for shaping the processes of impact assessments, whereas proportionality was proposed as an important principle for ordering assessment questions and lessons learned from other fields of application, such as environmental governance and risk management. Such concepts include purpose specification, legitimacy of purpose, fitness for purpose, and alternatives (van Dijk & Gunnarsdottir, 2014), (van Dijk, Gellert & Rommetveit, submitted).
Conclusions: Epinet and RRI
The concept of RRI is now moving from its visionary phase towards having to face tough questions about implementation and practice. In particular, three practical challenges stand out from the European policy perspective:

1) The diffusion of RRI across Europe into R&I funding and practice, as proposed e.g. by the recent Rome Declaration on RRI.
2) The mainstreaming of RRI across European research funding programmes, notably the various work programmes of Horizon 2020.
3) Promotion and monitoring of RRI through quality criteria and indicators for RRI.

As noted in above, EPINET has studied assessment practices and not assessment policy-making and cannot give much tactical advice for the latter. Still, because EPINET has empirically explored the question of what happens to assessments (of various kinds) as these become implemented in putting practitioners to practice, we believe that our experience provides clues to the way forward with the current challenges. EPINET has focused on the practices and the networks in which assessors will have to work if they are to integrate and come together with other groups. The corollary of the analysis provided in this text is that there is a need for a deepening of practices and networks in which RRI comes into being, if it is going to achieve long-term legitimacy, epistemic and normative authority. This we have explored in relation to some of the basic coordinates with which assessors have to do their work:

They have to deal with different commitments (epistemic and normative). These differences cannot be ignored but must be regarded as fundamental to providing checks and balances, and as part of the working conditions of assessors. Hence, we briefly outlined some main differences (explicated in the use of epigrams ) when it comes to the possibilities for integration into innovation, research and policy. Such epistemic and normative commitments are not mere divergences of opinion about the prospects for RRI; they are rather part and parcel of the deeper commitments that provide different assessment disciplines with their validity and legitimacy. There is a need for fundamental research to unravel what RRI can be, what it can achieve, and how the different approaches it already comprises, serve as repositories for reflection, knowledge creation and public legitimacy.

Similarly, we have seen that interdisciplinarity emerges as a central aim, and is itself a kind of integration or cross-cutting principle. However, it cannot be taken for granted as a default option for research and policy, but must rather be considered the achievement to some degree of well-orchestrated procedures, communication and media production, as well as tough negotiations and learning processes between the involved disciplines. This insight resonates for instance with the view that indicators for promotion and monitoring of RRI should be constructed from a network- and practice-oriented perspective, in which dynamic and living learning processes are more central than the metrics of centrally collected data variables. It also resonates with the view that mainstreaming processes for RRI should take a network and learning-based focus rather than pre-made formulae and schemata.

Finally, we argued that there is an under-developed potential for law to be regarded as more of an autonomous actor. Law should not merely be part of a “regulatory mix”, but should also be seen as potentially a constitutive part of the broader ecologies of practice within which assessments of research and innovation take place.

Taken together these elements point towards the need for a firmer embedding of RRI within broader ecologies of practice, in which mutual checks and balances can be exercised: between different epistemic and normative commitments, between disciplines, and as provided for by firmer legal guarantees.

Potential Impact:
The impacts of the project can best be read through the extensive list of publications, talks and workshops, as listed in the Dissemination report. Also, the inventory of dissemination activities is submitted in the final project report of EPINET. This text serves a slightly different purpose: It aims to provide a concise analysis of what has been achieved and of the challenges of “disseminating results” from a research project such as EPINET. We attach in part B, however, the cumulative list of dissemination activities of EPINET.

The analysis may indeed be brief since communication is a central topic in our 50-page set of policy reports (D8.3) and the submitted 180-page set of final scientific deliverables (D8.6). We have two conclusions:

First, EPINET has successfully communicated results into various networks (academic,
policy-makers, R&I sector). The dissemination reports witness a large and varied
dissemination activity, and as is usual for research, there is a time lag so that there will be a substantial number of published research papers in 2015-2017. Furthermore, our results and our methodological approach, which indeed is a key part of the results of this essentially methodological research project, have been introduced into a number of European institutions and policy processes, including the EC, the services of the European Parliament, and the Council of Europe. Also in this sense EPINET is a success.

Secondly, EPINET results point towards the need to rethink the science-policy-interface.
EPINET has studied technology assessment through the lens of EPIstemic NETworks – that is, knowledge networks. Scientific experts, including those of our own kind, may have the tendency to think that our job is to produce the knowledge, give the advice, and then it is up to policy-making institutions to be interested and grateful and readily implement the knowledge. When this does not happen, scientific experts may tend even to blame policymakers for “not understanding” or paying attention.

Also EPINET has encountered the difficulties in communication at the science-policy
interface, not only as a study object but also as a challenge to our own work. While we can document broad interaction e.g. with the EC, with the European Parliament and the Council of Europe, we cannot boast of an impressive uptake of our ideas (yet). The novel and innovative messages from EPINET, about integration of styles of technology assessment (TA), about integration of TA and research and innovation, and about the consequential need for institutional development, do not fit easily into the day-to-day political agendas of the mentioned institutions and what may be called commonsensical notions of the science-policy interface. This affects the uptake; we have also experienced, and have had to negotiate, the expectancies from notably the EC DG RTD and the STOA about “ready-made”, easily implemented “answers” from our research that can be taken up by policy-makers without any need to change or doubt the current methods and institutional arrangements. To do our job, we therefore have had to find the right balance between providing advice within existing methods and institutional arrangements, and provide advice pointing towards the need of changing them.

There is no easy “quick fix” to this problem. One major point from EPINET, however, is to think and act in terms of a network and process approach. This is why we place emphasis on our ongoing, long-term interactions with R&I networks (such as robotics, wearable sensors, smart grids and in vitro meat R&I communities) and policy-making institutions on the national and European level. Perhaps more important than the one-off final project conference, to which, admittedly, quite few European policy-makers came in spite of massive invitation on our side, are the long-term processes with training, discussion and debate, report writing, etc with the European institutions. The “New Narratives for Innovation” process initiated by Andrea Saltelli at the EC-JRC is a particularly good example. While not a part of the EPINET project as such, this process is a fine example of how one can work in a network- and processoriented fashion. We will over the next years continue to work through that and the other listed processes. In this way, EPINET will have an afterlife and a continued effect – and will continue to set an example of methodological innovation in the field of assessment of societal impacts of new and emerging technologies.

Kjetil Rommetveit, kjetil.rommetveit@uib.no
Roger Strand, roger.strand@uib.no

Centre for the Study of the Sciences and Humanities
P.O. Box 7805
N-5020 Bergen, Norway

Documenti correlati

final1-epinet_final-report_results.pdf

Final Report Summary - EPINET (Integrated Assessment of Societal Impacts of Emerging Science and Technology from within Epistemic Networks)

Documenti correlati

Condividi questa pagina Condividi questa pagina sui social network

Scarica Scarica il contenuto della pagina