The wrong model, and why it's wrong
The dominant models of information spread are borrowed from epidemiology — the SIR family, adapted to treat stories and ideas as a contagion moving through a population. They have produced real insight, but they rest on one assumption that does not hold: that the medium information travels through is roughly homogeneous, a uniform population of similar people connected by similar contacts.
Social landscapes are not homogeneous. They are topographically complex terrain — shaped by culture, history, economics, identity, and institutional trust. A story that spreads freely across one region hits an impermeable boundary at the next. Epidemiological models have no mechanism for terrain: they model spread but not direction, velocity but not the phase change when a story stops spreading laterally and instead sinks in, accumulates, and conditions how the next story is received. The claim of this framework is narrow and specific: treating the social landscape as uniform produces predictions that are miscalibrated in the same systematic directions that treating air as uniform would miscalibrate a weather forecast.
What meteorology offers instead
Atmospheric science earned its predictive power by taking the heterogeneity of its medium seriously. Weather is fluid dynamics in a system with variable pressure, temperature, humidity, and topography — chaotic at short time scales, statistically characterizable at longer ones. That is a far better description of how information actually moves than any contagion model.
But the analogy carries a commitment the first version of this framework left implicit: in the cases that matter most, the medium is not a fixed stage. Narrative does not merely flow over the social landscape — it alters the landscape, and the altered landscape redirects what comes next. So the right question is not only "what are the atmospheric conditions?" but "how do the conditions change as a consequence of the weather they produce?" That second question is what the formal core below is built to answer.
The mapping
The framework lays out a full correspondence between atmospheric phenomena and their information-propagation equivalents. A selection of the load-bearing ones:
The paper is explicit that these are intended as analytical correspondences, not decoration — the bet is that the mathematics describing the atmospheric side describes the information side closely enough to be useful. Which of them survive formalization is exactly the open empirical question, taken up below.
The testable core — barometric pressure
The sharpest, most operational idea in the framework is the barometric-pressure signal: the most reliable leading indicator of a narrative storm is the accumulating divergence between what dominant narratives claim and what measurable ground truth shows. When economic narratives say one thing and the data says another, pressure builds. Pressure builds until it releases — as a storm, a rapid uptake of a new explanatory framework that resolves the accumulated tension. The storm looks sudden from inside the discourse, but the pressure that produced it was visible in the data for months or years.
Coordinated campaigns show up in the same instrument as seeded weather: when a storm forms that is far larger than the existing pressure gradient would predict, manufacture is the parsimonious explanation. The signature is measurable — synchronized timing, concentrated source, precise targeting — without having to name the actors or judge the content true or false. The measurable claim is only whether the weather was natural or seeded.
The water cycle — why narratives recur
Static network models describe one-way spread and then stop. The water cycle supplies the missing temporal dimension. A narrative that saturates the discourse evaporates — loses force, lifts off active belief, returns to cultural latency — where it persists as potential, recondenses in a new moment, and falls again on a generation that experiences it as new. The specific facts change; the underlying shape (the villain structure, the moral arc) cycles through the same patterns. Events are weather, patterns of dominance and collapse are seasons, and the deep structure of what a civilization can believe is climate — the same laws at three time scales.
What it forecasts, and what it doesn't
The predictive claim is deliberately honest. It does not predict which specific story will dominate on a given date — chaos theory rules that out past a short horizon, the same reason a meteorologist cannot place each raindrop. What it forecasts is conditions: given the current pressure gradients, saturation levels, terrain, and measured divergence, the probability distribution over what kinds of narrative events are likely, where, and over what window — most reliable in the near term, degrading to climatological characterization at the long range. Exactly the structure of a weather forecast.
The formal core — the plastic propagation medium
The fix is a coupled system on a graph. Belief is a fast variable that diffuses across the network; terrain — the receptivity lean of each node, the formal object the metaphors call sediment and salinity — is a slow variable that is rewritten by the belief flowing through it, and the rewritten terrain gates what flows next. In one line: in the old picture the terrain is a constant; here the terrain is changed by what crosses it, and process and memory become one substance. Set the plasticity rate to zero and the model collapses back to the fixed-medium (epidemiological) null exactly — so the two can be compared directly.
The right mathematics for this is smaller and more specific than the original atmospheric-physics list: a graph Laplacian for diffusion and watershed structure, coupled slow–fast dynamics for the plasticity, and bifurcation theory — whose early-warning signals (rising variance and autocorrelation before a sudden jump) are the operational form of the barometric-pressure forecast. Chaos theory contributes one idea only: the Lyapunov exponent bounds how far ahead the forecast can reach. Reaching for more than that is where frameworks like this become unfalsifiable.
The test that matters — three claims against the null
Both models run on the same clustered graph; the only difference is whether the terrain equation runs. Each of the framework's three signature claims was checked, with the fixed medium as the null:
All three are produced by the one mechanism and vanish without it. They are not three phenomena requiring three explanations — they are three faces of a medium rewritten by the flow it carries. That is the formal core the original draft described in metaphor and did not yet have in equations.
Where the mechanism came from
The plastic medium did not originate inside this framework. It was reached independently in a separate thread of this corpus — chasing the geometry of a buckyball down to a plastic connection, a self-modifying transport where the substrate is shaped by what flows across it — and recognized there as the object this framework needed. That a medium-rewriting transport arrived from an unrelated direction and fit the gap exactly is part of the reason to think it is the right object rather than a convenient one.