Monadological Sociology

In the AI Story

The framework's contemporary relevance emerges precisely when we try to understand what a large language model is doing. The model processes a corpus of texts, each text a trace of a specific monad's perspective — a specific mind's encoded relationship to the whole of its experience. The model's output is not a new perspective but a statistical integration of many perspectives, weighted by the probabilistic regularities in how those perspectives have historically been expressed. In monadological terms, the model simulates a perspective without occupying one. It produces outputs that look perspective-like but lack the positional specificity that defines a genuine monad. This is why AI output tends toward the mean: it integrates perspectives but does not occupy any, and therefore cannot produce the specifically positional modifications that a genuine perspective would introduce.

The framework also illuminates why human-AI collaboration can produce genuinely valuable work despite this limitation. The human in the collaboration occupies a specific perspective — a specific monad's position in the social flow — and her modifications to the model's output reflect that positional specificity. The collaboration is not a replacement of human perspective with machine perspective; it is the interaction of a genuine perspective with a statistical integration of many perspectives. When the human brings genuine positional commitment — genuine stakes, genuine relationships, genuine participation in the living flow — her modifications reconnect the model's perspective-less output to the perspectival flow from which all genuine social contribution emerges.

Bruno Latour's development of actor-network theory explicitly drew on Tarde's monadology, recognizing it as the metaphysical foundation most compatible with ANT's refusal of the macro/micro distinction and its insistence that social reality consists of connections between specific actants rather than structures imposed above them. Latour argued that Tarde's monadology had anticipated the central insights of contemporary network science and digital sociology by more than a century. The framework's rehabilitation has accelerated as its empirical plausibility has become demonstrable through digital tools Tarde could not have imagined.

Origin

Tarde articulated the monadological framework most fully in Monadologie et sociologie (1893), a short book that attempted to ground his sociology in an explicit metaphysics. The framework drew on Leibniz's 1714 Monadology but adapted it to social rather than metaphysical concerns. The book was considered eccentric by Tarde's contemporaries and was ignored for most of the twentieth century. Its rehabilitation began with Deleuze's engagement with Tarde in the 1960s and accelerated with Latour's explicit development of the framework for actor-network theory in the 2000s.

Key Ideas

Every individual is a perspective on the whole. No individual contains the whole, but each represents it from a specific position that no other individual occupies.

The social is continuous mutual modification of perspectives. Not a structure above individuals but the flow of perspective-modification between them.

The macro/micro distinction dissolves. At sufficient resolution, the difference between individual perspective and collective pattern becomes a question of observational scale, not of ontological kind.

AI simulates perspective without occupying one. The model integrates many perspectives statistically but does not occupy a position in the flow, which is why its output tends toward the mean rather than toward perspectival specificity.

Human collaboration reconnects perspective to flow. The human partner's specific positional commitment is what transforms the model's perspective-less integration into a contribution to the perspectival flow.

Debates & Critiques

The framework's metaphysical ambition makes many contemporary sociologists uncomfortable — empirical sociology generally prefers to bracket metaphysical questions. But the AI transition has made the bracketing increasingly difficult. Questions about whether AI systems are conscious, whether they have perspectives, whether their output constitutes genuine social contribution cannot be answered without some implicit or explicit metaphysical commitment. The monadological framework offers one such commitment that accommodates both the empirical observations about AI capability and the intuitive recognition that something important is missing from purely statistical integration.