– Knowledge: there is not a single piece of knowledge, but varying forms of knowledge, which may or may not be scientific, may not always be compatible and may sometimes be in competition to aid understanding of the “causes and consequences” of harmful events (Mercer 2012). The social sciences are equipped to decipher the meanings encapsulated in the different acceptations of the notions mobilized (Demeritt 2001; O’Brien et al. 2007; Hulme 2009; Bonneuil 2015). Knowledge never evolves alone, independently of the social world in which it is produced and conveyed.

       – Technocracy and dogmatism: the decision founded in science remains a political arbitration. On the one hand, the practice of science and the legitimacy it enjoys – not to mention the heterogeneity of the scientific field – are a matter of social relations and therefore of politics (Bourdieu 1975). On the other hand, recourse to science to decide is always selective. It depends on the moment, on the context or on the underlying conceptualizations that presume a reading of the world. If they are not made explicit, they are ideology. Pretending to avoid arbitrations on the grounds that geology or atmospheric sciences would know neither morality nor subjectivity is an act of faith.

       – Scientific polyphony: for risks as for the Anthropocene, the diversity of social science productions is considerable. This introduction has mentioned work on both risks and the Anthropocene. In both cases, there is no consensus on the analysis of the social dimensions of environmental issues, and there is a wide range between utilitarian reflections, in line with Earth science approaches, on the one hand, and the analysis of the environmental dimensions of the social, on the other hand. This diversity, characteristic of the epistemology of the social sciences (Passeron 1995), enriches interpretation but can feed confusion and so be detrimental to such environmental research.

       – Interdisciplinarity and integration: it is now commonplace to denounce the fragmentation of research into silos, particularly in the case of subjects such as risk. We distinguish two forms of integration called for in response. One concerns the conditions of research; the other concerns the decompartmentalization of disciplines around the society–environment relationships that are the hallmark of geography and constitute one of its historical markers.

       – The power of capture of one legitimate research sector over another is known (Castree et al. 2014). Some promote the broad field of “sustainability sciences” (Lorimer 2017) or embrace the unifying dynamics of “environmental sciences” with the Anthropocene (Brondizio et al. 2016). Others denounce it, see the opportunity to think differently (Palsson et al. 2013) and resist falling into line with a poorly controlled framing of research. In particular, the debate stumbles over modeling or indicators, in fact over reluctance to put an irreducibly diverse social world into functional equivalence (numerical or not) (Wisner 2016). Extracting the figure from its context also cuts the data from its meaning (Rebotier et al. 2019) and contributes to dehistoricizing the analysis (Moore 2003).

       – Historically, risk studies have endorsed the separation between nature and culture (between hazard and vulnerability). This is particularly visible in the use of conceptual models that struggle to articulate physical processes and social mechanisms in a complex and explicit way (Pigeon and Rebotier 2016; García Acosta and Musset 2018). The Anthropocene moment may offer an opportunity to think about the post-great divide between nature and culture. But integrated ways of considering the reciprocal and non-deterministic relationships between societies and environments were already among the founding debates of geography at the turn of the 19th and 20th centuries (Berdoulay 2008). The trajectory of the notion of adaptation is an illustration of this (Rebotier 2020), and calls for more integrated research on the environment, at least in Francophone geography, and predates the Anthropocene moment (Mathieu and Jollivet 1989).

       – Transformations of the social: the theme of social change as a practice has sensibly receded in the research agenda (Castel and Martin 2012). Regardless of what it is called, (transformative development (Lavell and Maskrey 2014); adaptive transformation (Pelling 2010)), social change is either reduced to the inoffensive perimeter of small reformist steps (Pahl-Wostl 2009) whose modest consequences are continually promised to become significant; or it falls outside the scientific field, on the grounds that it would cover an ideological bias unacceptable in the majority representation of science (Metzger and Robert 2015). Research copes differently with this normative tension. Critical analyses identify asymmetrical development or structuring inequalities as obstacles to effective, sustainable and just improvements in risk conditions (Lavell and Maskrey 2014). The same criticisms are, nevertheless, blunted in official documents to preserve the consensus required by international institutionalization (GAR 2015). Critical thinking seems neutralized between its inclination to say things and a form of inability to change the world. In spite of the impasse, some people take up the issue of social responsibility in two ways. On the one hand, they assume that changing the world (not just making it work or describing it) is part of the mandate of research (Robbins 2004); on the other hand, they engage with other knowledge producers and with social sectors (other than those composed of scientists) that are more action-oriented (Mercer 2012; Shaw 2013). On the subject of the Anthropocene, should we dismiss the critique of capitalism on the grounds that its postponement is unrealistic (Latour 2017)? Along the narrow path that combines critique of capitalism and social transformation, some explore a movementist, collective option (Wisner 2019), which relies on critical initiatives and claims interdependencies as pledges of liberation (Moore 2017b).