The prefix post in postdigital makes its first appearance in the abstract of the second paper of this dissertation. It is not yet combined with propaganda, but with warfare. I hesitated before adding it: the prefix felt more rhetorical than analytical, a marker of novelty rather than a generator of insight. Yet as empirical cases accumulated that could not be adequately explained by existing frameworks, the term postdigital proved its analytical worth. As Jandrić et al. acknowledge, the postdigital “is hard to define; messy; unpredictable; digital and analog; technological and non-technological; biological and informational” (Jandrić et al., 2018:895) — a term, as Cramer and Jandrić (2021) put it “that sucks but is useful”.

This chapter develops that argument. It moves from pre-digital to digital to postdigital propaganda — not as a chronological narrative but as a sequence of infrastructural transformations that reorganised the conditions under which propaganda operates. It identifies the limits of existing digital-era frameworks and proposes a working definition of postdigital propaganda.

2.2 The Limits of Existing Frameworks

The digital-era frameworks that emerged in response to networked political communication — computational propaganda, participatory propaganda, hybrid media systems analysis — represent genuine theoretical advances. By foregrounding platform dynamics, coordinated amplification, and the co-productive role of audiences, they moved decisively beyond transmission models that treated persuasion as a linear process of strategic encoding and passive reception. The question is not whether these frameworks illuminate the phenomena this dissertation examines — they do — but whether they theorise those phenomena at the structural level that postdigital conditions require. Four limitations suggest they do not adequately.

First, the dominant frameworks recognise that platforms shape political communication, but theorise platform architecture as an environment through which strategic actors operate rather than as the generative mechanism of persuasion itself. Where environment implies a space within which things happen, ecology implies a system that constitutes the conditions under which political communication becomes possible. Lin (2024) identifies this as a “new wine in old bottles” problem — the observation that computational propaganda research, despite its digital framing, remains anchored in traditional transmission paradigms, with much of its empirical focus on bot detection and coordinated inauthentic behaviour during the 2016 US electoral cycle (e.g., Woolley & Howard, 2018). Analyses of the broader media ecosystem of that moment, such as Network Propaganda, pointed instead to asymmetric partisan amplification within networked media systems. Lin's own expanded theory (2024) represents the most recent attempt to address this limitation, broadening the analytical scope beyond state actors to include commodification, priva­tisation, connectivity, and virality. Yet even here, platform algorithms appear primarily as amplification channels — mechanisms that extend the reach of strategically produced content — rather than as generative infrastructures that prestructure the conditions under which any content achieves political salience. TikTok makes this distinction analytically decisive. Its recommendation logic, sonic affordances, and remix infrastructure are not merely channels through which propaganda travels — they are conditions of what propaganda can be.

Continue reading.

2.3 Postdigital Propaganda: A Working Definition

The choice of the prefix postdigital rather than alternatives such as platform or algorithmic propaganda requires justification. Terms like platform propaganda or algorithmic propaganda would accurately name two of the conditions this dissertation theorises — infrastructure and recommendation logic — but they would leave the deeper structural shift unnamed. What distinguishes the current moment is not simply that propaganda now travels through platforms or is amplified by algorithms, but that the digital infrastructure enabling these processes has become the continuous environment within which political life is organised rather than a distinct communicative sphere — as unremarkable and constitutive as electricity or running water.

Postdigital propaganda refers to the infrastructurally conditioned organisation of political alignment within algorithmically governed platform ecologies. Strategic and non-strategic actors contribute to this process, but what their contributions become is shaped less by individual intent than by platform infrastructure, platformized affect, participatory recursion, and synthetic plausibility. Each of these four dimensions is elaborated in Section 4 through the empirical cases the dissertation examines.

Political alignment, as used here, is a property of the platform ecology rather than a measurable outcome at the level of individual users. It becomes observable when three conditions co-occur: ecological saturation, in which a political imaginary spreads until it achieves ambient density; directional coherence, in which this spread consistently reinforces similar political orientation without evidence of institutional coordination; and permeation, in which content reaches algorithmically delivered feeds of users with no prior affiliation to the originating community.

Each condition alone is insufficient to produce alignment. Saturation and coherence within a bounded niche describe community reinforcement; permeation without coherence describes ordinary algorithmic amplification. Postdigital propaganda becomes observable as these three conditions converge, with full alignment representing their strongest co-occurrence.

Continue reading.

This dissertation began with hours spent on TikTok trying to comprehend what was happening in Ukraine before and immediately after the Russian invasion on 24 February 2022 — experiencing a conflict through a multiplicity of perspectives, from dancing soldiers to propagandistic sound memes, that I had not encountered on any other platform. The research logic across all four papers follows from that origin: driven by a recurring sense that political content on the platform exceeded what existing frameworks could adequately name, across contexts as different as wartime conflict, electoral campaigns, and synthetic media ecologies. The methods documented in this chapter are formalizations of that practice.

3.1 The Dachschrift

The cumulative format was a deliberate choice. Each paper was designed to capture a different context, actor configuration, and analytical perspective on propaganda, while remaining publishable in its own right. The argument that connects them became fully visible only in retrospect: what the four papers document separately is not merely four related phenomena but a single structural condition appearing across contexts that differ in the dimensions alternative explanations would normally invoke. The Dachschrift is therefore not a summary of the papers but the theoretical argument their accumulation makes possible.

3.2 Methods

The four papers that constitute the empirical core of this dissertation are summarised in Table 1. The papers are designed contrastively rather than additively, maximising analytical variation across the dimensions most relevant to the postdigital propaganda framework: two conflict contexts (Ukraine, Gaza) and two non-conflict contexts (German federal election, cross-national synthetic media study); two cases primarily involving state or institutional actors (IDF, US federal government) and two involving decentralised or unofficial actors (TikTok sound networks, fan accounts and AI accounts).

This contrastive design logic is the foundation of the structural argument. Read retrospectively, it reveals whether the four dimensions of the postdigital propaganda framework — platform infrastructure, platformized affect, participatory recursion, and synthetic plausibility — appear consistently across empirical contexts where competing explanations would normally expect different dynamics.

The argument is not that each paper arrives at the same empirical finding, nor that each paper was designed within the postdigital framework — the framework emerged retrospectively from the accumulation of empirical work, and the theoretical register of each paper reflects the point in that process at which it was written. What the contrastive design makes visible is that the empirical phenomena documented across all four papers exceed the frameworks through which they were initially approached — and become fully legible only when read through the analytical coordinates the dissertation develops as a whole.

3.3 Empirical Scope and Limitations

TikTok is the primary platform across all four papers. Its sound-driven architecture, short- form vertical video format, algorithmic For You Page, and vernacular affordances — reusable sounds, duets, stitches, editing tools — make it the site where the mechanisms of postdigital propaganda are most directly observable for the current research question. Paper 4 (Bösch, forthcoming) extends the empirical scope to Instagram, justified by the cross-actor comparison at the center of that paper: including Trump and the White House alongside @dobermanncloe1 on TikTok strengthens the structural argument by providing preliminary evidence that comparable dynamics may emerge on platforms sharing similar engagement logics.

Across the four papers, the unit of analysis is not individual pieces of content, actors, or platform features in isolation, but observable configurations of propaganda dynamics within platform environments. Individual videos, sounds, accounts, or campaigns function as analytical entry points that make these dynamics visible. The analysis therefore focuses on how platform infrastructures, affective circulation, participatory practices, and synthetic media interact in specific empirical instances to generate observable patterns of political alignment.

The temporal scope spans 2022 to 2025, from the Russian invasion of Ukraine and the October 2023 Gaza war through the 2025 German federal election and the first months of the second Trump administration. The four papers constitute a multiple-case study of a structural condition documented at different points within this period, not a longitudinal study of change across it.

Two limitations are carried forward into the theoretical synthesis. First, the research design focuses on observable platform practices rather than internal production processes. Without access to campaign organizations, platform moderation systems, or internal decision-making structures, the study cannot determine how specific actors intentionally design propaganda interventions or how platform companies shape these dynamics through internal governance mechanisms. Second, the research does not measure users interpretation, persuasion, or behavioural effects. Engagement metrics and participatory practices are treated as indicators of participatory uptake rather than as evidence of attitude change or political persuasion. The study therefore examines the conditions under which propaganda circulates rather than its effects. The four papers that follow are therefore interpreted in the synthesis not as isolated case studies but as empirical observations of the same structural condition under these scope constraints.

4.1 Structure and Logic of the Synthesis

The four sub-chapters that follow share a common argumentative structure. Each begins from empirical phenomena that existing digital-era frameworks could observe but not fully account for. This is not because those frameworks are wrong, but because their theoretical architecture does not capture the structural level at which postdigital propaganda operates. Each then identifies the specific assumption within prior theory that the finding unsettles: that infrastructure is merely a condition rather than a generative mechanism, that affect is produced through user practice rather than by algorithmic infrastructure, that propaganda effects are attributable to strategic actors rather than emergent from participatory structure, and that propaganda operates through factual deception rather than the accumulation of affective plausibility.

Each section closes by showing how the empirical record requires the concept the framework provides — infrastructure as generative mechanism, platformized affect, participatory recursion, and synthetic plausibility — demonstrating that existing frameworks cannot account for what the cases document without it. The movement in each case is the same: from what the papers found, through what prior frameworks cannot explain, to what the empirical record requires. Several cases recur across sections because the same phenomena are analytically multidimensional: each case is read through a different analytical lens to demonstrate that the four dimensions are co-present within single empirical instances. Each dimension therefore corresponds to empirical dynamics documented in one or more of the four papers, which together reveal the structural configuration the concept of postdigital propaganda names.

4.2 Platform Infrastructure as Generative Mechanism

Existing frameworks for computational propaganda, most prominently articulated by Woolley and Howard (2018), center analytical attention on strategic actors who deploy computational tools — bots, automation, and algorithmic targeting — to influence political communication. Platforms in this account structure the communication environment, but propaganda remains analytically tied to the intentional activities of organised actors who deploy these techniques.The empirical record across the four papers challenges this assumption at multiple stages of the production–circulation cycle.

The “Russian warship go f*** yourself” sound page analysed in Paper 1 (Bösch & Divon, 2024) accumulated more than 80 billion views across derivative videos through a propagation mechanism that exceeded what any actor could have managed or controlled after the initial upload. TikTok’s sound page architecture automatically aggregates all videos using the same audio into a single feed, producing recursive visibility and circulation across thousands of derivative posts. The Ukrainian government was the origin actor with clear propagandistic intent, but the subsequent propagation phase shows no evidence of coordination or management by any institutional actor — operating instead through the platform's own recommendation logic and participatory vernaculars.

Continue reading.

4.3 Platformized Affect

Much of classical persuasion research treats affect as something messages produce in receivers — an emotional response that shapes attitudes and judgments (Petty & Cacioppo, 1986). Papacharissi's account of affective publics (2015) moves beyond this by showing that affect is not only a response to individual messages but something that emerges collectively through platform-mediated practices — hashtag streams, retweet structures, and connective storytelling. Affect, in this account, is constitutive of how publics form rather than a byproduct of how messages land.

What remains comparatively undertheorised in both accounts is the infrastructural layer beneath these practices: the recommendation systems that shape the distribution of visibility across content on platforms (Bucher & Helmond, 2018; Gillespie, 2018). Affect functions not only as an outcome of communication but as a ranking signal within algorithmic recommendation systems. Aggregated engagement behaviour feeds back into distribution, meaning that affective resonance becomes a structural condition of visibility rather than merely a response to content.

In this dissertation, platformized affect is identified through observable platform vernaculars and aesthetic registers that precede engagement outcomes. Indicators include the recurrent use of platform-native affective formats (e.g., dance challenges, thirst traps, slowed-down edits, lip-sync performance), aesthetic markers associated with specific affective registers (e.g., intimacy, humour, irony, or admiration), and stylistic conformity with dominant platform vernaculars such as short-form narrative compression, soundtrack-driven pacing, and meme-based audiovisual editing. These indicators are observable at the level of form and performance prior to circulation outcomes and therefore provide an analytical basis for identifying affective alignment or misalignment independent of engagement metrics.

Continue reading.

4.5 Synthetic Plausibility

Influential disinformation frameworks (e.g., Wardle & Derakhshan, 2017) organise analysis around the epistemic status of information, distinguishing between true and false content. Frameworks organised primarily around the true/false axis offer less analytical vocabulary for capturing the propaganda dynamics documented across the four papers: content circulates and produces political alignment through affective plausibility rather than factual deception.

Synthetic plausibility is distinct from platformized affect: where platformized affect names the infrastructural conditions under which affective resonance determines visibility, synthetic plausibility names the epistemic condition under which repeated aesthetic instantiation stabilises a political world as perceptually coherent — independently of whether that world corresponds to documentable reality. The mechanism is not deception but accumulation: not the falsification of individual claims but the construction of a felt political world through serial aesthetic repetition.

Synthetic plausibility does not originate with generative AI. Rather, AI completes a propaganda dynamic already visible in earlier platform practices: the stabilisation of political worlds through affective plausibility rather than factual verification. Propaganda has always operated through the construction of felt political realities — worlds in which certain actors, threats, and allegiances feel self-evident — rather than through the patient accumulation of verified evidence. What previously constrained this dynamic was the requirement that content maintain some indexical relationship to documentable reality: real events as narrative anchors, real bodies as footage, real figures as recognisable referents. Generative AI removes that final constraint — not by introducing a new propagandist logic but by eliminating the last structural limit on how completely and self- sufficiently a political world can be constructed without documentary grounding.

The Russian warship meme documented in Paper 1 (Bösch & Divon, 2024) spread as a heroic narrative — Ukrainian soldiers defying a warship, the sound page accumulating 80 billion views — without circulation depending on whether the details of the exchange were documentarily established. The IDF thirst trap content in Paper 2 (Bösch & Divon, 2026) produced a humanised image of military actors through dance challenges and platform vernaculars without making any verifiable claim about the conflict it was embedded in. In neither case did factual accuracy or inaccuracy appear to govern circulation or political effect. What mattered was not whether the content was true but whether it produced affective resonance with a political imaginary that made particular responses feel warranted.

The serial Trump images documented in Paper 4 (Bösch, forthcoming) — gangster, conductor, pope, king, Sith lord, Superman — do not advance explicit propositional claims. Together they produce a world in which executive authority feels mythic, sacred, and extra- democratic — not argued but aesthetically stabilised through serial repetition. The @dobermanncloe1 account constructs a völkisch political imaginary through dozens of synthetic videos without advancing explicit propositional claims — blonde children with eagles, warrior women with national flags, Putin as guitar-playing ally. Generative AI does not introduce this dynamic — it completes it. Each image adds another layer to a political world that accumulates through aesthetic density rather than argument — felt as coherent before any individual claim is evaluated.

The Trump Gaza AI video documented in Paper 4 (Bösch, forthcoming) constructs a complete political world — Trump as redeemer, Gaza as luxury resort, displacement as transformation from chaos to order — by subordinating a real political proposal to affective overwriting. Before generative AI, synthetic plausibility required real events as narrative anchor (Bösch & Divon, 2024), real bodies as footage (Bösch & Divon, 2026), or real political figures as recognisable referents (Bösch, 2026). Generative AI in this specific case does not remove the political referent — the Gaza proposal exists — but the evidentiary constraint on what images can be made to show. What disappears is the requirement that images maintain an empirical or indexical referent to events that actually occurred. Political worlds can now be constructed through algorithmic synthesis without the requirement that an image record something that happened. The propaganda effect is therefore not simply more intense — it is more closed: a self-sufficient political world that requires the world as referent but no longer as evidentiary limit.

Across the four papers, synthetic plausibility emerges as a structurally necessary dimension of postdigital propaganda — present in Papers 1, 2, and 3 as affective resonance that circulates independently of factual status, and completed in Paper 4 as the construction of self-sufficient political worlds no longer bounded by documentary grounding.

4.6 Co-Constitution and the Limits of Analytical Separation

7. References

…