One common way of understanding arguments, and logical fallacies, is to view them as abstract bits of discourse, as propositions collected together that imply some conclusion. This is the way many popular informal logic or critical thinking texts discuss them. On this view, logical fallacies are patterns of reasoning that are in some way defective. To speak very generally, the defect is internal to the argument structure or scheme. An ad hominem for instance, is an argument form wherein one misdirects criticism to a person making an argument rather than the argument the person makes. It has long been understood that such treatments of arguments and fallacies in particular are inadequate (Hamblin 1970). In the first place, what may look like a fallacy in one instance seems, on closer reflection, completely reasonable in another circumstance. This means that it is inadequate to characterize informal arguments outside of the dialectical contexts in which they take place. The context is the dialectical tier wherein on anticipates and answers objections to one’s views or cases to the contrary (Johnson 2000, p. 165). However one understands the dialectical tier of argumentation, it is very difficult to represent it schematically (van Eemeren and Grootendoorst 2004 p.160). One prominent way to confront this difficulty is to represent the dialectical feature of argumentation as basically a kind of dialogue, as in pragma-dialectics or Walton’s dialogue theory. While according to pragma-dialectics fallacies can be understood as wrong moves in a dialogue, according to Walton and Krabbe (1995) the concept of dialogical shift is the only way to achieve a deeper understanding of many informal fallacies (p.108).
Walton’s concept of a dialogue is a normative one. But it is a normative program that takes into account how dialogues actually go and what their purposes are (Walton 1998 p. 29). Dialogues, as Walton and Krabbe (1995) discuss them, have three key elements. First, they have participants, often characterized as “proponent” and “exponent”. Second, the participants verbally engage each other for some purpose. One such purpose is for one person to persuade another of the truth of some claim. This is a persuasion dialogue (also called “critical discussion.”). There are (roughly) five other types of dialogue, as defined by their main purpose: negotiation, information-seeking, inquiry, deliberation, and eristic dialogue (which is an airing of grievances). But there are perhaps innumerably many kinds of mixed dialogue because each type of dialogue can involve others as sub-types. Third, formal dialogues are normative models, such that an exchange consists in a series of game-like moves (Walton 1998 p. 29).
Central to this idea is the dialogue shift. Since dialogues are normative models for conversation, i.e., for how conversations ought to go as defined by their purposes, errors will occur when participants illegitimately (i.e., without permission or warning) shift from one kind of dialogue to another. Since dialogues are modeled on the idea of a game, and games have rules, and different games have different rules, to shift in a dialogue, then, is like shifting from the rules of one game to the rules of another. To give a crude analogy, it would be like playing chess with someone who at some point decides to make their moves according to the rules of checkers (while claiming they’re still playing chess). Chess and checkers require two players, they share the same board, involve pieces of differing value, and rules of movement. Naturally, such a shift would be completely legitimate, and only a little odd, if both players agreed to the change (Walton and Krabbe 1995: 102–103). But if that were the case, the game played would no longer be checkers or chess, but something else. Dialogue shifts are evaluated retrospectively based on a couple of key principles. First, the dialogue shift should be consistent with the original goals of the exchange. To shift from chess to checkers, in the present case, maintains the objective of playing a board game. A shift is legitimate, in other words, if it provides another way of doing the same thing. A second, more critical, rule is that both participants need to recognize it and consent to it (Walton 1998 p. 205). By analogy, if one starts playing checkers during a game of chess without telling the opponent, then one has undermined the purpose of the exchange. Both of these considerations together show that we can mutually agree to undermine our purpose. If our goal is to play a game and we come to blows instead, we’re not playing a game any longer.
Walton and Krabbe envision dialogue shifting as a model for characterizing informal fallacies (1995: 2). Traditionally, a fallacy is “an argument that seems valid but is not so” (Hamblin 1970 p. 12). This definition naturally raises the question of what makes an argument valid in the first place. For dialogue theory, this is answered by considering the dialogue types to be normative models. Like games, dialogues that follow the rules are valid.
A more difficult, and more persistent, question for fallacy theory is how to characterize the seeming validity of fallacious arguments. How is it, in other words, that they are not just silly? Walton and Krabbe cite two factors in answering this question. The first is that fallacy schemes are often legitimate in the appropriate context (as in the case of the ad hominem noted earlier). The second reason is that there is a dialectical shift (1995: 114–115). The first of these factors has given rise to the now standard view that fallacious arguments are context-dependent. Formally or structurally identical arguments can be interpreted differently depending on the context. To return to the example of the ad hominem, a person’s personal conflicts can be relevant considerations in the right context, and so the charge of ad hominem may be mistaken. The second factor leverages the shifting context to explain the success of the fallacy deployment. Because moves are fallacious in one context but appropriate in another, an interlocutor may be duped into taking them to be valid because appropriately used in the proper context. Critically, in its fallacious use, the shift of contexts of dialogue is covert, unilateral, or not agreed upon. The heart of the deception, on this new account, is to shift the context of the argument in a way that the interlocutor doesn’t notice. This feature—the deception feature—which is a central part of the traditional account, retains its place in this dialogical account.
To illustrate this let’s first consider an example of licit shift in a persuasion dialogue. The main goal of a persuasion dialogue is to resolve a conflict of opinion by verbal means. In pursuit of this goal, each participant aims to persuade the other of their standpoint (Walton 1998 p. 37). In the course of such a discussion, it might be necessary for the participants to shift into another mode of dialogue, such as an information-seeking dialogue (where one participant questions the other) in order to serve the overall purpose of dispute resolution. A reason for this might be to discover what one participant takes their commitments to be. Two reasons show this shift to be licit: it is consented to and it serves the purpose of the original dialogue (to resolve the dispute by verbal means) in helping to define the dispute in the first place. One thing that makes dialogue shifts difficult to detect at times is that the various kinds of dialogues have overlapping purposes. As we’ve just seen, an information-seeking dialogue can share the purpose of dispute resolution, so it is not surprising when a dialogue shifts from persuasion to information seeking. Such a shift might be illegitimate, however, if the inquiry is meant to trap the respondent into giving an embarrassing answer, as is the case with the fallacy of complex question. Such a move might be legitimate in another circumstance (for instance, interrogating a prisoner of war).
With this understanding of dialectical shift, or dialogue shift, we can see how this applies to Bothsiderism, where the concept of dialogue shifting would seem to be perfectly descriptive. As we’ve seen above, Walton and Krabbe call the “Fallacy of Bargaining” when one attempts “to replace an offer for an argument” (1995 p. 104). In a very general sense, the fallacy of bargaining occurs when a critical discussion or persuasion dialogue illicitly slips into a negotiation. As we’ve discussed, there is more than one way this can occur. Given a dialogical approach to fallacies such as that of Walton and Krabbe, the most direct way for this to occur is when one participant in a critical discussion demands of the other that they meet half-way, or compromise, on some standpoint.
The trouble, as we see it, is that this explanatory approach to the phenomenon is lacking an account of how someone would make the error. Consider Walton and Krabbe’s example for the Fallacy of Bargaining: the doctor tells the patient that he must not drink or smoke, and the patient tries to bargain the recommendation down to quitting smoking and a few drinks here and there. The problem is that the example is better as a joke between doctor and patient based on the shift in dialogue—as we might see the number of gods going to splitting the difference as a kind of humorous tale, too. That the reasoning is good in the context of negotiation does not explain the fact that the shift from critical discussion to negotiation was not detected. This seems a harder error to explain, as (again) the fact that the doctor-patient case works better as a joke than as an example of fallacious reasoning.
A better explanation for the patient’s reasoning in the case is one captured by a meta-argumentative approach. The doctor has given the patient some sobering news and a consequent directive, and the patient has indirectly given the doctor feedback on that directive—that it is not one he is likely to follow, given that it calls for total abstinence. A limit condition on advice is that the person who must take it can follow it, and the patient’s feedback is that it’s unreasonable to think that he will be able to take the advice of total abstinence. However the recommendation of cutting smoking out and reducing alcohol consumption is more plausible and in the spirit of the doctor’s original directive. Read as such it is not a dialogue shift to negotiation. Or, if read as negotiative, it is only on the surface structure of the exchange, but its deep structure is one devoted to deliberation about an effective and practicable plan. The patient, with the joke on the level of a dialogue shift, has introduced the meta-argumentative reminder that doctor’s orders must be ones the patient can follow. The patient’s feedback, then, is that by his lights, the doctor’s advice of abstinence is too extreme. And what makes the case fallacious, then, is not the dialogue shift, but whether the patient’s counter-proposal accurately represents whether the doctor’s reasons were for total abstinence only or there in fact was room for moderate alcohol use. If the reply is not attentive to the reasons for total abstinence only, then going meta-argumentative is only a diversion. And if the reply is attentive to the doctor’s reasons and those reasons do have room for moderacy, then going meta-argumentative by invoking the role of practicability (through the surface structure of a shift to negotiation) is actually argumentatively appropriate. Sometimes the joke of negotiation is a good meta-argumentative move, but not as negotiation but as communication about what has been proposed within the norms of that discussion. The result, then, is that the meta-argumentative approach here can not only explain why someone would make such a move (and think it is a good one), but it can also identify fallacious and inappropriate instances. In contrast, dialogue-shifting can function only as an explanation for why the move is illicit, but it cannot explain why someone would be motivated to perform it nor can it explain when it might be appropriate.
In summary, there are two main reasons the concept of dialogue shift cannot explain bothsiderism. The first is that bothsiderism is often not an error within dialogues, it is rather a misbegotten judgment about the dialogues of others or the dialogue one is in. It is a meta-dialogical or meta-argumentative error. The second reason is that shifting to bargaining fails to capture the deceptive feature of bothsiderism, because such shifts are both obvious to participants of the dialogue and their shifting must not be noticed in order to be successful. Dialogue shifting does not capture the fact that fallacies are deceptions. The deceptive feature of bothsiderism, we argue, is an epistemic one in that it mistakes the status of disagreements for meta-evidence of some kind about the subject of the disagreement. Bothsiderism is a failure of evidence management, or more precisely, that of taking evidence about the dialogue to be meta-evidence about the matter of the dialogue. The problem is that this shift seems more obvious to dialogue participants, and can function more plausibly as a surface structure shift and as indirect meta-argumentative feedback about the state or purpose of the dialogue. Dialogue shift is better read as communicating that something’s gone wrong with the dialogue, so it’s a meta-argumentative piece of feedback. Again, in the doctor-patient interaction, it is more plausibly read as the patient giving feedback on the objectives of the dialogue by jokingly making a dialogue shift. Once we see the meta-argumentative function of the humor of the case as a comment on the dialogue’s frame, we can not only see why someone would use such an argumentative tactic thinking it good but we can also identify when it is appropriate and when it is inappropriate.