# Working Thesis Once we adopt the revised [[Sheryl Tuttle Ross]]-style account of [[propaganda]] defended by [[Bonard, Contesi and Marques]], [[Robin McKenna]]'s attempt to isolate a politically troubling "best case" that is not propaganda becomes unstable. Either McKenna's "best case" is idealised so far that it is epistemically non-defective in the broad Ross sense, in which case it is not propaganda and arguably benign; or the "best case" remains politically troubling in the way McKenna wants (scale + concentration dominating the marketplace of arguments), in which case it predictably produces [[epistemic defectiveness]] at the campaign level and so is propaganda after all. --- # Section 1: McKenna's Project This section presents McKenna's paper and his central worry, ending with a clear statement of why he thinks the best case is troubling without being propaganda. Begin with McKenna's starting observation: our intellectual lives are shaped by external factors; propaganda and indoctrination are obvious bad cases; persuasion is supposed to differ from overt coercion but remains philosophically suspicious. Present the two extremes he rejects: [[Luciano Floridi]]-style hyper-persuasion alarmism versus scepticism that AI changes nothing. Emphasise McKenna's central claim—the key worry is not superior persuasiveness but scale and concentration of capacity. Quote the passage: "The fundamental reason why we should be worried about AI persuasion is not that AI has, or soon will have, significantly more persuasive power than humans. The reason we should be worried is that any AI persuader or AI-generated persuasive content is (a) relatively easy to roll out en masse and at scale when (b) you have the resources (financial as well as technological) to create it in the first place." Give a brief account of the empirical picture: LLMs are often as persuasive as humans; personalisation can increase effectiveness; but persuasion effects are typically modest. There may be inherent limits on persuadability given human [[epistemic vigilance]]. Then turn to McKenna's conceptual move. He brackets misinformation and deception and focuses on a "best case" in which AI is used to offer genuine arguments and reasons aimed at [[rational persuasion]]. The point of this move is to show that we should worry even when the outputs are framed as rational argumentation rather than deception. Present the "sophistry" framing: AI persuaders are like the ancient [[Sophists]]—arguers for hire who can generate arguments for any conclusion regardless of commitment to its truth. Quote the key passage: "The purveyors of persuasive AI technologies can generate huge numbers of 'arguers for hire' (AI agents) or huge amounts of 'persuasive content' within minutes, with any dilution in the quality of arguments often more than made up for by the massive increase in quantity." McKenna introduces the "marketplace of arguments": the pool of arguments that provide the materials we use to discuss moral, political, and societal issues. A healthy marketplace would supply a representative balance of arguments on different sides. The danger is that a small number of actors (primarily big tech firms) can dominate this marketplace, shaping what arguments are available, salient, and widely circulated. The section ends by making explicit that McKenna does not treat his "best case" as propaganda. He mentions propaganda in the paper, but his worry is structural and political: concentrated power over the argumentative environment. His "best case" is designed to bracket deception and focus on a scenario where the outputs are genuine arguments. The question for the remainder of the paper: does this middle category—politically troubling but not propaganda—survive scrutiny once we apply an epistemic-merit account of propaganda? --- # Section 2: The Revised Ross Account This section presents the refined Ross-style conditions and clarifies the central role of epistemic defectiveness. Present the refined conditions as developed by Bonard, Contesi, and Marques. A communicative act counts as propaganda only if: (i) it conveys an epistemically defective message (broadly construed), (ii) it is produced by or on behalf of an institution or cause with intent to persuade for direct political ends, (iii) it targets a socially significant group, and (iv) the producer is aware (or should have been aware) of the defectiveness. The innovation central to this paper: "epistemically defective" extends beyond falsity. A message can be defective by misleading through omission, by generating false pragmatic inferences (including [[Gricean]] Quantity-style effects), by presupposing something false, or by inducing unwarranted beliefs or unwarranted connections among beliefs. The key example from Bonard et al.: cherry-picking true war evidence can still be misleading because the audience infers completeness from the presentation. This is epistemic defectiveness without falsehood. The term "message" should be understood broadly: it includes not just locutionary content but what is implicated, what is omitted, and what is done illocutionarily and perlocutionarily. State the conditional that drives the rest of the paper: if McKenna's "best case" is genuinely non-defective in this broad sense, then it is not propaganda on this account. Anyone who wants to classify McKenna's troubling scenario as propaganda must show that the troubling features—scale and concentration—generate defectiveness at the system or campaign level. --- # Section 3: The Dilemma and the Collapse This section presents the dilemma and argues for the collapse result. The fine-grained structure of the argument proceeds as follows. Begin by stating the dilemma. McKenna wants a middle category: scenarios that are troubling but not propaganda. On the revised Ross account, this requires scenarios that are troubling but not epistemically defective. The question is whether such scenarios exist in the space McKenna is concerned with. Present the benign horn. Describe an idealised scenario—call it the "Open Flood." An AI produces large volumes of arguments but is transparent, symmetric across sides, explicitly flags limits, ensures exposure to counterarguments, and is contestable. For any contested question, it generates arguments for all major sides; it labels arguments by side and strength; it forces symmetry; it includes a representativeness module that tells you whether you are seeing a representative sample; it gives equal prominence to competing positions; it explicitly warns against the inference "this is all there is." In such a case, there is no epistemic defectiveness: no omission, no misleading implicatures about balance, no crowding-out of alternatives. On Ross, this is not propaganda. But notice: it is also not troubling in McKenna's sense. The political worry evaporates along with the propaganda classification. If everything is genuinely epistemically above board, the system is a benign distributor of arguments—something like a well-functioning deliberation tool. Present the troubling horn. McKenna's actual worry concerns a different scenario: a small number of actors deploy persuasive AI at scale, creating armies of arguers-for-hire and dominating what arguments circulate. This is not the Open Flood. The question is whether this troubling scenario can remain epistemically non-defective. Argue that it cannot. The key move is to treat the relevant communicative act at the level of the campaign or ecosystem, not the individual utterance. On the revised Ross account, what is conveyed includes pragmatic implications and omissions that arise from distribution, prominence, repetition, and crowding-out. Individual arguments can be sound while the campaign-level communicative act is epistemically defective. Present the first route to defectiveness: omission-by-crowding-out. Even if counterarguments exist somewhere, flooding can make them practically inaccessible, producing a systematically incomplete evidential basis for belief formation. The audience ends up with a skewed sample of the argument space. This is misleading omission in the broad sense endorsed by Bonard et al. Present the second route: Quantity-style misleadingness about representativeness. Saturation on one side predictably licenses false inferences—"these are the main considerations," "there are no serious alternatives"—even without explicit denial. People infer from ubiquity that something is widely supported, socially central, or evidentially decisive. This is exactly the structure of the Gricean Quantity example: presenting true information in ways that make the audience infer something false about completeness. Present the third route: defective maps of salience and evidential weight. Controlling what becomes salient predictably distorts what audiences take to be decisive or important. Humans treat prominence and repetition as evidence of importance. A system that massively amplifies some true considerations while burying others can make audiences rationally but incorrectly update on "importance." The resulting beliefs are connected to other beliefs in ways that are inapt or unwarranted. Briefly show how McKenna's target cases satisfy the other conditions: institutional provenance (big tech actors, campaigns, interest groups), intent to persuade for direct political ends, targeting of socially significant groups, and awareness of these predictable effects. Any sophisticated actor controlling the attention architecture should know that flooding shifts beliefs by altering salience and perceived consensus. Draw the interim conclusion: the troubling version of McKenna's "best case" is propaganda on the revised Ross account. The mechanism by which it becomes troubling is the mechanism of epistemic defectiveness. Return to the dilemma and state the collapse result. McKenna's purported middle category—worry without propaganda—is unstable. If we idealise his "best case" to remove all epistemic defectiveness, we get something benign. If we preserve the troubling features he cares about—marketplace domination—we get epistemic defectiveness and hence propaganda. There is no stable middle ground. --- # Conclusion Restate the collapse result. McKenna's attempt to identify a politically troubling "best case" distinct from propaganda faces a dilemma once we apply the revised Ross account. Either the scenario is genuinely epistemically non-defective and benign, or the scenario remains troubling in McKenna's sense and becomes propaganda via campaign-level epistemic defectiveness.