1I am deeply grateful to Gijs van Maanen, Samuel Ferey, and Antoinette Baujard for their thoughtful engagement with my book, and to Cyril Hédoin for organizing the symposium. One of the purposes of writing the book was to bring together different discourses on the relation between politics and knowledge. I am therefore very happy to engage in the interdisciplinary discussions that these comments invite. As van Maanen rightly notes, the book tries to paint a ‘big picture’, which sometimes comes at the cost of detailed engagement with specific phenomena, arguments, or literatures. The comments by van Maanen, Baujard, and Ferey offer a welcome opportunity to dig deeper into a number of important places.

1. Response to Comments by Gijs van Maanen

2Gijs van Maanen raises three very interesting issues from the perspective of data science and political theory. His first question is about the role of ‘big data’ and ‘data governance’, inviting me to reflect on the “role, importance, and value of data-sets for an ‘epistemically well-ordered society’” (van Maanen, this issue). He points out that many governments have started to publish big data-sets as a way of practicing ‘open’ and ‘transparent’ government (van Maanen, 2023).

3My response to this—very timely—question about ‘big data’ is that the way in which governments deal with data can indeed be addressed by help of several key arguments of Citizen Knowledge. One is that the crucial question is not about data, or pieces of information, but about knowledge (Chapter 2). Most of the time, simply publishing data is not sufficient for governments to create genuine openness and transparency (though it can be necessary condition, depending on the topic in question). Data-sets do not speak from themselves; it takes expert knowledge to extract the key information from them, and to understand the political implications, of, say, huge amounts of data from measurements of fine particle pollution in different parts of a city. Without experts interpreting such data for citizens and discussing what could be done to reduce particle pollution in certain neighborhoods, the mere publication of data sets creates a kind of pseudo-accountability that can in fact obfuscate the facts and the policy options that can be derived from them (e.g. O’Neill, 2002).

5Van Maanen’s second comment or question turns around the notion of infrastructures, not only in the metaphorical sense in which I use the term ‘epistemic infrastructures’ in the book, but also in the sense of ‘cables and routers’ (van Maanen, this issue) and other forms of materiality. As van Maanen rightly notes, infrastructures only become ‘visible’ (in the sense of garnering our attention, rather than functioning in the background) in the moment they break down. He discusses a fascinating example of ‘council information systems’ which are used in Dutch municipalities to run city council meetings, and which are provided by only a handful of private companies, for all municipalities in the Netherlands. This raises questions about the relation between commercial interests and their potential impact on democratic decision-making, with the public sector lacking the expertise to provide such infrastructures itself.

6This comment opens up a number of fascinating ways in which the ideas from my book could be combined with empirical research on infrastructures, in all their material and immaterial dimensions. One first thing to acknowledge is that all the infrastructures I have discussed in the book also have a material side. If public schools, for example, are hosted in crumbling buildings, this can not only hamper the learning experiences of students, but also lead to an exit of well-to-do parents into private schools, thereby contributing to the widening of the educational class-divide that we see in many Western democracies. When it comes to online infrastructures, there is equally a material underbelly which involves not only ‘cables and routers’, as van Maanen rightly notes, but also the sourcing of rare metals, often at grueling working conditions, in many countries of the Global South (see e.g. Riofrancos, 2020). One can here also draw connections between infrastructures and sustainability in an environmental sense (which van Maanen raises in a different context, see below): How can the infrastructures that democracies rely on be provided in ways that do not harm the environment or local communities, and that can be sustained in a future that will hopefully, one day, be fossil-free?

7The other line of thought concerns the relation between the public and private sector when it comes to the provision of (material and immaterial) infrastructures. ‘New public management’ has led to the outsourcing of many infrastructural activities to private providers. Another example, in addition to the one provided by van Maanen, is the grip that the private providers of research infrastructures have on Dutch universities (e.g. Boomsma, 2024). As Mazzucato and Collington (2023) have also recently argued, the widespread use of consultancies in the public sector meant that many public institutions no longer have the relevant expertise in-house. What kind of vulnerabilities does this create? How safe are the contracts that are concluded between institutions that are meant to serve the public good, and private companies, e.g. when it comes to the risk of data leakage? What makes this area particularly difficult, politically speaking, is that because of the ‘invisibility’ of infrastructures, the topic is not likely to achieve high political salience. In the term coined by Culpepper (2011), this makes it an issue of “quiet politics”, in which influential interest groups often get their way.

9Moreover, while it may not be possible to make such infrastructure questions salient to the general public, alliances between experts and those groups directly concerned (e.g. council members, or scholars at public universities) might create what has been called ‘issue publics’, to increase the scrutiny of the practices in question (see e.g. Ryan and Ehlinger, 2023). In the case of academic publishing, those concerned are academics and university administrators who no longer want to accept the grip that private companies have on the publication process. This, I take it, is the best chance of creating democratic accountability for infrastructures that are, indeed, best not seen because they function without problems in the background, enabling other processes, such as deliberation in municipal councils or research and publication processes in academia.

10The third topic that van Maanen raises concerns the temporality of democratic decision making in the face of the urgent, and massive, chances that are needed to act against climate change (or, insofar as a certain degree of global warming is irreversible, adaptation policies). Have democracies ‘talked enough’, might it even be necessary to adopt a ‘dictatorial stance’ (van Maanen, this issue) to get going on climate policy? I am in two minds here. On the one hand, I am firm believer in democracy as the system that can best make use of ‘the knowledge of the many’, and that can also best generate legitimacy for far-reaching policy measures. This will require that the ‘talking’ continues. But I understand and share the urge to ‘get into action’. For me, the relevant question is not whether to replace democracy by something else, but rather how to get democracies into action. In my next book, one of the proposals I discuss is a reduction of work time and the encouragement of ‘time for democracy’, i.e. civic and volunteering activities in which citizens can, together, take care of common spaces or address environmental issues in their cities and regions (Herzog, 2025, Chapter 5). While environmental tasks are not the only ones that could be addressed by such an approach—and while not all tasks that are needed for getting our societies in line with the Paris agreements can be taken up by groups of citizens volunteering together—this could be an important step forward with implementation. But it would require acceptance of the fact that we are indeed facing a climate emergency that requires drastic measures. And this, admittedly, leads back to the question of whether democracy is too slow and caught in endless ‘talking’. But I wonder whether this is the greatest problem—or whether it is the stranglehold of vested interests over the democratic process? Of course, there can be unholy alliances between both, with vested interests, who benefit from the status quo, prolonging public discussions by obfuscation and confusion. This was part of the infamous ‘tobacco strategy’ through which tobacco companies delayed regulation, and which has also been taken over by fossil companies (Oreskes and Conway, 2010). Would a ‘climate dictator’ be better able to break the power of these vested interests than democratic forces? At least this is an open question. I continue to believe that democratic strategies are our best bet.

11Lastly, let me comment on the beautiful metaphor of a “democratic garden in decline” with which van Maanen concludes his reflections. I see the power of this picture: the garden of democracy needs care, and we have neglected it for too long. At the same time, I think it remains almost too beautiful a metaphor—for in a neglected garden, nature takes over again, and a bit of neglect can be extremely alluring. This is not what we are currently facing; we are facing, instead, more and more ransacking of nature through capitalist actors. If we democrats are the gardeners, can we stand up against the bulldozers and the concrete casters, as it were? Unfortunately, that needs to happen in addition to the care for the plants, to protect this garden not only from the erosion of institutions, but also from their active destruction through anti-democratic forces.

2. Response to Comments by Antoinette Baujard

12Antoinette Baujard approaches my book from a very different disciplinary angle: that of welfare economics and its methodological questions. Let me first try to summarize her points, and then provide my response.

13Welfare economics is built on the assumption that normative questions should be decided on the basis of the impact they have on individuals’ utility, which is expressed in their preferences. This sets aside, from the start, other normative criteria, for example human rights or environmental concerns: everything needs to be translated into individual utilities to be ‘visible’ through the lens of welfare economics. But this leaves a fundamental question unresolved, namely on which informational basis individuals form their preferences. What if they fall prey to misinformation, for example? This leads to the question of how to think about the possible revision of preferences. But who would have the right to make claims on behalf of individuals, or to request them to change their preferences? Isn’t there a risk of problematic forms of paternalism if one moves away from the metric of unreformed individual preferences?

14Baujard uses the example of Corona policies and the question of which experts were invited to provide their expertise to politicians. She argues that it is a question of shared values which kind of information—and hence which direction of policy—is given priority, for example child psychology of epidemiology. In her words: “the control of the agenda is itself key” (Baujard, this issue). To prevent undemocratic forms of expertocracy, she emphasizes the need to be transparent about the values that decide about which experts get invited. The “collective ethical norm” (Baujard, this issue) that determines priorities in such cases needs to be made explicit.

15In response, I agree with Baujard that my arguments can be funneled into a critique of a purely welfarist approach to policy making (one may want to defend, instead, certain combinations of welfarism with other normative approaches). Individual preferences are indeed often misinformed and biased. Behavioral economics and psychological research have long emphasized the many ways in which individual behavior deviates from the models of fully informed, fully rational behavior that is used in many economic models. Real human beings do not always show “subjective ordinal, non-comparable, unidimensional, monist utilities” (Baujard, this issue)—one may well wonder how often they do show them.

16I share the worry about paternalism that can suggest itself once one gives up the sacrosanctity of individual preferences. But I also have another worry, which Baujard does not articulate (although it is not in contradiction with her arguments). This is the worry that if one does not try to correct badly formed preferences, one often leaves individuals at the mercy of other actors who try to benefit from such badly formed preferences, e.g. by misinforming them and then selling stuff to them that they do not need or want. In the context of the ‘liberal paternalism’ debate, this argument has been made by Andreas Schmidt (2019): if the government does not nudge citizens, this often means that they are being nudged by others, e.g. corporations, and the government might, for example, want to provide ‘counternudges’. If this sounds as if an infinite back and forth might be lurking here, then I think that this intuition is correct: companies are likely to always use new insights, for example from psychological research, to nudge individuals into buying their products, or giving their attention to the advertisement they are shown. Governments will always need to be on the lookout for such maneuvers, and, if necessary, regulate the use of highly manipulative strategies. (Of course, there is also a question about the possibility of governments using such manipulative methods—in such cases, the checks and balances of a democratic order would, hopefully, block such strategies).

17With regard to the need to make values explicit in the interactions between experts and citizens (or parliamentarians or governments, as their elected representatives), I could not agree more. I should have made more explicit than I did that this also matters for the power of agenda setting, the ‘second face’ of power in Lukes’s famous typology of power ([1974] 2005). And there is, indeed, also the ‘third face’ of Lukes’s typology: the exercises of power that function via the choice of framings, the options never considered, etc.

18With Baujard’s focus on agenda setting, groups of actors come into the picture which I have not discussed in detail, but whose power is undeniable, and whose role in democracies thus also deserve to be discussed: for example, the moderators and facilitators that organize citizen assemblies or exchanges between experts and citizens. As I mention briefly in the book, empirical research draws a somewhat paradoxical picture of this group: its members are often driven by democratic, egalitarian values, but they are also very much aware that their own role in such processes can give them opportunities for influence that are difficult to square with this egalitarianism (Lee, 2015). Another group are the civil servants who act as gate keepers in the sense that they invite experts to interact with politicians.

19One might here return to the perspective on socio-economic class that I discuss in Chapter 10 of Citizen Knowledge. In a class society, it is likely that powerful people in politics will choose those experts as their advisors who come from similar backgrounds and have similar types of credentials—the well-off professor of epidemiology rather than the nurse with practical knowledge who ‘only’ has practical training, let alone the single mother who could report what it is like to try to homeschool three kids in an overcrowded apartment during a lockdown (which should also count as a form of expertise, after all!). This can lead to value judgements not being made explicit, as Baujard rightly emphasizes. But it can maybe also lead to a very basic kind of empathy gaps that lead to blind spots, with privileged people simply not seeing what the lived reality of others is, and how important, for example, seemingly small differences in legal regulation or seemingly small amounts of money can be for people’s day-to-day experiences.

20The question that Baujard’s contribution leaves me with is this: How can one ensure democratic accountability in all the processes that precede democratic decision-making itself, such as the choice of certain experts over others? Good management of such processes and transparency about the values that guide decisions are certainly key. But I also wonder what role the media, with their watchdog function, need to play here: journalists can critically scrutinize the composition of expert bodies, for example, while expert communities themselves can try to ensure that when they are asked to send delegates into policy advice contexts, they ensure variety of backgrounds and perspectives (see also Citizen Knowledge, Chapter 8.4.3). To some extent, accountability will also have to come retrospectively: the science-policy-nexus needs to be monitored and there need to be ways in which ‘lessons learned’ can be shared, discussed, and the results implemented in the next rounds. Of course, when politics happens in a crisis modus, as was the case during the Corona pandemic, there will be less attention to such reflexive questions—but this makes it all the more important to review the processes and decisions later on, to see what could have been done differently. But many exchanges between experts and citizens (or the latter’s representatives) take place in a less pressured pace, and so one can create room for deliberating on the composition of teams, on the setting of agendas, and on making value judgments explicit, in the process itself. This is itself also a learning process: how best to manage this interface, in ways that respect the value of expertise, but also the democratic imperative not to overstep it. Democracies need to here take the experimental stance that Dewey, among others, recommended: always remaining attentive to what works and what not, and what could be done better.

3. Response to Comments by Samuel Ferey

21Samuel Ferey starts his reflections on my book with a reference to the ancient discourse on mixed constitutions. He argues that Citizen Knowledge, in a parallel way, makes a case for a “Mixed Epistemic Constitution.” This is a flattering comparison, but I must admit that the case for mixed institutional solutions, with mutual checks and balances, has always had more allure for me than seemingly simple institutional tools (e.g. direct democracy). The latter, in my eyes, often end up failing to fully take seriously our complex social reality. Or to put it in the form of a question: wouldn’t it actually be surprising, given the complexity of our social life, if a non-mixed constitution offered the best of all possible solutions? I thus share Ferey’s sympathy for the Polybian tradition, but it does come at a price: it requires attention to many more dimensions of democracy (mixedly constituted, that is), in all their historical and sociological variety, and thus leads to messier arguments, than abstract arguments “for and against” democracy (which I indeed do not discuss in the book, as Ferey rightly notes; Ferey, this issue).

22Ferey also comments on the fact that I combine the liberal and the republican approach in Citizen Knowledge. The latter brings in a focus on institutions and the risk of institutional corruption. This also leads to more questions about “the responsibilities of individuals participating in epistemic life” and calls for “new forms of regulation” (Ferey, this issue). It is, of course, true that I have been inspired by these two traditions. But it is worth asking what it is in these traditions that leads to this change of focus. It is not that liberals do not assume the existence, and functioning, of institutions—it is more, I think, that they are interested in discussing principles and the justification of principles, maybe based on certain assumptions about the division of labor between philosophy and the social sciences. The republican tradition—in addition to having a somewhat different notion of freedom and a stronger focus on republican shared self-rules—has a stronger focus on institutions and all that can go wrong with them. But if pressed, I would like to defend the claim that liberals—at least if you mean ‘liberal egalitarians’ with this term, but maybe even ‘classic liberals’ —also need to take questions about institutions seriously. Maybe, in the post-WWII years, they have simply presupposed more or less well-functioning institutions (see also Claassen and Herzog, 2021). But in today’s world, this seems an extremely dangerous form of negligence. Even liberals without any republican leanings need to ask questions about how democracy, the rule of law, the welfare state, and other institutions that can realize the principles of justice they argue for can be protected against the onslaughts of non-democratic forces.

23Moreover, as I have explored, together with colleagues, in other work since Citizen Knowledge came out, institutions often have a tendency to decay on their own, even when not attacked from the outside (Herzog et al., 2024). It takes constant effort to keep up the human motivation to do what is right from the perspective of an institution, rather than what is most convenient or delivers personal gains. This does not mean that it would be impossible to maintain institutions—there are many successful examples to the contrary, of course. But understanding how to do this requires a richer account of human motivation than the one offered by rational choice models. The predominance of economistic thinking in management, including public management, has done a lot of harm in this respect (see also Ghoshal, 2005). Protecting institutions and making sure that they continue to realize the principles they are meant to realize, must thus be a concern for liberals and republicans alike.

24A second point that Ferey raises has to do with the way in which individuals are situated in institutions, and what this means for the distribution of responsibility. He uses the example of epistemic limitations such as, for example confirmation bias (the tendency to pick up mostly information that confirms one’s own views). He argues there is a “tension between individual and institutional responsibilities” (Ferey, this issue) (in such cases: one might appeal to individuals to try to combat their own biases, or one might look for institutional solutions that help doing so, e.g. explicit hints about the risk of confirmation bias in reporting about certain issues).

25In response, let me say that I do not see this as an either-or, but as a matter of constructing the best synergies between the individual and the institutional level. Some individual responsibility certainly needs to be part of the picture, and this requires, for example, that individuals have some learning opportunities (including opportunities to take wrong decisions and learn from them). If institutions try to protect individuals from any kind of bias (assuming that this would be possible), this is likely to lead to forms of paternalism that would be problematic for other reasons. But this does not mean that there is no scope for institutional solutions. It is in particular when the stakes are high—e.g. when people’s health might be put at risk—that there should be a stronger focus on institutions to prevent misinformation and biases as much as possible. This is why, in many countries, public speech (e.g. advertisement) about medical products is more strictly regulated than that about products that do not have such a great impact on people’s lives.

26I would add a third dimension to this discussion, however. In addition to the individual and institutional levels that Ferey distinguishes, there is also the interpersonal level, which has to do with how people react to each other, what informal norms develop between them, etc. When it comes to the fight against biases and blind spots in people’s epistemic situations, then this level also seems crucial to me. We often make up our minds in conversations with others, and in such conversations, we have a responsibility towards one another to support everyone’s epistemic situation (this may not be true for all types of conversations, of course—but for many, it does hold). Ferey rightly noticed that I had decided not to go for a framing in terms of (epistemic or moral) virtues in the book, but here it might be helpful to draw attention to the concept of “other-regarding epistemic virtues” (Kawall, 2002) that describes virtues through which the epistemic situation of other people is improved.

27Of course, other-regarding epistemic virtues are highly context-specific. And there is a question about the extent to which they can play the positive role one would expect from them in contexts in which a population is divided into two or more ‘epistemic camps’, as one might call them: different groups who, among themselves, share different worldviews, sources of information (‘epistemic infrastructures’ in my vocabulary), priorities of discourse, etc. When such epistemic camps deviate more and more from each other, both epistemically and socially, it becomes difficult to see how a res publica as a ‘shared thing’ can continue to exist. This is why keeping the epistemic house of democratic societies together, as it were, is so crucial to protect democracy.

28Lastly, Ferey asks an important question about the role of the law and its own epistemic conditions in the argumentative architecture of my book. Having distinguished the three different logics of knowledge creation and transmission (markets, experts, deliberation) allows me to identify two types of problems: corruption within one of the spheres, so that it does not fulfil its own epistemic and social functions properly, and forms of imperialism, or trespassing of boundaries, between the three spheres in which the three logics should be at home. Addressing these questions, Ferey argues, will have various regulatory implications, but I here leave some crucial questions open.

29The first, in his view, is the “libertarian objection” (Ferey, this issue) that sometimes, when it comes to regulation, the cure is worse than the disease. This objection might be brought forward, in particular, against my arguments about the need for market regulation. If markets fail because of epistemic problems, it might be better to let markets actors come up with their own solutions, Ferey suggests. In response to this objection, let me distinguish two kinds of market failures that look rather different in this respect. The first are informational failures that harm both sides of a market, for example by leading to the complete breakdown of a market that would otherwise be mutually beneficial. In such cases, there are indeed market-internal incentives for the participants to come up with solutions. For example, in his famous discussion of the “market for lemons,” Akerlof (1970) himself discusses solutions such as brands or guarantees to overcome information asymmetries (see also Citizen Knowledge, 94). I would grant that such mechanisms can often work without the need for external regulation.

30But a second kind of market failures is such that one side benefits while the other loses. A typical case here would be misleading advertisement or other forms of information gaps on the part of consumers. Here, only one side in the market has an interest in changed—companies, in contrast, may be perfectly happy to sell more products even if this may not be fair (or efficient!) from the perspective of consumers. In such cases, waiting for market-internal solutions is likely to be futile. The only non-government actors that might stand up against such practices would be consumer organizations, in which the interests of consumers are bundled and defended. But it is for a good reason that such organizations often call for legal regulation: they do not have sufficient sanctioning power to really enforce better market standards.

31This is why regulation is often needed if one wants to have markets that are truly epistemically efficient and beneficial for all participants. However, I would grant that many forms of regulation that we see today are indeed not well executed. But in my view, the quality of regulation is often undermined precisely because of lobbying activities and undue influence by the actors that should be regulated—if only in the form of pleading for numerous exceptions, which makes regulatory frameworks complex and difficult to understand for lay people. The problem that I am concerned with throughout the book—the impact of market thinking and the power of market actors over democratic politics—thus returns here as well. Let me also note that when calling for regulation, I do not necessarily mean ‘more’regulation in the sense of ever more rules, but rather regulation that genuinely addresses the market failures in question. Often, such regulation can be rather simple, for example by placing the burden of proof and the liability for lack of correct information on the actors who need to carry these responsibilities.

32The second question that Ferey raises about calls for regulation concerns the role of the law as “an epistemic sphere of its own” or as “a shared normative foundation underpinning the other three epistemic infrastructures” (Ferey, this issue). Ferey writes that “Much of the law’s effectiveness in regulating the three other epistemic institutions depends on its own epistemic processes” (Ferey, this issue), and this is undeniably true. I had not discussed the law separately from the sphere of deliberation, seeing it as a task of democratic politics to not only deliberate well, but also to give shape to the results of deliberations in the form of well-designed laws. To avoid a problematic infinite regress of mechanisms, I had argued that the deliberative sphere has the meta-responsibility of deciding about the delineations between the spheres (Citizen Knowledge, 52). I had certainly also meant to include the dimension of legal regulation here, but without making this sufficiently explicit and thinking about the additional challenges that lurk here.

33This is indeed a gap in my argument. I am fully aware that many of the processes of corruption that I am concerned with play out in the legal sphere – for example in the form of the “product defense industry” in which misinformation about products is spread not only in the public in general, but also in legal strategies that include, for example, the hiring of one-sided expert opinions (see Michaels, 2008). But I have not discussed how such problems could be prevented, e.g. by discussing explicitly the intersection of experts and the legal system.

34I lack the expertise to fully grasp what it would take to better equip the legal systems of democratic societies with the right ‘epistemic infrastructures’ for fulfilling their functions within democracies in the right way, rather than becoming another tool in the hands of the powerful. Such a discussion would require detailed engagement with the legal traditions and legal practices of different countries. My cautious guess is that countries with a Roman law tradition, and a strong professional ethos of legal experts as having a public role (such as expressed, for example, in the fact that in some countries they have a state exam rather than a bar exam) have a better chance of tying the legal system to democratic values and the rule of law. But more empirical work and more arguments would be needed to back up this claim.

Conclusion

35Let me conclude by thanking the three commentators again for their insights and arguments, which gave me the opportunity to delve deeper into some questions that the book raises but does not answer. The search for more arguments—and for how to apply the arguments in concrete political contexts—will certainly need to continue. I have been writing this commentary in the week the US people elected Trump for president, and the centrist German government crumbled, probably to be replaced by one more to the right. Over the last years, the threats against democracy seem greater than ever. Improving the epistemic dimensions of democracy is therefore urgent, but—as I also argue in the book, and am even more convinced now—ultimately, the epistemic and the social dimensions of societies cannot be kept apart. We need to find ways to come together again, as citizens, to discuss the values that divide us, but also to understand which values we share and what we want to preserve. And this will not be possible without also paying attention to normative values such as respect and justice. In the end, despite all attempts to draw conceptual distinctions, social, political, and moral epistemology have to come together.