In the children’s story Doctor Doolittle, one of the fantastical creatures the good doctor encounters is the pushmi-pullyu, an ungulate with a head on each end that eats with one and converses with the other (so as not to be rude by talking with its mouth full!). In the familiar film version, the pushmi-pullyu is portrayed as a llama, and it can talk and eat from both ends. It is, however, always challenged by movement. Only one head can go forward at a time, thus condemning the other head to trail along backward. Pushmi-pullyu is thus a walking pun, a ‘di-llama’ if you will, suffering the dilemma that the desires of its respective heads to govern its direction can never be reconciled.

The Collingridge dilemma, also known as the dilemma of social control (DSC), describes the desires of humans to control technology and the irreconcilable forces that thwart this goal. Ribeiro et al. (Citation2018) in their discussion paper summarize the two-part challenge posed by the DSC as ‘that of anticipating and controlling the potential consequences of emerging technologies’ (p. 315).Footnote¹ In one temporal dimension, let’s call it upstream, technologies are still formative but we don’t have enough information to direct them toward good social outcomes and away from bad ones, because the implications of the technologies are not yet apparent. In another temporal direction, let’s call it downstream, we have sufficient information about good and bad implications, but the technologies are so deeply embedded in interests and systems that they are much, much harder to change. Thus, the dilemma: We can act ignorantly on malleable technologies, or we can act intelligently on reified technologies. Neither is apt to be effective. We are damned if we do and damned if we don’t.

Ribeiro and co-authors attempt to offer up a ‘dilemma of societal alignment’ (DSA) to complement the Collingridge dilemma. I write ‘attempt’ because despite the broad, informed survey of a great deal of contemporary literature on responsible research and innovation (RRI) performed by the authors, I do not believe that they succeed in making the case that a DSA exists. My response comes in four short parts: First, I attend to the difference in our understanding of what a dilemma is. Second, I describe to what extent I believe that Collingridge’s DSC is a true dilemma and thus proper model for the DSA. Third, I offer up (and critique) one possible version of a dilemma of societal alignment. Fourth, I conclude with reference to other work that supports the idea of societal alignment.

The overall purpose of the discussion paper is to identify the DSC as a model for another dilemma, that of societal alignment, and to frame the DSA as a way of advancing informed conversation about RRI. I am exceptionally sympathetic to this agenda, having used the Collingridge dilemma as a starting point for some of my own work (see, for example, Guston [Citation2014] and below) and having contributed to the RRI agenda for more than a decade (beginning with Guston [Citation2004],Footnote² which takes on the issue of RRI in public institutions like universities). My fundamental issue with the discussion paper, therefore, is that while societal alignment is a difficult and important challenge, a challenge is not the same thing as a dilemma. Far be it for me to dispute the Oxford English Dictionary, but I think its definitions as quoted by Ribeiro and coauthors obscure more than they enlighten. A dilemma is not just a choice between undesirable alternatives. It is rather a choice between two lemmas (or lemmata) – which are mid-level, proven propositions in a logical argument. Thus, they are not merely undesirable, but they are importantly in some logical conflict with each other. It is not simply, ‘would you prefer bad 1 or bad 2?’, but more like, ‘you must choose either impossibility B or impossibility not B.’ That’s a dilemma worth having. And while I believe that we’d be damned for not pursuing societal alignment, I don’t believe we’d be damned if we do. Indeed, I believe that such an admittedly difficult but not illogical pursuit is the only way we’ll get the kind and quality of innovation we need to salvage our current situation.

A second concern about the discussion paper and how it uses the Collingridge dilemma as a model is that even the DSC is not the soundest of all logical dilemmas. The lemmas of information and control are not demonstrated as robustly as one might think, and they are relatively easy to weaken by qualification. While not all of the societal aspects of technologies can be predicted, many or some of them can be foreseen in some ways. Think, for example, of how a fleet of autonomous vehicles would create pressures for somewhat predictable changes to urban form and finance. Thus, the information at early stages of technology development is not entirely absent or unusable. And while reified technologies, embedded in deeply held interests and sprawling sociotechnical systems, maybe be highly resistant, change is still possible. Here think of changes to the automobile with catalytic converters, seat belts, air bags, electric engines and, soon, autonomous guidance. While these changes mostly leave the core of the innovation intact, they are changes of no small consequence and have ideas of real public value behind them.

So even though social control is clearly rendered and logically structured as a dilemma, I tend to think of the two horns of the DSC instead as tendencies or boundary conditions that must be battled against, even if the logical pincers don’t always bite hard or definitively. That some foresight is ignored or some changes are hard-fought merely demonstrates that the really interesting action may be in the fuzzy politics between the horns of the dilemma.

Supporting this interpretation is also the model Collingridge elucidated that is related not to policy for science but to science in policy. With this model, perhaps a dilemma but not patently so, he argues that science does not influence political decision making because, under conditions of political consensus, no science is necessary to drive decision making (the over-determined case); under conditions of political dispute, evidence from multiple perspectives is mobilized and frequently deconstructed and thus ineffective (the under-determined case). Here, as in the DSC, the description of the logic and dynamics of science in policy is canny, but the reality lies in the fuzzy space between the logical extremities. An article by Genus and Stirling (Citation2018) – cited by Ribeiro et al. but not incorporated into their analysis, presumably because of how recent it is – explores how the extent of Collingridge’s work in fact attends to this fuzzy space and is thus much more subtly political than the caricatures of the DSC and the overdetermined/underdetermined model would indicate.

So far I’ve argued that my understanding of a dilemma is different than that of Ribeiro and coauthors, and that the DCS, while well-structured as a dilemma, is a somewhat soft one. Now I will explore what the discussion paper really offers in description of its DSA and, alas, it is minimal. In contrast to what they characterize as the technical and temporal elements of the DSC, ‘the challenge [in the DSA] is one of engaging multiple and often diverse publics, framing societal needs and aligning the objectives and configurations of science, technology and innovation for meeting those needs’ (p. 324). In their words again, the DSA is less concerned with consequences ‘as it is with the challenges of shaping science, technology and innovation to ensure that their development processes are aligned with the values and needs of different publics’ (p. 318).Footnote³

In my reading of the discussion paper, I cannot find a clearer statement of the DSA, and thus the authors never quite approach the idea of a built-in conflict or an opposition of two necessary functions in the dilemma of societal alignment. Challenge, yes, and an important one, but dilemma, no. Yet even if there were such a dilemma, it is not clear to me that it would operate in the same way in which the Collingridge dilemma has come to operate. That is, the DSC has come to be understood to mean that social control is illusory and should not be pursued or expected to be pursued. It is not clear to me that any well-articulated dilemma of societal alignment would similarly mean that we should not pursue or expect to pursue societal alignment. It also does not seem that this is the aim of the authors, and so the pursuit of the articulation of a dilemma is quixotic.

One could imagine going beyond Ribeiro and coauthors to articulate an actual dilemma, if one were truly motivated to do so. Here is the outline of a dilemma of societal alignment that one might attempt to construct:

Social alignment around innovation is therefore a contradiction, a dilemma, and should not be pursued because it is as illusory as social control.

Do I believe in such a dilemma? Only to the extent, again, that it helps set some boundary conditions and allows for the fuzziness of reality in between – but for this one even less so. Innovation is about change and disruption, but it derives from contemporary socio-technical arrangements as well as seeks to surpass them. So, at a minimum, engagement activities can reflect those same arrangements. Moreover, engagement activities, even those involving lay publics rather than experts or stakeholders, can be futures-oriented and similarly seek to surpass those limits (see, for example, Davies et al. Citation2013; Selin and Sadowski Citation2015; and Selin et al. Citation2017). It is possible that such a dilemma, or even a non-dilemma phrased as one, risks sapping some of the energy and commitment necessary to work toward meeting the challenge of societal alignment.

Indeed, an awful lot of energy and commitment has gone into trying to meet this challenge already. The very vision of anticipatory governance, as articulated through the work Arizona State University’s Center for Nanotechnology in Society (CNS-ASU), which I directed and which was funded by the US National Science Foundation from 2005 to 2016, was to demonstrate how the capacities of engagement, anticipation, and integration and their ensembles can help societal alignment processes manage innovation responsibly (Barben et al. Citation2008).Footnote⁴ Like many of the references in Ribeiro et al., Guston (Citation2014, 266) cites favorably the Collingridge dilemma. But then it emphasizes not some incisive solution to the dilemma, but rather an ongoing intellectual and pragmatic effort – the practice of building the three capacities – to perform this societal alignment. CNS-ASU engaged lay publics in future-oriented deliberations about emerging technologies through such activities as its National Citizens’ Technology Forum (e.g. Cobb Citation2011; Cobb and Gano Citation2012) and the Futurescape City Tours and other material deliberations (e.g. Altamirano-Allende and Selin Citation2016; Davies et al. Citation2012) to explore with publics options for responsible nanotechnology. As I have written (Guston Citation2014, 226), ‘This vision of governance is sympathetic to broader STS concerns that have emphasized the contextual nature of knowledge, democracy, the interactive nature of policy making, and, perhaps most importantly, the centrality of “uncertainty, doubt and indeterminacy to such processes”’ (citation omitted).Footnote⁵

To be sure, Ribeiro and coauthors are not the only ones to gloss over some of this energy and commitment. Beyond my sympathies for their agenda, I would also join in their opening lament that

Their attempt to reframe governance in terms that go beyond the DSC concept is laudable, but the fact that their attempt is anchored in the very language of its temporal logic points to the importance of getting our concepts right.