Arguments from Popularity: Their Merits and Defects in Argumentative Discussion

How to understand and assess arguments in which the popularity of an opinion is put forward as a reason to accept that opinion? There exist widely diverging views on how to analyse and evaluate such arguments from popularity. First, I define the concept of an argument from popularity, and show that typical appeals to the popularity of a policy are not genuine arguments from popularity. Second, I acknowledge the importance of some recent probability-based accounts according to which some arguments from popularity are epistemically strong arguments, but also contend that despite these strengths such arguments have at most limited value in argumentative discussions. Finally, I show that there are at least five different ways that arguments from popularity can be fallacious, and examine what this means for an account of the Fallacy of Popularity.


Introduction
How to understand and assess arguments in which the popularity of an opinion is put forward as a reason to accept that opinion? There exist widely diverging views on how to analyse and evaluate such arguments from popularity. According to strong critics, such as David Godden (2008), Trudy Govier (2005), and Henrike Jansen (2020), arguments from popularity are always or typically unconvincing and fallacious. According to enthusiasts, like Don Dedrick (2019), and Ulrike Hahn and Jos Hornikx (2016), there are important settings in which arguments from popularity are legitimate and rationally persuasive arguments. Mild critics, such as James Freeman (1995), Douglas Walton (1999), Ralph Johnson and Anthony Blair (1994), and Frans van Eemeren and Rob Grootendorst (1992, p. 166), hold that arguments from popularity, as I will conceptualise them in this paper, can provide reasonable and useful contributions to an argumentative discussion although they are, on their own, weak arguments that often are used in a faulty and fallacious way. This paper has three aims. It deals with the conceptualisation of arguments from popularity, and aims at clarifying a possible confusion in the concept of argument from popularity. It deals with evaluation of arguments from popularity, and aims at showing that it has a modest epistemic role to play in argumentative dialogue. It deals with settings where it makes sense to charge the proponents of such arguments with committing the Fallacy of Popularity, and aims at finding a solution to the problem that on the one hand there does not seem to exist one Fallacy of Popularity and on the other that the concept of the Fallacy of Popularity often is useful and important to keep an argumentative exchange on a good track.
In Sect. 2, I will define "argument from popularity" as an instance of a specific argumentation scheme (see for the theory of argumentation schemes from where I start: Walton et al. 2008). This definition enables me to show that many arguments in which a specific policy is justified by the policy's alleged popularity are-despite appearances-not genuine arguments from popularity. Drawing this conclusion has consequences for the evaluation of such appeals to popularity. For one, it casts doubt on the assumption that arguments from popularity have a surplus value in democratic deliberation about public policy (Walton 1999).
In Sect. 3, I will examine the epistemic value of arguments from popularity. Some theorists point to the fact that when we need to make up our mind about something, we can be justified in assigning some degree of reliability to other people and their opinions. They then use probability 1 3 theory to argue that the popularity of an opinion can, in certain circumstances, provide an epistemically strong reason for the acceptability of that opinion (Dedrick 2019;Hahn and Hornikx 2016). On the basis of their considerations, I will acknowledge the epistemic potential of arguments from popularity (pace the strong critics). But I will emphasize that having or taking a position in a discussion involves more than believing the opinion to be acceptable (true) or unacceptable (false), and contend that in typical argumentative dialogues vis-à-vis an opponent who adopts a position of her own, arguments from popularity provide at most very weak epistemic support (pace the enthusiasts).
Clearly then, the question when arguments from popularity are fallacious is an open question, which I address in Sect. 4. I will go over a number of very different circumstances in which a specific argument from popularity could be qualified as fallacious. What is more, none of these grounds are specific to arguments from popularity. This raises the metatheoretical issue whether a theory of argumentation should retain the concept of the Fallacy of Popularity. Despite these findings, I will make a pragmatic case in support of maintaining the theoretical status quo, and for retaining the Fallacy of Popularity as a useful part of argumentation theory.

Defining "Argument from Popularity"
First, I will delineate the concept of an argument from popularity. I will do so by elaborating on the theory of argumentation schemes (Walton et al. 2008). Being somewhat precise about what counts as an argument from popularity will allow us to distinguish arguments from popularity from a related but different kind of argument, in which the popularity of a policy is advanced as a reason for the policy's democratic legitimacy. This is a point worth making, as it implies that the virtues of the latter cannot without further ado be attributed to the former. (1) "This vaccine is unsafe. Why? Almost everyone thinks so." (2) "We should stop administering this vaccine. Why?
Almost everyone thinks we should!" I am interested in how such arguments fare in critical examination dialogues. A critical examination dialogue is modelled as a conversational exchange in which: (a) a proponent advances a thesis; (b) an opponent critically challenges the thesis; (c) the proponent justifies the thesis by responding to the challenges; (d) the opponent critically challenges the justification; (d) and where these dialectical moves-back and forth-is part of a cooperative effort to arrive at a shared view on what is a correct judgment regarding the thesis (cf. the concept of a critical discussion, van Eemeren and Grootendorst 2004; cf. the concept of persuasion dialogue, Walton and Krabbe 1995). 1 The framework of critical examination dialogue is meant as a general setting, and as applicable to numerous forms of argumentative discourse (cf. van Eemeren and Grootendorst 2004, Chapter 5). The proponent and the opponent can be individuals engaged in a two-person conversation, either speaking on their own behalf, or on behalf of institutions, such as health agencies or governments. An argumentative text in a paper, a letter or a book can be reconstructed as an implicit critical examination dialogue, such that the author in the role of the proponent of a thesis attempts to rationally persuade members of his readership in various opponent roles by anticipating their critical challenges. The thesis may concern the acceptability of a more or less complex theory, or of a policy proposal, and the exchange could take place at the conference table during a lunch meeting, or be extended over several months and spread out over various speech events and forums. I abstract from such contextual features, and simply speak about "the proponent", and "the opponent", and "the moves" they make in their "dialogue".
I need to be more precise about what it means to justify a thesis by responding to, or anticipating, a challenge. The response only counts as argumentation if the proponents presents his reasoning for inspection and assessment by the addressed opponent. Not any attempt at persuasion merits to be qualified as argumentation. If in a video a politician advances an argument in support of a standpoint, and the video also shows that she shakes hands with some rich and famous people (or alternatively with some common folks), this mis-en-scène may exert a causal effect on the addressee's reception of the politician's reasoning. But the hand shaking is not part of the proponent's argumentation, because it has not been communicated as including the proponent's reason for his thesis.
Arguments from popularity have been discussed under a variety of labels, such as: argumentum ad populum (van Eemeren and Grootendorst 2004), argument from popular opinion (Walton 1999), bandwagon appeal (Freeman 1995), and populist argument (Jansen 2020). I refer to it as "argument from popularity", which I use as a non-evaluative term, reserving "Fallacy of Popularity" for fallacious instances. 2 I follow a tradition of regarding arguments from popularity as instances of a specific argumentation scheme (Walton et al. 2008;Wagemans 2016;Hinton and Wagemans 2022), and shall in this paper define "argument from popularity" by means of the following pattern of reasoning, that I label "argumentation scheme from popularity", where T may stand for any proposition (i.e.: of any kind, and of any complexity): T, because most people in group G accept T. I define "argument from popularity" as any instance from the argumentation scheme from popularity. Before proceeding I make a few comments on how to understand this definition.
(1) Like a formal scheme of inference, an argumentation scheme has variables, so that it is an abstract pattern that can be specified to create an indefinite number of real-life arguments. Unlike formal schemes, what remains constant are not the logical terms such as "or" and "not" but some material term (or terms), in our case: "most people in group… accept." Because there is no limit to the material terms that we could find useful for argumentative purposes, there is no natural limit to the nature or number of argumentation schemes. In a specific company of arguers some highly specific and idiosyncratic argumentation scheme could be useful.
Other such schemes are more general and more widely adopted, such as argumentation schemes with which to appeal to expert opinion or to an analogy. The argumentation scheme from popularity is amongst very general schemes, and though it seems to be quite popular, it is certainly not a universally adopted scheme. (2) Argumentation schemes can be useful for characterizing the nature of argumentative discourse, or for explaining the choice for an argumentative strategy. But they also serve a normative function in critical examination dialogue. More in particular, they can be seen as giving significance to arguments that are defeasibly, and not deductively, valid. When the dialogue participants adopt an argumentation scheme, it plausibly functions as a dialogue rule. Suppose that the proponent presents an argument (e.g. an argument from popularity) that instantiates that scheme (e.g. the argumentation scheme from popularity). Then the opponent can still challenge the argumentative connection between the premises and the conclusion of that specific argument (e.g. "if you concede that most people accept that the vaccine is unsafe then that's a good reason to also concede that the vaccine is unsafe"), but by having adopted the scheme as prima facie appropriate, the opponent must be seen as having challenged a presumption. That she has challenged a presumption implies that she incurs special dialogical obligations, such as obtaining a burden of proof for the denial of the argumentative connection (as contended by: Walton et al. 2008), or at least an obligation to provide some explanation for not accepting the connection (as contended by: van Laar and Krabbe 2013). By the way, the proponent can also use arguments if they do not fit any adopted argumentation scheme, but in that case the proponent misses the benefit of relying on a presumption. However that may be, the choice to adopt or not to adopt the argumentation scheme from popularity has consequences for how the critical examination dialogue may develop (See Sect. 4).
(3) In arguments from popularity, what is advanced as support is that a manifold of people accepts the thesis. The premise can be, and has been, specified and detailed in different ways. The premise may read "the people accept T," or "most people believe so", or "it is widely believed that T", and more. Such differences matter in specific settings, but for my general discussion they do not. I only adopt the current formulation as it makes it more easy to show in Sect. 3 how the Condorcet jury theorem bears on the evaluation of arguments from popularity. (4) In real-life arguments from popularity, the reference group is often left implicit, but sometimes it gets mentioned, even if only vaguely: "most normal people believe…", "any decent person thinks…", "the whole of our country sees that…" When having to evaluate such arguments, the reference group cannot be left unexamined, and this underlines the importance of a correct interpretation of a real-life argument from popularity. In this paper, I will not discuss the issue of how to arrive at a justifiable interpretation of real-life arguments from popularity. (5) The conclusion, "T", is to be understood as short for something like that proposition T merits acceptance by the addressed opponent, or that T is true for all we (proponent and opponent) know, or that we (proponent and opponent) ought to commit ourselves to T on epistemic grounds in so-far as we have access to, and understand these grounds. This also applies to cases where T evaluates or prescribes some course of action or policy. (6) I only deal with positive versions of arguments from popularity, and we can see negative versions, where the popularity of the denial of a proposition provides support for the denial of that proposition, as a special case. This paper does not deal with cases where a lack of popularity is advanced as support for denying a thesis.
A natural distinction between two kinds of argument from popularity is between those that concern practical issues, such as issues of policy choice, and those that are not actionoriented, such as those dealing with the truth or correctness of some descriptive or evaluative claim. The idea that arguments from popularity are more easily to be evaluated as reasonable when they concern policy choice in a democracy has been supported (Walton 1999) as well as criticised (cf. Jansen 2020). But how about the presupposition shared by both positions: are such arguments really arguments from popularity?

Popular Support for a Policy
Are arguments from popularity more reasonable when they concern policy choice in a democracy? Johnson and Blair (1994) warn us not to confuse popularity with "majority rule" or "popular sovereignty" (see also Jansen 2020). This prompts me to zoom in on a possible ambiguity that results when in the argumentation scheme from popularity we specify "T" to something such as "Policy P should be implemented." Only some of the possible disambiguations of the argumentation scheme are producing real arguments from popularity. As a result, the merits and defects of appeals to the popular support for a policy do not easily translate to the merits and defects of arguments from popularity.
When we specify the argumentation scheme from popularity by replacing "T" with "Policy P should be implemented" we obtain the following candidate argumentation scheme: "Policy P should be implemented, because most people in group G accept that policy P should be implemented." Jansen sees the arguments that result from such a scheme as forming a special subtype of arguments from popularity, which she calls "deliberative populist arguments." She then develops an interesting critique of such arguments. In her view, they typically function as conversation-stoppers, and they are thereby fallacious, precisely in the context of a democratic deliberation. In Sect. 3, I will explain why I think that some real arguments from popularity can be reasonable and convincing arguments, also when they deal with policy choice, but here I contend that typical appeals to the popular support for a policy are, despite appearances, often not really arguments from popularity.
The phrase "should be" in the candidate argumentation scheme can be regarded as ambiguous. After all, there are situations where it is both natural to say that one endorses and also that one rejects that some policy "should" be implemented. Suppose, my government implements a policy that I personally disagree with on the basis of its being ineffective, having bad side-effects, and serving no worthwhile purpose. But suppose further that I also see that the policy results from a democratically impeccable procedure: it clearly has received the support of a majority of fellow citizens. In such a situation I could, without contradicting myself, say something like "well, it really is a shame, but this policy should be implemented" or "of course this policy should not be implemented when evaluated from a substantial point of view, but then, on the other hand, it should be when evaluated from a procedural point of view." As a result, the two occurrences of "should be implemented" in the candidate argumentation scheme can be disambiguated in two directions. This ambiguity of "should" as it figures in the candidate argumentation scheme is related to the ambiguity that Koszowy and Walton identify in the term "authority" as it is involved in arguments from authority (2019). In their view, the scholarly treatment of arguments from authority and the related Fallacy Ad Verecundiam has, with some exceptions, been mainly concerned with "epistemic authority," whereas some such arguments appeal to a quite different kind of authority, which they label "deontic authority," and these latter arguments merit their own distinctive treatment. A person or institution has deontic authority over someone when this latter person incurs an obligation to act in a particular way, or to refrain from acting thus, if according to this former person or institution we should, or should not act in that way.
I discuss two of the four possibilities for disambiguating the candidate argumentation scheme. The first reading is a fully substantive reading, and the popularity of the policy is then used, to use Johnson and Blair's phrase, as "the criterion of its [i.e., the claim's] truth or plausibility" (Johnson and Blair 1994, p. 178). When we apply the Koszowy and Walton terminology to such appeals to popular policy, the proponent appeals to the epistemic authority of the majority.
The result of this disambiguation then results in an argumentation scheme that could be labelled "Substantive Argument from Popular Policy": "Policy P should be implemented, when evaluated from a substantive perspective, because most people in group G accept that policy P should be implemented, when evaluated from a substantive perspective." An example could be: "I wholeheartedly believe that banning this vaccine is the correct choice to make. After all, that's what most people believe." The conclusion does not regard the democratic legitimacy of banning this vaccine, but its correctness, or truth, or epistemic justifiability. This may be a very weak argument, but it really fits the constraints of the argumentation scheme from popularity, and thus forms a real kind of argument from popularity.
As an aside, I want to note that in special circumstances, such reasoning can become highly convincing. Suppose that the issue is something like "should we-in the Netherlands-abandon Dutch as the national language, and change to Frisian?" Suppose further that almost everyone in the Netherlands would welcome this change, and that no stakeholders abroad object to the proposal. Then the popularity of the new language policy could make it simply "true" (or "worthy of acceptance" or "correct") that is a change for the better. When there is no state of affairs out there as a touchstone, then a collection of people can be seen as not only politically, but also as epistemically sovereign, and in such cases popularity adds epistemic value to the conclusion.
But, when talking about the democratic value of appeals to popularity, most scholars seem to have in mind a quite different, procedural disambiguation. which we can label "From Substantive to Procedural Acceptability": "Policy P should be implemented, when evaluated from a procedural point of view, because most people in group G accept that policy P should be implemented, when evaluated from a substantive point of view." Here the proponent clearly appeals to the deontic authority of the majority judgment, rather than any possible epistemic authority. An example could be: "Banning this vaccine is the democratically legitimate choice to make, because most people wholeheartedly believe that banning the vaccine is the correct and prudent choice to make." This way of reasoning has some plausibility, even though Jansen is right in critiquing it as typically functioning as a conversation stopper. But it is not an argument from popularity, in our sense. Because, what the people are said to accept is not the same as what the proponent is defending. Yes, there is an appeal to popularity. But no, it is not the popularity-of-the-thesis that is advanced as support for the thesis. Instead, the popularity of the substantive correctness of the policy is support for the legitimacy of the policy.
Having sorted out some elements of the concept of argument from popularity, and how that argument connects with appeals to the popular support for policy, I now turn to the issue of assessment.

Are Arguments from Popularity Rationally
Persuasive?
What is the value of arguments from popularity? Can they provide a probatively relevant, or even a sufficiently convincing, reason to accept its conclusion, within the setting of a critical examination dialogue? Can it ever be a good idea to adopt the argumentation scheme from popularity as a prima facie correct kind of defeasible reasoning? As we have seen, this seems to be a controversial issue within argumentation studies. Strong critics consider them as intrinsically fallacious, either for being devoid of epistemic value (Godden 2008) or for being dialectically inadequate (Jansen 2020). There are also enthusiasts, who defend that within appropriate settings arguments from popularity have genuine argumentative worth, on probabilistic grounds (Dedrick 2019;Hahn and Hornikx 2016). Finally, there are mild critics, who consider arguments from popularity to be risky and typically unconvincing or even fallacious, yet not intrinsically so-allowing for the possibility that in special circumstances, such as when deliberating about policy, arguments from popularity have a positive role to play (Walton 1999;van Eemeren and Grootendorst 1992;Johnson and Blair 1994;Freeman 1995). 3 I will contend that arguments from popularity can offer genuine argumentative support, both within and outside of contexts of practical deliberation, so that it need not be bizarre to adopt the underlying argumentation scheme as a prima facie persuasive, albeit defeasible, way of reasoning. But I will also explain that in normal argumentative situations, they at most have minimal probative relevance, so that their premises only provide an itsy-bitsy of argumentative support for their conclusions, because arguments from popularity typically do not provide an answer to the concerns of the opponent.

The Reliability of Others
Whether a specific argument from popularity is convincing depends on details about the reference group in which the thesis is popular: how reliable, on average, are the members of this group as far as the matter at issue is concerned? That somebody is reliable to a certain degree with regard to some proposition, I take to mean that this proposition has a probability of being true or acceptable to that same degree, if that person asserts the proposition. The background can be that we trust the person at hand in what he says, so that we assume him to be well-intentioned and that we rely on his discharging the commitment to speak sincerely (cf. Hawley 2012, pp. 10, 17). Or it can merely be that a person is reliable by having deferred her judgment to that of a credible opinion maker, or to a top scientist with an excellent track record. I distinguish three kinds of argumentative situations: (1) where the proponent has no, or not sufficient information available about the (average) reliability of the people in the reference group; (2) where the proponent has information about the (average) outright reliability, for example where a case can be made that with respect to the issue at hand the members of the reference group can be taken to adopt a correct judgement in about two of the three cases; and (3) where the proponent has information available about the (average) reliability relative to the addressed opponent, for example information that shows that members of the reference group are on average equally well-informed about the issue as the addressed opponent is.
In the first case, the proponent lacks (sufficient) information about the reliability of the members of the reference group mentioned, or taken for granted, in the argument from popularity. As a result, the proponent will be unable to adequately answer any of the opponent's questions regarding the reliability of the people referred to, so that the proponent's argument from popularity remains without any further support in this respect. An argument from popularity in any such situation can be called the blatant version of this kind of argument (Johnson and Blair 1994, p. 177).
Godden's position seems to be that only blatant versions are genuine arguments from popularity, which makes his view that arguments from popularity are hardly, if at all, epistemically useful (2008, p. 108) a plausible position. In his view, any additional information about the reference group must be seen as an added premise that changes an argument from popularity to something richer than a mere argument from popularity. He develops his position in distinction from Walton's view, who holds that adding specifics about the reference group may bolster the argument from popularity, and on the basis of how arguments from popularity get bolstered he distinguishes a number of subtypes of what he labels "ad populum arguments", such as the "position to know ad populum argument" (Walton 1999, Chapter 7;Walton et al. 2008, pp. 126-131).
In line with Walton's position, the argumentation scheme discussed in Sect. 2.1 includes a variable for the reference group, so that any specification of the reference group can be seen as included in an argument from popularity, rather than as an premise optionally to be added. But instead of regarding any useful specification as itself a bolstering of the argument, I believe that it would be more correct to regard the specification as a move that only prepares the proponent to further justify, and thus bolster, the argument from popularity by means of a separate, subordinate argument. The result then would be complex argumentation, composed of an argument from popularity, including the specification of the reference group, and a separate, subordinate argument that attempts to justify the connection premise (i.e. a proposition that expresses the particular argumentative connection between the other premises and the conclusion) of that argument from popularity by appealing to the alleged reliability of the members of that specific reference group.
For example: "This vaccine is unsafe, because most people working in health care seem to think so. The opinion of a majority of people working in health care provides a serious reason for my thesis, because these people have the requisite disciplinary background and hands-on experience." In this manner, I steer between Godden's overly narrow and Walton's overly broad account of "argument from popularity." 4 Now I can return to the issue of evaluation. Indeed, not much can be said in support of blatant versions of arguments from popularity in those settings where the proponent cannot appropriately specify the reference group. Only when the proponent is in a position to advance a subordinate argument in which he makes it plausible that majority acceptance is a sufficiently strong sign of acceptability, he can successfully respond to the opponent's critical probes of the connection between popularity and acceptability. Two distinct strategies may help the proponent in this regard: first the strategy by appealing to the outright reliability of a reference group; second the strategy of appealing to the relative reliability of the reference group. I discuss these in turn.

Outright Reliability
What I have labelled the enthusiasts for arguments from popularity, Hahn and Hornikx (2016), as well as Dedrick (2019), all deal with cases where there is information available, real or alleged, about the outright reliability of the members of the reference group. In those cases, the proponent could make it plausible that, on average, the members of the reference group are more likely to be right than wrong (to some degree). They then use probability theory 4 Walton, Reed and Macagno distinguish between the basic form of the ad populum scheme, and nine subtypes (2008). Some of these subtypes can plausibly be re-interpreted as a complex argument made up from an argument from popularity, and an argument that supports the former's connection premise by pointing to the reliability of the reference group: "Position-to-know Ad Populum Argument" (p. 126), "Expert Opinion Ad Populum Argument" (p. 126), and "Deliberation Ad Populum Argument" (p. 127). Other subtypes, I would be inclined to see as non-argumentation, because the reasoning plausibly expresses a causal process affecting the addressee's mind (See Sect. 2.1), namely: "Common Folks Ad Populum Argument" (p. 129), "Rhetoric of Belonging Ad Populum Argument" (p. 130), "Snob Appeal Ad Populum Argument" (p. 130) and "Appeal to Vanity Ad Populum Argument" (p. 131). In other words, this account insufficiently heeds the insight that "argumentum ad populum" comes in two very different forms: it can refer to a form of non-argumentation, but also to a failure of reasoning (Krabbe and van Laar 2015). The two remaining subtypes, "Moral Justification (Excuse Subtype) Ad Populum Argument" (p. 127) and "Moral Justification Ad Populum Argument" (p. 128), I regard as too ambivalent between an epistemic and a deontic reading as to allow for classification. to explain when and how arguments from popularity can be strong arguments. 5 In their own words, Hahn and Hornikx provide a Bayesian foundation for the evaluation of arguments, among which arguments from popularity. In their view, the purpose of an argument is to increase the level of rational belief in the thesis. The strength of the reasons R, offered in support of a thesis T, is then identified with a conditional probability, namely the probability that the thesis is true, provided that the reasons are true: Prob(T/R). Such a conditional probability, in this case the argumentative strength or epistemic merit of the argument, is itself also matter of degree.
Bayes' theorem then shows how to determine the conditional probability that in this way constitutes the strength of the argument. The bottom line is that the strength of the reasons in support of the thesis, called the posterior, depends on three items of information. First, the prior probability of the thesis, independent of the available reasons: Prob(T). The higher the prior probability of the thesis, the more quickly the reasons provide strong argumentative support for the thesis. Second, the probability of the reasons if the thesis were true: Prob(R/T) (compare the sensitivity of the evidence to the truth of an hypothesis when R stands for evidence and T for an hypothesis). Third, the probability of the reasons if the thesis were false: Prob(R/ ~ T) (compare the false positive rate when R stands for evidence and T for an hypothesis). The probability that the reasons are true if the thesis is true must then be higher than the probability that the reasons are true if the thesis is false. The theorem specifies the relations exactly: Some natural objections against this approach to argument strength allow of an easy answer. First, one could object that argumentation is not about degrees of belief because it deals with commitment and that's a yes-no affair. But that could be fixed by regarding commitment as a degree of rational belief above some threshold. What is more, a case can be made that commitments are gradual. Second, we lack the information for assigning any such specific values to the requisite items of information. Yes, but we can make educated guesses, or work with imprecise values within certain intervals. Third, one may challenge the idea that the strength of an argument also depends on the initial probability of the thesis. This can be answered by pointing to the fact that argument strength here is seen here as specified for a context where a specific thesis is at issue. As a result, when an agent assigns a probability to a thesis T that is very close to (or even at) the threshold of acceptability, then in that specific context even a very weak reason R, such that the probability that the reasons are true-if-the-thesis-is-true is only minimally higher than the probability that the reasons are trueif-the-thesis-is-false (or even: zero weighty argument, where the two variables are equal), suffices to tilt the agent's belief in T above that threshold of accepting it.
What results when applying this approach to arguments from popularity? How to determine the argumentative strength of an argument of the form: "T is popular in group G, therefore T"? It depends on the answer to the following questions, as specified by the above theorem: -How probable is "T", in itself? -How probable is "T is popular in group G" if T is true? -How probable is "T is popular in group G" if T is false?
(From now on, I omit the phrase "in group G") Thus these are three critical questions with which to critically examine the argumentative strength of any argument from popularity, and Hahn and Hornikx show how they are specifications of the kinds of critical questions that can be made to apply to any application of any argumentation scheme. 6 A virtue of their approach is that it makes it clear how the strength of a specific argument from popularity depends on contextual details. If T is very improbable then plausibly the argument has insufficient strength. And if the people are likely to be collectively misled, then the argument also falters. But there are situations where it may provide a strong argument. For example, if the prior probability of T is 0,4, and the probability that T is popular if T is true is 0,66, whereas the probability that T is popular if T is false is 0,33, then when we apply Bayes' theorem, the strength of the argument is 0,57thus in such a setting: an argument to be taken seriously.
Does this make Bayes' theorem the foundation for the evaluation of arguments from popularity? 7 I will approach this question from the viewpoint of the dialogue participants: does it provide a foundation for their evaluation of these arguments? Yes, if they only mean to say that in any situation in which the requisite information is available this provides a way in which the proponent can try to make a case in support of the argumentative strength of his argument. But No, if they-less modestly-mean to say that without the Bayesian justification the proponent's attempt at rational persuasion is doomed to fail, or that any proper such justification can be understood as really an application of the Bayesian principles, or that typically it provides a sufficient justification. Before discussing whether the possibility of a Bayesian justification is a necessary or a sufficient condition for successfully defending the argumentative merit of an argument from popularity, I must deal with the issue of how to understand the Bayesian justification from the viewpoint of the participants in a critical examination dialogue.
We can understand the Bayesian as having two interlocked arguments, put forward by the proponent in response to challenges by the opponent. I will illustrate this by way of an (easy to generalize) example, where "&" represents the juxtaposition of reasons that together are presented as sufficient argumentative support for a proposition: Argument B is the subordinate argument that bolsters Argument A, the argument from popularity. Argument B is an interesting kind of argument, also because it suggests similar justifications for other argument types, and further because it connects to the rich field of Bayesian epistemology. But then, the proponent does not need to advance this very argument to convince an opponent who challenges a connection premise (i.e. A3), so that a Bayesian justification is not indispensable for the participants of a dialogue in which this foundational issue has arisen. Alternatively, the proponent can justify the connection by appealing to some relevant expert ("that what my teacher says"), or to an analogy ("the majority was right regarding that other issue"), or by way of some well-chosen examples ("look at her, her opinion is right"), or by means of statistical information ("this percentage in that group answered question such-and-so correctly").
One special way of providing a justification of premise B3 can be provided by yet another result in the intellectual history of probability, namely Condorcet's jury theorem-also discussed by Hahn and Hornikx. This theorem states that groups may outperform individuals as regards the correctness of their judgments. If under specific conditions a majority of people (the "jurors") judges a proposition to be true rather than false, then the probability of the majority being right is higher than the probability of each of the individuals being right, and the larger the group, the more the group outperforms each of the individuals. Under what conditions? First, the individuals make an all or nothing choice between exactly two rival options, such as between T's truth and T's falsity. Second, each individual judges the issue independently of others, and there are no individuals who simply follow what others do by deferring to their judgements. Third, each individual in the reference group is equally likely to be correct as any other. Fourth, each individual is more likely to judge correctly rather than wrongly, though the theorem holds even if they are right in slightly more than 50% of cases. In such settings, as Dedrick says, "Popularity is a truth tracker" and "It is not a fallacy to appeal to the popularity of a belief as evidence for its truth" (2019, pp. 158-159).
To elaborate further on our example, the proponent could justify Premise B3 by claiming that the relevant reference group has three members (jurors), each of which with a reliability of 60.7% when it comes to the issue of whether or not T. In that case, the proponent can contend that the probability of a majority of the members believing that T if T is correct is (about) 66%. 8 In short, if information about the outright reliability about the reference group is available then a Bayesian justification becomes a serious option for any user of an argument from popularity. In those cases, can a Bayesian justification be expected to provide a sufficient justification? I conclude this section by discussing two related considerations that suggest that it can only be of limited use in critical examination dialogue.
First, the point of engaging in an argumentative discussion is to examine an issue on the basis of considerations that the participants in the dialogue accept and understand. Nothing else than an improved understanding of the issue, and of the perspectives on the issue, should prompt an opponent to withdraw her critical stance, and to concede the proponent's thesis. But, an argument from popularity, even if the outright reliability conditions are ideal, at most yields the conclusion that the majority verdict is (probably) correct, without shedding any light on why the outcome is correct. For all the opponent knows, the members of the reference group could just be reliable by sheer luck. The common 8 Suppose that T is true, and that the reference group has just the three members A, B and C, each of which with an individual reliability of 60,7%. Then the probability that A, B and C all judge that T = 60,7% * 60,7% * 60,7% =22,3%; the probability that A and B judge that T but where C judges that not-T = 60,7% * 60,7% * 39,3% = 14,5%; the probability that A and C judge T and B not-T = 60,7% * 60,7% * 39,3% = 14,5%; the probability that B and C judge T and A not-T = 60,7% * 60,7% * 39,3% = 14,5%. These are all options in which the majority is right. Thus, the reliability of the majority judgment is 22,3 + 14,5 + 14,5 + 14,5 = about 66%. (I am following Dedrick's explanation of the Condorcet jury theorem (Dedrick 2019), with different numbers.). goal of the proponent and the opponent is not so much to arrive at a correct outcome, but to arrive at an outcome that both perceive and understand to be the correct outcome. 9 As a result arguments from popularity suffer from a lack of transparency problem. 10 Second, the opponent of a thesis typically has considerations that motivate her not to yield to the thesis, but to critically engage with it. These motivating considerations may be explicit and well-developed, yet sometimes they are only some uneasy feeling that something isn't all right. In order to develop argumentation that really addresses and rationally persuades the opponent, the proponent should try to tailor his arguments at the opponent by responding to whatever, vague or precise, makes her doubt the thesis. Typically, an argument from popularity does not speak to the opponent's concerns. Even in ideal outright reliability conditions, arguments from popularity suffer from a lack of responsiveness problem.
Both reasons for assigning at most a bit justification to arguments from popularity derive from the fact that when people adopt the position of a critic or an opponent, this involves more than just the expression of critical doubt towards the proponent's standpoint. In real-life controversies, people have doubts that relate to the way they understand and judge the various aspects of the controversy. In all such cases, there is something that underlies the critical position: possibly quite articulated motivations, yet at least some more intuitive feelings or intuitions. When the opponent succeeds in (partly) explaining her opposition to the proponent's thesis, the proponent's argumentation must in some way or other address these concerns in order for the argumentation to provide a genuine response. Transparency and responsiveness are natural requirements of argumentation. Any argument that suffers from a lack of transparency and a lack of responsiveness, whatever its other merits, flops in discussion.
A third consideration for lowering one's expectations regarding arguments from popularity has to do with the fact that the epistemic support of an argument from popularity only bears on the issue indirectly, rather than directly, but that consideration is best developed when dealing with cases where the proponent appeals to the reliability of people relative to the addressee.

Relative Reliability
Let's then turn to settings where the proponent appeals to the relative reliability of people. The epistemological literature on so-called "peer disagreements" typically deals with vignettes where the one person has a particular belief, B, but then learns that an "epistemic peer" dissents by having a contrary belief, say not B. That two persons are epistemic peers means something like that both individuals have more or less the same information at their disposal, and are more or less equally capable of rationally processing it. As a result, when the opponent learns that an epistemic peer believes some proposition not B, the opponent cannot dismiss the other's dissent with an appeal to his lack of information ("if only he would read a decent newspaper!"), and neither with an appeal to his lack of rationality ("if only he would think critically!"). There has been much debate about the issue whether an person who learns about a dissenting peer should conciliate by suspending judgment or by lowering her credence in her initial belief B, or whether it would be more rational to steadfastly maintain her belief (Frances and Matheson 2019).
For the purpose of this paper it suffices to mention that a serious case can be made for assigning probative importance to the fact that an epistemic peer dissents from you. According to Kelly's total evidence view (2011), for example, when a peer dissents this provides some special kind of higherorder evidence that needs to be taken into account, though without disregarding one's first-order evidence. The upshot of this moderate conciliatory view is that it is probatively relevant for the opponent to be informed by the proponent that somebody who is no less reliable than the opponent happens to believe that T, instead of being uncommitted to T or committed to not-T. The proponent could strengthen such reasoning in at least two ways: first, by trying to appeal to the opinion of someone who has more epistemic standing than that of merely a peer by being better informed or a better inquirer; second, by appealing to more than one epistemic peer (or epistemic superior), such as a majority of people in a reference group of epistemic peers (or superiors). If the protagonists of a conciliatory approach are right, there is epistemic value in arguments from popularity, even if the group members are only known to be reliable relative to the opponent addressed.
Could a justification of the argumentative connection of an argument from popularity by reference to the relative reliability of the reference group suffice in a critical examination dialogue? The two problems we identified when discussing outright reliability also apply to the current kind of justification. Despite the majority view having probative relevance, the proponent's reasoning is free of evidential substance and there remains the issue that in normal situations the proponent's argument lacks transparency and responsiveness. A third problem can be added.
The higher order evidence provided by disagreeing peers (superiors) need not directly affect the opponent's own beliefs. Instead, it could be reasonable to revise only the higher order attitude towards her belief (or commitment) that T by lowering the confidence that the initial credence (or strength of commitment) is correct. In practice, this could mean that the opponent is more willing and inclined to explore the issue anew in dialogue, yet without prematurely changing her credence (commitment) before that exploration has been carried out and brought to a conclusion. Thus, the higher order evidence can have a bearing on the opponent's attitude to her own belief (commitment) without necessarily changing anything to the belief (commitment) and its strength itself, except that it affects her confidence with respect to that credence (cf. Henderson 2021).
Similarly in a discussion. When being presented with an argument from popularity, it may suffice for the opponentinstead of yielding to it-to rethink her position. Imagine that I believe that our prime minister deserves yet another term and decide to vote in his favour. But then I learn that my well-respected neighbours decide to vote against this candidate. This could be a reason for me to spend a second thought on my position, and inquire further into my choice by having a chat with these neighbours. But there seems to be no immediate need to change my position and defer to their stance, nor to suspend my judgment, nor to assign any lower probability to the candidate's deserving another term. Thus, an argument from popularity is indirectly relevant by prompting the opponent to reconsider, but it suffers from the problem of having no direct bearing. This is also problem for appeals to outright reliability, and also in those cases the addressed opponent best judges that instead of prematurely adjusting her credence, she better lowers her confidence that her initial credence is correct.
To conclude this section on the epistemic merits of arguments from popularity: These arguments do provide support under conditions that are not outlandish. But there are also reasons to think that they only provide a bit of support when the dialogue participants wish and need to get to the bottom of a disagreement themselves, from their own perspectives, so as to approximate a resolution on the basis of considerations as they understand them.
Of course, just one weak tiny argument can be important, and even decisive, if that is what tilts the balance. Walton gives an example where that is the case. According to his "maxim of non-disputativeness", the weak support provided by an argument from popularity suffices when the proposition at hand is not very controversial, so that a weak argument from popularity can be helpful after all (1999, pp. 238-241). Other situations can be added, for example where the issue is highly controversial yet the pros and cons are more or less in balance. Here, I can set this issue aside and conclude that arguments from popularity can provide serious contributions to critical examination dialogue, but that there are reasons for assigning them at most minor argumentative weight.

The Fallacy of Popularity
In the previous sections, I have contended that, instead of being dismissive of arguments from popularity, we should acknowledge their potential, yet limited merits. In line with dialectical approaches to fallacy, I use the term "fallacy" for moves that undermine the point and purpose of critical examination dialogue. That an argument from popularity has intrinsic weaknesses is insufficient ground for charging its proponent with having committed a fallacy, because weak arguments that do not add much to the rational persuasiveness of the proponent's case need not thwart the exchange. But here, I hasten to add that arguments from popularity easily fall prey to fallacy. How to distinguish between situations where an argument from popularity is weak yet dialectically harmless and situations where they can rightly be charged with being harmful and fallacious?
It is possible distinguish quite a number of different sorts of perils, and when characterising them I will employ the rule-based approach by van Eemeren and Grootendorst (2004). It is possible to advance a variety of fallacy charges against a proponent who, allegedly, uses an argument from popularity fallaciously. 11 What to conclude from this? One thing is that there's not much unity in the way arguments from popularity can decay to fallacies. This then prompts the question whether it makes sense to include anything like "the Fallacy of Popularity" in our normative theory. But, even though the Fallacy of Popularity proves to be a mixed phenomenon, I will contend that it can make sense to charge an interlocutor with having committed that fallacy, and that therefore critical opponents should have this fallacy charge in their inventory of dialogue moves.
I will make my case with a real-life example. It concerns a fragment from a speech from Geert Wilders, a right-wing, populist member of the Dutch parliament's second chamber. He held the speech during the general budget debate of 2015, an occasion for him to attract attention and to hit the news with one-liners. The speech dealt with the high number of asylum seekers from war-ridden Syria. Wilders party, the Party for Freedom (PVV), strongly opposed the reception of these people, mainly for the reason that many of them were Islamic, which was framed as a threat for the country's national identity: I said so yesterday: if anything, it has become clear during this general budget debate that this government and this chamber [the second chamber] no longer represent the Dutch people. An enormous catastrophe is coming towards us. The whole of the Netherlands feels it. The whole of the Netherlands sees it. The whole of the Netherlands is crying out for action. But no action is coming from the chamber or the government. (Handelingen der Staten Generaal 2015).
One element of this fragment is the standpoint that it is really the case that a catastrophe consisting of the arrival of asylum seekers from Syria is imminent, and that what justifies the correctness of this thesis is that the whole of the Netherlands feels and sees that this catastrophe is heading their way. I focus on this specific argument from popularity concerning a standpoint that is partly descriptive and partly evaluative, and I will disregard other embedded reasonings, such as the policy oriented argument (the people crying out for action) or the argument chastising the colleague politicians (this government and chamber do not represent the people). How could a critic object to the dialectical legitimacy of that specific argument from popularity, and thereby raise a fallacy charge?
I start from the pragma-dialectical idea that fallacies can be understood as specific kinds of violations of rules for critical discussion (van Eemeren and Grootendorst 2004, pp. 190-196). From the perspective of dialogue participants, this implies that any fallacy charge must be capable of being explained or justified as being a specific violation of at least one of the rules for critical discussion. How about a setting where the proponent presents an argument from popularity and the opponent responds by charging him with committing the Fallacy of Popularity? How could a metadialogue (Krabbe 2003) starting with the charge of the Fallacy of Popularity reasonably develop?
In what follows, it becomes clear that the fallacy charge at hand can be sustained in various ways. The proponent of the argument from popularity can be charged with having violated a number of rules for critical discussion, and none of these violations is exclusively reserved for arguments from popularity. In line with this, I will not develop any theory regarding the proper nature and definition of the Fallacy of Popularity as being a specific kind of violation of one of the rules for critical discussion. Instead, I will contend that the concept of the Fallacy of Popularity serves the practical purpose as an entry point for a metadialogue about the merits or defects of some specific argument from popularity. If the dialogue participants want to get at the bottom of that issue, they need to turn to a more fine-grained terminology, such as for example provided by the pragma-dialectical rules for critical discussion. This will enable the opponent to explain and justify in a number of different ways any initial charge to the effect that the Fallacy of Popularity has been committed.
I will discuss five different ways in which the opponent can justify her initial charge that the proponent's argument from popularity amounts to a Fallacy of Popularity: (1) she can claim that it forms an instance of an inappropriate argument scheme, in which case the proponent violates the Argument Scheme Rule: "Standpoints may not be regarded as conclusively defended by argumentation (…) if the defense does not take place by means of appropriate argument schemes that are correctly applied" (van Eemeren and Grootendorst 2004, p. 194);(2) or that it amounts to an incorrect application of an appropriate argument scheme by presenting a problematic reason as a matter of course, in which case the proponent violates the Argument Scheme Rule in a different way, and also violates the Starting Point Rule: "Discussants may not falsely present something as an accepted starting point or (…)" (p. 193); (3) or that it is an incorrect application of an appropriate argument scheme that presents a problematic connection premise as a matter of course, so that the proponent in yet a third way violates the Argument Scheme Rule and in addition also the Starting Point Rule; (4) she can claim that the proponent uses the argument from popularity to evade his real burden of proof, in which case he is said to violate the Obligation-to-Defend Rule: "Discussants who advance a standpoint may not refuse to defend this standpoint when requested to do so" (p. 191); (5) and finally she can explain that the argument from popularity functions as a non-argumentative intimidation tactic, thereby violating both the Relevance Rule, "Standpoints may not be defended by non-argumentation (…)" (p. 192) and the Freedom Rule, "Discussants may not prevent each other from advancing standpoints or from calling standpoints into question" (p. 190).
(1) The opponent could point out that the proponent employs an argumentation scheme (the argumentation scheme from popularity) that she does not accept, and that she regards as unacceptable for the conversation at hand. What is more, she could add, the proponent presents the scheme as unproblematic, thereby falsely suggesting that the reasoning has more credit than the dialectical situation warrants. In our example, Wilders could be told that it is up to the representatives themselves to weight all the substantive considerations and how they bear of the various policy options, and that any such argument from popularity, and thereby the use of the very underlying scheme, is inappropriate in this conversational context. In the pragma-dialectical theory, this charge amounts to the claim that the proponent has violated the Argument Scheme Rule by using an-in that context-inappropriate argumentation scheme. 12 (2) The opponent may point out that, even though there may not be anything intrinsically wrong with the argumentation scheme in the conversation at hand, the scheme has been applied incorrectly by having a premise that is false or doubtful. This kind of criticism can be further specified in two directions, depending on the specifics of the case. Characteristic of contemporary populism as a political movement, is that it claims that there exists something like "what the people want," and that their spokesmen have privileged access to this, and thus can act as the mouthpiece of an allegedly unified will of the people. Arguments from popularity that include such a claim as a premise are fallacious when the highly questionable, if not false, premise gets presented as a matter of course. Wilders could be charged with falsely representing his view as being supported by almost all Dutch citizens, and that is what actually happened in the follow-up of the debate, where it was pointed out that a majority of citizens actually voted for representatives who favoured a much more lenient policy regarding asylum seekers. We could label this mistaken usage of the argument from popularity a populist fallacy. 13 The proponent may (implicitly) appeal to the idea that the people who support his thesis have independently arrived at that thesis, so as to boost the epistemic value of that majority view (in line with the reasoning behind the Condorcet's jury theorem). But possibly, the persons involved lack the required independent judgment, and merely parrot some propagandistic opinion leader, surrender to their herd mentality, or let themselves be manipulated by the economic or algorithmic design of online social media. Then, what looks like an epistemically impressive majority of independent 'jurors', each with their own positive individual reliability increasing the value of the collective judgement, may on closer inspection turn out to be the judgment of only a minority, thus spoiling the applicability of this appeal to the wisdom of the crowd. Wilders could be told that in so-far as citizens reject sheltering asylum seekers, this is in part due to a sustained propaganda campaign. In addition, the condition of the reference group members having positive reliability (outright or relative) may not be satisfied. Possibly, they judge things badly, or they are epistemically inferior to the addressed opponent. Note that the jury theorem also has a negative side, and when the members of the reference group are more probably wrong than right, then the likelihood of the majority being right swiftly reduces to zero with increasing group size.
When in such cases a problematic premise is presented as unproblematic, the opponent can charge the argument from popularity as fallacious. The opponent could provide a two-fold diagnosis. First, she could explain that the proponent applies the argumentation scheme from popularity, which in itself need not be bad thing, and that he does so in an incorrect way, by including a premise that is unacceptable. Thereby, the proponent violates the Argument Scheme Rule, but in a different manner than before. Second, she could point out that this unacceptable premise is presented as a matter of course, which violates the Starting Point Rule. 14 (3) Freeman stresses that the limited degree of support that popularity can provide, can easily be inflated (1995, pp. 266-267), in which case the proponent of the argument from popularity can be charged with committing the Fallacy of Hasty Conclusion (Johnson and Blair 1994, pp. 70-75). The proponent may assign a higher probability to his thesis than is warranted by the level of popularity and the (average) reliability of the reference group. Or, the proponent may ask the opponent to assign too much weight to the higher order evidence provided by dissenting peers, or too little to the available first order evidence counting against his thesis. Or, the proponent may act as if the critical consideration that motivates the opponent to challenge the thesis no longer requires a response. In all such cases the argumentative strength of the argument gets inflated, or so the opponent may explain when charging the proponent with the Fallacy of Popularity. Wilders could be told that even when all these people do believe that the said catastrophe is imminent, the conclusion cannot be drawn that that we cannot but accept that the catastrophe is, indeed, imminent. After all, even when on average these citizens have some reliability, the degree of certainty with which the conclusion gets drawn may not be warranted.
Again, the opponent could provide the same twofold diagnosis, but now with a focus on the justifying force of the combined reasons in support of the conclusion, rather than on the one popularity premise. Thus, she could explain that the proponent applies the argumentation scheme from popularity incorrectly by relying on an argumentative connection that is too weak. Thereby, the proponent violates the Argument Scheme Rule. Because the sufficiency of the argumentative connection has been presented as a matter of course, the proponent also violates the Starting Point Rule.
(4) The opponent could claim that the proponent's argument functions to supress an in-depth discussion about the merits of the thesis. The indirect evidence provided by the beliefs of people then is said to divert from the examination of substantial evidence and considerations directly bearing on the issue at hand (cf. on the Fallacy of Diversion: Johnson and Blair 1994, p. 93). In Kelly's terminology, the higher order evidence provided by disagreeing peers can be misapplied by "swamping" the first order evidence (2011). In Jansen's terminology, the popularity of a policy gets misapplied when used to beat a discussion to death and to bypass genuine deliberation on the policy (2020, p. 362).
Wilders could be told that he is diverting attention from considerations such as the basic needs of people, human rights or the willingness of Dutch citizens to support refugees. His contribution can be regarded as a violation of the Obligation-to-Defend Rule, which requires a proponent not to evade his burden of proof: even if the argument from popularity provides some epistemic support, in these cases they also turn the attention away from considerations that (also) merit attention.
(5) The proponent can commit a Fallacy of Intimidation by advancing an argument from popularity (Johnson and Blair 1994, pp. 167-190). The argument may exert pressure on the opponent to yield to the thesis, and to defer to the proponent's point of view-motivated by a fear for being considered a social, or an epistemic, or a moral aberration, the odd one out, difficult and uncooperative, impertinent and disrespectful, abnormal, and so forth. What does the trick is the veiled threat of leaving the addressed opponent alone and abandoned, outside of the community of people who are in their right mind and have a proper (moral or epistemic) sense-if she does not yield to the thesis. In such cases, the argument from popularity exploits the human need for recognition, and damages the critical self-steered thinking and decision making of the addressed opponent. This intimidating function of arguments from popularity is a big danger, and Johnson and Blair are right in attending to this aspect. What makes this type of derailment worse than any of the other ones discussed earlier, is that the fear of being seen as an outcast may discourage the opponent from identifying and pointing out the fallacy in the first place. Wilders could be criticized on account of his trying to blacklist people who tend to empathise with war refugees, and would be willing to accept them as asylum seekers; after all, they apparently do not befit the Dutch community.
When the opponent wishes to point out that the proponent's argument from popularity functions as an intimidation tactics, she can admit that on the one hand the proponent offers an argument, so that he invites the opponent to inspect and assess his reason in support of his thesis. But then she can explain that on the other hand, due to the way the proponent frames and words his argument, the argument from popularity functions as a non-argumentative, manipulative, and primarily emotion-based device to causally affect the opponent, rather than to rationally persuade her.
Thus, the opponent can sustain her charge that the proponent has committed the Fallacy of Popularity, by showing that it violates two further rules for critical discussion. The proponent can be criticized as violating the Freedom Rule, according to which it is impermissible to prevent others from advancing their positions by exerting pressure on them. And given the causal role of the emotions of wishing to belong to a community, of being appreciated by peers, and of the fear of being ostracised, the opponent may equally well point out that the proponent violates the Relevance Rule, according to which (among other things) it is not allowed to merely exploit pathos when argumentation is required. 15 Even though arguments from popularity are epistemically interesting and potentially even strong arguments, it is a type of argumentation that easily derails. What is more, there are at least five quite different ways in which such arguments can turn out to be fallacious. What to conclude from this?
First, there is a question about what policy to recommend to any company of arguers regarding the ground rules to follow. Should they, as a matter of an argumentative "better safe than sorry" policy, adopt a dialogue rule that bans all arguments from popularity? If arguments from popularity are normally weak and if they easily decay to so many kinds of fallacy, would the proper policy for any company of arguers then be to ban such arguments altogether? Even if a particular act of presenting an argument from popularity may function well in some special critical examination dialogue, when dialogue participants need to decide between a rule that allows such arguments and a rule than bans such argument, it is the latter that must be regarded as producing the best consequences-or so one could argue. However, there is a serious consideration to be brought against this style of argumentative rule-utilitarianism.
The rationale for engaging in an argumentative discussion is that there is some dialectic between the proponent and the opponent. It is by engaging in a limited form of competition that they cooperate in their quest for acceptable beliefs, opinions or standpoints (cf. Dutilh Novaes 2021; cf. Stevens and Cohen 2021). After all, if the opponent fails to be sufficiently critical, there is the risk that an incorrect proposition gets accepted, whereas if the proponents fails to be sufficiently persuasive, a correct thesis may risk to get rejected. What is more, by allowing participants to, as it were, push the limits of reason in their advocacy, they may outperform themselves, and thereby enhance the quality of the argumentative exchange. A dialectical division of labour thus may lead to an outcome with increased epistemic significance. Absent a clear, univocal and mutually accepted criterion for determining the fallaciousness of an argumentation scheme, such as the one from popularity, the dialectical method is best served by not banning such moves, but by allowing the proponent the freedom to use them, and leave it up to the arguers themselves to check whether any limit of reasonableness has been transgressed, for example by means of advancing a fallacy charge and examining the justifiability of that charge.
Second, there is a question about what fallacies to include in a normative theory of argumentation. Arguments from popularity can be unreasonable, and dialectically inadmissible, in a variety of ways. This could be seen as a reason for dispensing with the concept of the Fallacy of Popularity altogether, and for either using a much more fine-grained fallacy classification, or else evaluate specific arguments from popularity directly in terms of whether they comply with the rules for critical discussion (or with other evaluation criteria, such as acceptability, relevance and sufficiency: Johnson and Blair 1994, p. 54). Or it could be seen as adding to a more general scepticism regarding fallacy theory, based on the finding that when we apply initially plausible scheme-based or rule-based definitions of fallacies to real life arguments, it often is hard if not impossible to maintain one's view that the fallacy has indeed been committed (cf. on 'the fallacy fork ': Boudry et al. 2015).
But we do not need to become fallacy sceptics, because we do not need to start from overstrained expectations about fallacy theory. We can take a practical approach towards any classification of fallacies, even when taking a more principled stance on the importance of fallacies and on the definition of "fallacy." The concept of a popular opinion is a highly salient one in public discourse, and it is easy to find both cases where real-life arguers use it to support their own opinion, and where they use it to diagnose what they regard as fallacies of popularity. In public discourse, people seem to have an ambivalent stance towards the argumentative merits of popularity, but the term is a salient part of the argumentative meta-language used by real-life arguers. The concept of the Fallacy of Popularity is useful, because it allows the opponent to say (quoting Johnson and Blair): "Hold on, are you saying that because everyone in your group believes it, therefore it is true?" (Johnson and Blair 1994, p. 177). Having this multifarious fallacy charge at one's disposal is useful, for it provides a salient entry point for a more finegrained evaluation, and if the need arises, the charge "Fallacy of Popularity!" can be specified along the way. 16

Conclusion
This paper dealt with the conceptualisation of arguments from popularity, with the epistemic assessment of such arguments, and with the question in what circumstances the proponent of any such argument could reasonably be charged with having committed the Fallacy of Popularity.
After having presented an argumentation scheme based definition of "argument from popularity", I showed that not all argumentative appeals to popularity are really arguments from popularity, and that some quite plausible democratic appeals to popularity fall off the wagon.
Then I contended that arguments from popularity can have real epistemic merit, because in some circumstances people can be regarded as reliable, so that an appeal to the popularity of a thesis may become a serious reason for that thesis. But in the context of a critical examination dialogue, arguments from popularity have a number of serious drawbacks, and as a result, there is neither cause for enthusiasm about such arguments, not for being dismissive about them.
Finally, I distinguished five different ways in which an argument from popularity may be fallacious. At the same time, for practical reasons I stopped short of concluding that we better drop the concept of the Fallacy of Popularity.