Antonio Scala

The claim that hostile epistemic environments threaten liberal democracy is now widely accepted. In the sense developed in recent social epistemology [@nguyen2023hostile], such environments are hostile because they systematically exploit cognitive vulnerabilities, entrenched prejudice, fallacious reasoning, and adversarial structures of media and political discourse. This paper contributes to that discussion by isolating one mechanism through which hostile epistemic environments operate: the fragmentation of what I call *shared causal grammar* -- the minimally common conditions for attributing agency, responsibility, and legitimacy to public events. Existing accounts tend to under-specify this level, privileging content-level mechanisms while leaving a structurally prior layer under-addressed. Drawing on Paul Watzlawick's concept of *punctuation*, I argue that digital platforms do not merely distribute information: they organise the interpretive sequences through which events become causally intelligible as provocation, reaction, threat, defence, or victimhood. The same evidence can therefore generate divergent accounts not only because one side is misinformed, but because communities may inhabit different causal frames that organise the same facts into incompatible attributions of agency, responsibility, and legitimacy. I develop this claim by distinguishing content-level from punctuation-level accounts of epistemic hostility, showing how successive technological regimes automate deeper layers of causal intelligibility, and arguing that the resulting fragmentation constitutes an infrastructure failure that procedural democratic remedies alone cannot address, because such remedies presuppose the shared causal grammar whose fragmentation is at issue. This argument does not depend on treating generative AI as a merely error-prone or hallucinating technology. Even where retrieval, grounding, alignment, and reasoning improve local accuracy, algorithmically mediated systems still organise which sources are connected, which sequences become salient, and which causal narratives become publicly contestable.

ABOUT THE NON EUCLIDEAN NATURE OF INTERNET'S BIG DATA AND ITS CONSEQUENCES In this paper we discuss the non Euclidean nature of navigable Big Data, hinting as it could introduce – thanks to its combinatorial richness – the possibility of multiple interpretation spaces. We also propose that, in such structures, exploration algorithms have a paradoxical nature insofar they either enhance human biases or are useless. Finally, we propose that Internet could be used as an artificial universe for observing constructivism at work.

Paul Watzlawick argued that reality is not discovered but constructed through communication. This paper extends his radical constructivist framework to the digital age, tracing how information technologies have progressively automated the construction of what Watzlawick called "second-order reality" - the domain of meaning, value, and significance. The analysis follows three technological regimes. Search engines transferred epistemic authority from human judgment to algorithmic ranking, making relevance a function of network topology rather than truth. Social media fragmented shared reality into parallel epistemic bubbles, where narrative resonance supplants factual accuracy. Large Language Models complete this trajectory: they produce fluent, coherent discourse without comprehension, intention, or reference - making "hallucination" not a malfunction but the default mode of a system that generates meaning without anchoring it to anything beyond statistical probability. The paper introduces the concept of algorithmic constructivism to characterize this new condition, and warns of its anthropological consequence: epistemic atrophy. As sense-making is progressively delegated to automated systems, humans risk losing the capacity to distinguish negotiated meaning from algorithmic simulation - and with it, the ability to construct reality through genuine communicative exchange.

The wide availability of user-provided content in online social media facilitates the aggregation of people around common interests, worldviews, and narratives. However, the World Wide Web (WWW) also allows for the rapid dissemination of unsubstantiated rumors and conspiracy theories that often elicit rapid, large, but naive social responses such as the recent case of Jade Helm 15––where a simple military exercise turned out to be perceived as the beginning of a new civil war in the United States. In this work, we address the determinants governing misinformation spreading through a thorough quantitative analysis. In particular, we focus on how Facebook users consume information related to two distinct narratives: scientific and conspiracy news. We find that, although consumers of scientific and conspiracy stories present similar consumption patterns with respect to content, cascade dynamics differ. Selective exposure to content is the primary driver of content diffusion and generates the formation of homogeneous clusters, i.e., “echo chambers.” Indeed, homogeneity appears to be the primary driver for the diffusion of contents and each echo chamber has its own cascade dynamics. Finally, we introduce a data-driven percolation model mimicking rumor spreading and we show that homogeneity and polarization are the main determinants for predicting cascades’ size.