2Lisa Herzog’s stimulating and inspiring book has convinced me that there is such a thing as a knowledge issue to cope with in order to defend the values of democracy and agency, and I do share much of her sentiment of urgency. I have then strived to scrutinize what particular form and priority infrastructures should have to circumvent the knowledge issue, which respects the values of democracy and agency.

4Let us recall why information is a key issue for welfare economics in the first place. When there are market failures—including externalities, spillovers, public goods—, then the information conveyed by prices is insufficient or unreliable, and one cannot straightforwardly assess what would be a good policy. Welfare economics is then called to intervene. Before all, it scrutinizes which information is (or is not) relevant for the evaluation of social states, the evaluation of public policy and the help of collective decision making. Questioning relevance implies whether the decision (or evaluation) based on this information will be able to be rational, adapted, efficient, fair and respectful or not of every kind of normative properties. Selecting an informational basis for decisions is both a commitment as to which pieces of information we look at and which ones we overlook. The debate on the informational basis of welfare economics has been popularized by Sen in the first place (Sen, 1977; 1979). He used the term “welfarism” in the 1970s to highlight the consequences of the restriction of information to the detriment of others. Welfare economics is welfarist because it considers that social welfare only depends on individual well-being, and also because it associates individual well-being only with subjective ordinal, non-comparable, unidimensional, monist utilities. There may be other valuable information, including the survival of a specie, the sustainability of a river, the objective quality of life of women, how humans are dependent from one another, species and planets, the fact that some subalterns are oppressed, that some are given asymmetric rights, and that others suffer from living in a world of un-adapted norms, and so on. But those are irrelevant from the perspective of welfarism, apart from their instrumental contribution to individual subjective utilities.

1. Articulating Utility and Knowledge

1.1 The Implicit Assumption of Welfare Economics on the False-True Demarcation

8For these various reasons, welfare economics evolved. Being welfarist does not imply to restrict its attention to utility information only. Sound knowledge also constitutes some relevant information, knowledge being an aid for better identifying well-being in the measuring of social welfare.

9However, the process of “laundering” preferences still requires serious justifications (Goodin, 1986), in particular in the case of liberal paternalism (Fumagalli, 2016). The liberal paternalism approach considers that individual preferences—e.g., the preferences that the individual would have, should they be rational and fully informed—are the ultimate justifications for these corrections. The circularity of the argument is nevertheless highlighted under the name of the “reconciliation problem” (Sugden and McQuillin, 2012; Lecouteux, 2022). One potential consequence is that preferences may be irrelevant, implying the definitive undermining of welfarism, in whichever definition (Sugden, 2018). An alternative—on which we focus here—is that, should preferences have any relevance at all, elaborating these justifications need information which, by construction, are not conveyed in subjective preferences (Goodin, 1986). Hence there is other information but preferences considered as relevant to justify preference laundering, and this raises novel fundamental questions: upon which criterion or which condition can a preference be corrected? Who is entitled to formulate preferences correction? What is ultimately a correct belief? Is there such a thing as a corrected preference based on true and complete information, uncontroversially axiologically neutral and respectful of the genuine individual preferences?

10Preference correction implies to imagine what the preferences of an individual would be under correct beliefs rather than false beliefs. A necessary (and not sufficient) condition to design corrected preferences is that true and relevant information can be demarcated from fake news, junk science and irrelevant manipulations based on vested interests, and that every stakeholder is able to properly identify it. Using available knowledge, if the fake was hardly sorted out from the true, it would be hard to guarantee that preferences are correctly revised. In most papers on preference correction, the assumption that the false-true demarcation is possible, comes as an implicit prior. Should experts tell an uncontroversial truth, then those who do have access to it are entitled to participate in the process of preference correction. Experts may eventually know better than the individuals whose preferences are being studied. The preference correction process results in an asymmetric power given to experts to define welfare, while it retrieves, at least partly, the ability of defining the relevant information on their well-being from the citizen themselves.

11This outcome comes as a strange paradox. Welfarism was primarily justified in order to respect individuals’ voices only, protect individuals’ freedom, and by the fear that the consideration objective information would give too much importance to experts and planners (all this in the context of the cold war opposing communism to liberal democracies, as highlighted by Amadae, 2003 or Baujard, 2017). Welfare economists have paid severe attention to kick out experts by the door; the necessity of correcting preferences now makes them come back by the window. In this context, the assumption of the false-true demarcation appears as a necessary brick in the wall, to avoid the complete collapse of the justification of welfarism in welfare economics.

1.2. The Entanglement of Truth and Knowledge in the Post-Truth Era

12Because she focuses on the knowledge issues in our concrete societies, Lisa Herzog does not retain the false-true demarcation assumption a priori. She undertakes a systematic scrutiny of the validity of this assumption on the basis of a considerable literature, among others from the social studies of science. As a consequence, she takes seriously what it means to now live in a “post-truth era”. A post-truth era does not imply that there cannot be any reference to truth: Truth can be and must remain a reference. The access to truth, by contrast, is more debatable. Herzog rather claims that knowledge as such should not be taken as obvious truth, wherever it comes from.

13Before all, there may be fake news. Some are introduced and spread on social media with an intention, where truth, to say the least, is not the ultimate reference. Fake news may influence everybody’s perception of the truth. This may not affect the inner scientific debates as such and the credibility of scientists for citizens, as those “continue to be seen as one of the most trustworthy professions” (1). However, that it may influence public debate on decisions is a matter of great concern.

14Secondly, the science process makes true information only partly known, or still unknown at different moments through time (76). Let some scientific results published in the scientific literature come with the conclusion that what is best for your well-being requires something you do not want—e.g. you do not want to take hydroxychloroquine because you have bad feeling about it—furthermore not yet supported by other published papers in scientific journals. Is an expert entitled to “revise your preference” and make you taking this drug in spite of your subjective preference? Waiting for a while would show that your intuitions were right, as this molecule has been soon proven to be inefficient and dangerous. Any scientific assertions, proven at one moment of the time, supported by few isolated articles, may not appear sufficiently decisive to say what should be—and therefore to justify revising individual preferences. On the contrary, at some point in time, an assertion becomes accepted beyond reasonable doubt. For instance, “doubt” regarding the anthropogenic climate change “is no longer reasonable” (120). On the other hand, this argument may also be used as a strategy to undermine the credibility of scientific results, as the well-known tobacco example presented below will illustrate.

15Thirdly, Herzog describes cases of more or less conscious and self-interested manipulation of scientific knowledge. It is now uncontroversially proven that the scientific knowledge on the tobacco impact on health has been intentionally obfuscated, slowed down and undermined by systematic strategies of doubts (117; 195; see also Oreskes and Conway, 2010). We cannot be blind to the fact that there is always a risk that this phenomenon may exist on every topic where scientific knowledge has important economic, social, environmental stakes, and which induces strong conflicting interests. Some may have interests in influencing the production of knowledge in a direction, which may generate some discrepancy, at least for a moment, between knowledge and truth. Whether we need to take current scientific knowledge in spite of the doubt, or without doubt is therefore not an easy question.

16Fourthly, knowledge is asymmetrically accessible, and the fact that some know more than others is unavoidable. Every public decision requires at least relevant description and understanding of the situation (Sen, 1967). The necessity of knowledge hence justifies to rely on experts, defined as those who do know more and will provide relevant knowledge for decisions. Besides, the interests of experts and their clients do not always exactly coincide. For instance, “an individual medical doctor may face a conflict of interest between the best advice for a patient and his own financial interests and may be tempted to, say, use more diagnostic tests than necessary.” (79) Experts are not just science tellers, they are also flesh and blood individuals with a life, personal ambition, embedded in a professional surroundings and constraints. Whether conscious or not, whether intentional or not, experts’ speeches may entangle interests. Because of asymmetric knowledge and conflicts of interests, experts have both a possibility and an incentive to decorate the spoken knowledge in their own interests. “This possible discrepancy between the incentive structures and the ‘right thing’ to do is particularly difficult to control by formal means because experts can always claim to have acted in good faith” (179). This situation explains both why knowledge integrity rules are essential, and also why they may not be sufficient in the end to guarantee there will never be such difficulty. As a consequence, citizens may not always believe everything that any experts may say. At the public level, that citizens cannot control experts’ knowledge and interests is likely to feed resentment and to raise suspicions against technocratic decisions (1; 177).

17Let us now reconsider the false-true demarcation introduced above. We have described how welfarist welfare economics needed to make the prior assumption that true and relevant knowledge could be demarcated from false knowledge, in order to justify preference revision. In light of these four characteristics of knowledge in the post-truth era, this assumption may be requalified as a confusion between truth and knowledge. Herzog’s book warns us against this confusion. Obstinate facts, as the phenomena of knowledge factories, as well as conflicts of interest in the context of asymmetric information between experts and citizens, undermine the assumption that it is always possible to disentangle truth and knowledge. We should acknowledge that this demarcation may be difficult. In concrete situations, the naivety of the confusion between truth and knowledge is a sin, which comes in contradiction with the basic justification of welfarism. The book rather calls for action: we need to be able to disentangle manipulated knowledge and correct information in order to reinstall the actual reference to truth in the post-truth era. This is a necessary condition to respect the values of liberal democracies and individual autonomy.

18Being aware that this assumption is not reasonable at the first place is a prior condition to design information processing for public decisions, in respect of the values of liberal democracies.

2. Disentangling Values, Preferences and Knowledge

2.1 The Diversity of Knowledge for Public Decisions

19If a given person were able to properly judge the coincidence of truth and knowledge, as in Plato’s ideal, we could all rely on this unique person to revise our preferences. Even ignoring potential conflicts of interests for now, Plato’s ideal is hardly acceptable because of the consequence of an inevitable fact, that expert knowledge is highly specialized (72; 177). For almost everything “we rely on information, knowledge, and expertise from others”; this divided epistemic labor induces a dependence on experts (177). We most often accept it, but we also observe growing situations of distrust and fear of technocracy, insofar as some may be seen “as holding too much power over laypeople, concentrated in too few hands, with too little accountability.” (182)

20Inquiry on the origin of this mistrust may come from a confusion between the respective role of experts and politicians, which today must be described differently than in Weber’s times. Herzog insists on the importance of reallocating roles: politicians (representing citizens) should be choosing goals, while experts should be only helping them in this direction:

Scientific knowledge can and often should play a role in political decision-making. What is problematic, … is, first, the assumption that it is scientific knowledge alone that should be counted as expertise, and, second, the assumption that expertise alone could decide questions that are, strictly speaking, beyond its scope. Political decisions can never be based on descriptions of facts, mechanisms, or causal relations alone; there always also needs to be a premise that concerns the goals that are to be pursued, which, ultimately, needs to be based on certain values or interests. The latter need to be combined with descriptions of facts to arrive at guidance for action, and factual experts are not experts in this regard. (83)

21The role of experts is not just to process knowledge mechanically, but, given goals, values and interests, to choose how to process it in order to come to a public decision. An issue therefore occurs if experts are influential in choosing goals themselves, rather than contending to help decision-making given such goals. To do this, I suggest shifting the question within the debate on information in welfare economics. In normative economics widely speaking, goals are assumed to be chosen by a benevolent planner, while economists contend with finding best means to achieve these goals (Robbins, 1932; Mongin, 2002; Atkinson, 2001; 2009). Value transparency is a key condition to guarantee the proper allocation of tasks between scientists and politicians (Baujard, 2021). Reformulating Herzog’s focus on the diversity of knowledge in this specific context shall help to first analyze the nature of the various different pieces of information which are decisive, and secondly check how relevant information emerges, is identified and processed.

2.2 Collective Norms as a Necessary Additional Information

23As highlighted by Herzog, the views of experts coming from various scientific communities may conflict with one another. Taking the case of Covid policies as an instance, health experts structurally focus on certain populations whose survival was made vulnerable to the pandemic; by contrast, education experts structurally focus on other populations, whose studies and long-term skills and opportunities were impacted by repeated lockdowns. Their policy recommendations typically diverge from one another. It does not uncover insufficient knowledge in health or in education. No better research could decide whether the policy driven by health concerns should be recommended rather than the policy driven by education concerns—or the converse. It would still hold in the total absence of conflict of interest and homogeneous individual preferences. Hence this conflict is somehow irreducible, because in the way we introduced this example, it uniquely results from the fact that public decisions should be based on a variety of perspectives, health and education (rather than health or education), each of which uncontroversially impacts different populations. Recommendations from both experts are inconsistent, but they are, however, equally valuable, and incommensurable.

24In such public decisions (e.g., regarding Covid policy) a choice of policy amounts to a choice of dimension. A strict lockdown reveals a priority to health over education, which will end in favoring the elderly to the younger generation; a (partial or) absence of lockdown reveals the converse.

26Facing the diversity of dimensions and the diversity of preferences, choosing which information is relevant and which is irrelevant is crucial. Let us repeat that this choice is neither implied by individual preferences nor by any pieces of knowledge. It requires in addition some information going beyond ethical individualism, which emerge at the collective level, because of the implication of a plurality of persons. Let us call this choice a collective ethical norm. This norm may include the priority of dimensions, the relevant standing of stakeholders, the trade-offs which are made among the interests of diverse populations. Should the ethical norm be different, for given knowledge and given preferences, the public decision should be different.

27This situation now raises a positive question: In the current situation, how does an ethical norm emerge? Who imposes it and eventually influences the policy decision? Identifying the origin of a norm is therefore crucial to discuss the legitimacy of final decision, and to understand why some citizens fosters some resentment, when they do.

2.3 Information in Deliberation

28To face the issue of resentment against experts and their asymmetric power, Lisa Herzog concludes that experts and laymen should be put at the same level, and deliberation would generate the re-articulation of their respective knowledge. The conditions under which the desired articulation is likely to emerge, however, remains underspecified.

32Third, because who decides which information is relevant is key in the final decision, the control of the agenda is itself key: what dimensions are to be discussed, which experts should be invited, in which order they shall present their conclusions. If this agenda choice is made silently, there is a potential for serving vested interests, manipulation, and in general unequal representation. Conversely, the agenda regarding the chosen dimensions could be an object, not only of publicity, but also of debate. Most decisive points concern how questions are asked, how the agenda is made, and how they are summarized by whom is in charge to do so. The tiniest details of the organization of the public debate may have their importance because everybody, experts and citizens, should be able to be aware of the importance of hierarchy of dimensions (and more generally ethical norms) in the resulting decisions.

33As a consequence, the issue at stake is not so much how knowledge should be shared or how it should be regulated. A necessary (and non-sufficient) condition for respecting the role of experts and laymen, and for an equitable participation of individuals in the collective decision is to uncover the importance of ethical norms and to make them transparent. These ethical norms are decisive to identify which information is relevant or not, and for determining the hierarchy of various relevant information. This perspective, in line with the typical approach of normative economics highlighted above, can contribute to Lisa Herzog’s quest of designing better knowledge infrastructure for democracy, by stating one primary necessary condition: any organization of deliberation and use of knowledge for public decision should primarily make transparent the collective ethical norms to be used in their screening of relevant knowledge.

3. A Conclusion on Democratic Infrastructure of Knowledge

34Herzog’s book takes knowledge issues seriously, and comes as an invitation to think which design of democracy and autonomy-friendly institutions are likely to circumvent knowledge issues.

35First, preference revisions in welfare economics rely implicitly on the assumption that truth and knowledge are the same; and, if not, fake and true assertions can be straightforwardly disentangled, such that it is possible to rely on true assertions only. Herzog shows that this confusion does not resist to facts. Taking her thesis seriously introduces a novel criticism of the welfarist approach and of the preference-based approach, and my reconstruction shows that a solution to it will require awareness and in-depth discussions.

36Second, while the decision process is based on the implicit assumption that positive and normative information can be disentangled, their actual entanglement should be acknowledged. Knowledge and preferences cannot come up to a public decision alone, because of the incommensurability of pieces of knowledge and the diversity of individual preferences. A decision ultimately requires some additional ethical norms, which are a matter of debate at the collective level. Information in society can hardly be considered in total ignorance of normative issues, because any public decision requires not only to process preferences and dimensions, but also derives from the choice as to which knowledge is relevant or not, and which is a priority compared with others. It is crucial that the public debate allows transparency on how these norms are introduced.

37The infrastructure of public decisions must account for the entanglement of both issues. Deliberation processes should reflect on their use of knowledge and promote transparency about the hierarchy of dimensions involved in every public decision.