The history of science presents a peculiar embarrassment for those who believe our best theories describe reality. Caloric fluid once explained heat transfer with mathematical precision. Phlogiston accounted elegantly for combustion. The luminiferous ether provided the medium through which light propagated. Each theory generated successful predictions—and each was eventually abandoned wholesale, its central entities revealed as fictions.

This pattern repeats with uncomfortable regularity. The entities posited by successful past theories have been systematically discarded, their names now serving as cautionary tales. Why, then, should we believe that electrons, quarks, and spacetime fare any better? The pessimistic meta-induction concludes we shouldn't. Our current theoretical furniture will likely join the cosmic junkyard of discarded ontologies.

Structural realism offers a sophisticated response to this challenge. It concedes that science may not reveal the intrinsic nature of things—what objects are in themselves—while insisting that something genuinely survives theory change. Mathematical structures, the web of relations that theories describe, persist even as our interpretations of what instantiates them transform. This suggests science tracks the world's relational architecture rather than its ultimate constituents. The question becomes: is structure merely what we can know, or is structure all there is?

Scientific realism holds that our best theories provide approximately true descriptions of the world, including its unobservable parts. Electrons exist. Spacetime curves. Quantum fields permeate reality. This seems like the obvious lesson of scientific success—how else could theories predict phenomena with such astonishing accuracy except by latching onto how things actually are?

Enter the pessimistic meta-induction. Larry Laudan's devastating 1981 argument compiled a list of once-successful theories whose central entities we now reject. The crystalline spheres of Ptolemaic astronomy generated predictions accurate enough for navigation. The caloric theory explained heat phenomena mathematically. The ether allowed Maxwell's equations to describe light as wave propagation through a medium. Each theory was empirically successful. Each is now false.

The inductive pattern seems clear: past theories' entities were discarded despite empirical success, so current theories' entities will likely suffer the same fate. The no-miracles argument—that realism best explains scientific success—meets its nemesis. If past successes weren't miracles despite being wrong about basic ontology, perhaps empirical success doesn't require true description at all.

Some realists respond by arguing the historical cases are cherry-picked, that mature sciences show genuine convergence. But the deeper problem remains. The history of science reveals wholesale ontological revision occurring alongside mathematical continuity. Fresnel's equations for light diffraction survived intact into Maxwell's electromagnetism and then into quantum electrodynamics, even as the underlying picture shifted from elastic ether vibrations to field oscillations to photon exchange.

This pattern—mathematical structure persisting while ontological interpretation transforms—suggests we may be asking the wrong question when we ask whether electrons really exist. Perhaps the question itself presupposes a kind of access to intrinsic natures that science cannot provide. Structural realism proposes a more modest claim about what theory change preserves.

Takeaway

Scientific success may not require describing what things fundamentally are—only how they relate to each other.

John Worrall's epistemic structural realism identifies precisely what survives theory change: mathematical structure. Fresnel described light using wave equations. Maxwell reinterpreted those equations as describing electromagnetic field oscillations rather than ether vibrations. The mathematics—the relational structure—remained constant. What changed was our account of what instantiated that structure.

This pattern generalizes remarkably. Newton's gravitational theory described inverse-square attraction between masses. General relativity reconceived gravity as spacetime curvature, eliminating force entirely. Yet Newton's equations emerge as limiting cases, accurate wherever gravitational fields are weak and velocities low. The structural relations Newton described remain approximately correct; only their interpretation as forces acting at a distance was abandoned.

The structural realist claims this is no coincidence. Science tracks structure—the pattern of relations among quantities, the mathematical form of laws, the symmetries preserved across transformations. What science cannot reliably access is intrinsic nature: what the relata are in themselves, considered independently of their relational properties.

Consider electrons. We know their charge, mass, spin—all relational properties, defined by how electrons interact with other things. Do electrons have some further intrinsic nature beyond these dispositional properties? Epistemic structural realism remains agnostic. Perhaps they do, but science cannot reveal it. Our epistemic access runs through structure alone.

This position preserves scientific realism's core insight—successful theories are non-accidentally successful because they capture something real—while conceding the anti-realist's historical point. Structure is real and knowable. Intrinsic nature, if it exists, exceeds our epistemic grasp. We believe in the equations, remain agnostic about what ultimately satisfies them.

Takeaway

Mathematical structures persist through scientific revolutions because they capture the relational architecture of reality—the pattern, not the substance.

Ontic structural realism takes a bolder step. Where epistemic structural realism says structure is all we can know, ontic structural realism claims structure is all there is. There are no objects with intrinsic natures that then stand in relations. Relations are ontologically primary; what we call objects are merely nodes in relational networks, individuated entirely by their structural position.

This sounds initially paradoxical. How can there be relations without relata—without things that stand in those relations? James Ladyman and Don Ross argue that modern physics already points this direction. In quantum field theory, particles are excitations of underlying fields. In general relativity, spacetime points have no intrinsic identity independent of the metric relations between them. The things physics describes increasingly dissolve into pure structure.

Consider identity in quantum mechanics. Two electrons cannot be distinguished by any intrinsic property—they share all their qualitative features. Their individuality, if they have any, cannot consist in something in them. Ontic structural realism suggests they never were individuals in the traditional sense. What exists are fermionic field configurations, structured patterns without underlying subjects.

The implications cascade outward. If structure exhausts reality, then the world has no hidden intrinsic nature that escapes mathematical description. Physics aims at complete structural characterization, not approximation to some deeper truth it cannot reach. The pessimistic meta-induction loses its force because there is nothing beyond structure that our theories might be missing.

Critics object that structure cannot exist free-floating, that relations require relata as a matter of metaphysical necessity. Ontic structural realists respond that this intuition reflects prejudices inherited from everyday object-talk, not deep metaphysical truth. Mathematics happily describes structures without presupposing intrinsically-natured occupants. Perhaps reality follows suit.

Takeaway

If relations are all that exist, then physics doesn't approximate reality—it directly describes it, and there is nothing beyond the equations.

Structural realism reshapes the realism debate by reconceiving what science reveals. Rather than fighting rearguard actions to defend the existence of current theoretical entities, it asks what genuinely survives revolutionary theory change. The answer—mathematical structure—suggests science tracks the world's relational architecture with impressive reliability.

The choice between epistemic and ontic versions marks a fundamental divide. Either structure is what we can access, leaving intrinsic natures forever beyond our reach, or structure exhausts reality, eliminating intrinsic natures entirely. Both positions honor the historical record while preserving scientific realism's core commitment to the non-accidental character of empirical success.

The stakes extend beyond philosophy of science. If ontic structural realism proves correct, the world has no hidden depths that escape structural characterization. Reality is the web of relations that physics describes. What seemed like epistemological modesty—admitting limits to scientific knowledge—transforms into metaphysical radicalism: there is nothing more to know.