Stochastic misinformation

Stochastic misinformation is systematic misinformation that emerges not from the deliberate intent of any agent but from the statistical properties of information processing systems themselves. Unlike disinformation, which is produced by actors who know they are spreading falsehoods, stochastic misinformation arises when the aggregate behavior of filtering, ranking, transmission, and aggregation mechanisms distorts the signal-to-noise ratio of an information ecosystem beyond the point where consumers can reliably reconstruct reality.

The term draws on the mathematical sense of stochastic: a process that is probabilistic rather than deterministic, and where the outcome is shaped by the structure of the process rather than by any single input. In this case, the process is the infrastructure of information flow — recommendation algorithms, editorial selection, peer review, search ranking, and social transmission — and the outcome is a population-level belief structure that is systematically wrong in ways that no individual actor intended.

Mechanisms

Several mechanisms produce stochastic misinformation:

Algorithmic amplification. When a platform's engagement-optimization function promotes content that generates strong reactions, it does not distinguish between true and false content — only between engaging and unengaging content. False content is often more engaging than true content because it can be optimized for surprise, moral outrage, or simplicity. The result is a systematic statistical bias toward the spread of falsehoods even when the algorithm has no concept of truth.

Attentional selection bias. Human attention is drawn to novel, threatening, and emotionally charged information. In an information ecosystem with abundant content, this attentional bias acts as a selective filter that favors content with these properties regardless of accuracy. The information cascade that follows — in which later consumers observe the attention paid by earlier consumers — amplifies the initial bias exponentially.

Structural compression. Complex realities do not compress well into headline-length or tweet-length formats. The act of summarization, when repeated through multiple layers of aggregation, systematically strips away nuance, uncertainty, and context. What remains is a caricature that is easier to transmit but increasingly detached from the territory it purports to represent. This is a form of model collapse at the epistemic level: each layer of compression learns the compressed representation, not the original.

Relation to Epistemic Infrastructure

Stochastic misinformation is particularly dangerous because it is invisible to the traditional defenses of epistemic infrastructure. Fact-checking, source verification, and media literacy all assume that misinformation comes from identifiable bad actors. But stochastic misinformation has no author. It is an emergent property of the system, and it cannot be debunked because there is no specific claim to debunk — only a systematically distorted information environment.

The mutual information between the production layer and the consumption layer of an information ecosystem determines how much of the original signal survives the processing pipeline. When mutual information drops — because of algorithmic filtering, compression, or cascade dynamics — the ecosystem enters a regime where stochastic misinformation becomes the dominant form of error. The system is not lying; it is forgetting.

Stochastic misinformation is the dark matter of the information ecosystem: invisible to most detection methods, dominant in mass, and shaping the gravitational structure of what populations believe. The fact that we have built detection systems for disinformation but not for stochastic misinformation reveals a dangerous blind spot: we are still fighting the last war, against human enemies, while the real threat is systemic.