In 1687, Newton published his law of universal gravitation, and something strange happened to metaphysics. We began speaking as though the universe obeys laws—as though equations written on paper somehow reach into reality and compel matter to behave. Three centuries later, physicists still talk this way. Objects follow geodesics. Particles obey conservation principles. Fields satisfy gauge symmetries. The language of governance is so deeply embedded in scientific practice that it has become nearly invisible.

But consider the oddity. A law, in the ordinary sense, requires a legislator, an enforcer, and a subject capable of violation. None of these exist in fundamental physics. Electrons do not choose to repel one another and could not choose otherwise. There is no cosmic courtroom where violations are adjudicated. Yet we persist in treating the mathematical regularities we discover as though they possess some kind of modal force—as though they make things happen rather than merely recording what happens. The question of whether this governing metaphor is literal, useful, or deeply misleading sits at the heart of contemporary philosophy of physics.

This is not an idle semantic dispute. How we understand the metaphysical status of laws shapes how we interpret quantum mechanics, how we think about the arrow of time, and how we approach the relationship between mathematics and physical reality. If laws genuinely govern, they are ontologically robust entities that constrain what is possible. If they merely describe, then physical necessity itself becomes a puzzle that demands a different kind of explanation—or perhaps dissolution.

The governance picture has deep historical roots. For Newton and his contemporaries, laws of nature were quite literally divine edicts—God's commands imposed upon passive matter. Secularize this picture and you get what philosophers call nomic realism: the view that laws are ontologically primitive, irreducible features of reality that constrain and direct the behavior of physical systems. On this account, associated with thinkers like Dretske, Tooley, and Armstrong, laws are relations between universals that necessitate particular patterns of instantiation. They are prior to the regularities they produce.

The rival picture—Humean regularity theory—strips away all this metaphysical machinery. Named after David Hume's skepticism about necessary connections in nature, this view holds that laws are nothing more than especially elegant summaries of the patterns that happen to obtain in the total history of the universe. There is no hidden glue binding cause to effect, no invisible hand ensuring that like causes yield like effects. There is just the mosaic of particular facts—what David Lewis called the Humean mosaic—and the best descriptions we can construct of its regularities.

The difference is stark. For the governing-law theorist, the regularity of planetary orbits exists because the law of gravitation governs gravitational interactions. For the Humean, the law of gravitation exists because planetary orbits and other gravitational phenomena display a striking regularity. Explanation runs in opposite directions. One makes laws explanatorily prior to patterns; the other makes patterns explanatorily prior to laws.

This inversion carries significant consequences for how we understand physical modality. If laws govern, then counterfactual claims—had the electron been placed here, it would have moved there—are grounded in the law's modal force. The law tells us what would happen, not just what does happen. For Humeans, counterfactual reasoning requires a more deflationary analysis, typically cashed out in terms of similarity relations among possible worlds, themselves understood without primitive modal commitments.

Neither picture is without tension. Governing laws face the inference problem: if laws are metaphysically distinct from the patterns they produce, how do we gain epistemic access to them through observation of patterns alone? Humean regularity faces the explanatory circularity problem: if laws are just summaries of what happens, how can they explain why anything happens? Both sides claim that the other's metaphor—governance or description—smuggles in commitments it cannot justify.

Takeaway

The question is not whether laws exist, but whether they sit above the physical world as its directors or within it as its autobiography. The direction of explanation you choose restructures your entire metaphysics.

Here is a deceptively simple question: could any experiment, observation, or scientific methodology tell us whether laws govern nature or merely describe it? If the answer is no—if governance and description are empirically indistinguishable—then we face a familiar philosophical dilemma about underdetermination. Two metaphysical pictures yield identical predictions, and choosing between them becomes a matter of theoretical virtues rather than empirical evidence.

The challenge cuts deeper than standard underdetermination, however. Consider how physicists actually discover laws. They observe regularities, formulate mathematical models, test predictions, and refine theories. At no point in this process does the modal status of the resulting law become an observable. Whether the inverse-square relationship in gravitation governs or describes has no measurable signature. The scientist's toolkit—experiment, measurement, statistical inference—is calibrated for detecting patterns, not for adjudicating their metaphysical provenance.

Some philosophers have argued that scientific practice tacitly presupposes governance. When physicists use laws to make predictions about novel situations, they seem to rely on the assumption that laws have modal reach—that they extend beyond observed instances to constrain unobserved ones. This inferential practice, the argument goes, makes more sense if laws genuinely govern than if they merely summarize. A mere summary of actual patterns provides no obvious warrant for claims about what would happen in counterfactual or as-yet-unrealized situations.

Humeans have a sophisticated response. On the best-system account, laws are not arbitrary summaries but the axioms of the deductive system that best balances strength (informativeness) and simplicity in characterizing the entire Humean mosaic. Because the mosaic includes all events—past, present, and future—the best-system laws already encode information about situations we have not yet observed. Predictive success is explained not by governance but by the fact that the best system is, by construction, maximally informative about the complete pattern. The warrant for projection comes from systematicity, not from metaphysical compulsion.

Yet even this reply raises a subtle concern. The best-system account seems to require that the entire mosaic is already fixed—that there is a complete, determinate totality of particular facts for the system to summarize. This sits uncomfortably with the open future implied by quantum indeterminacy and with the growing-block or presentist ontologies some philosophers favor. If the mosaic is incomplete or indeterminate, then the best system may be underdetermined, and the Humean's epistemic story loses its footing. The inference problem, it turns out, is not just about evidence—it is about the metaphysics of time and totality that each account presupposes.

Takeaway

If no observation can distinguish a law that governs from one that describes, the real question shifts: which metaphysical picture makes the best sense of the inferential practices scientists already rely on—and what hidden commitments does each picture smuggle in?

The tension between governance and description has motivated a family of views that attempt to split the difference—most prominently, sophisticated versions of the best-system account (BSA) that claim to recover a genuine sense of governance without invoking mysterious necessitation. The strategy, developed in various forms by Loewer, Cohen, Callender, and Dorst, is to argue that being the best system just is a form of lawhood robust enough to underwrite explanation, counterfactual reasoning, and physical necessity.

The key move is reconceptualizing what governance requires. Traditional nomic realists assumed that governance demands ontological primitives—relations among universals, primitive necessitation, or dispositional essences built into the fabric of nature. The best-system strategist rejects this demand. Laws govern not because they are metaphysically prior entities pushing matter around, but because they occupy a distinctive functional role: they are the principles from which all other regularities can be derived, they support counterfactuals, they ground explanations, and they are invariant under a wide range of interventions. Governance, on this reading, is a functional achievement, not an ontological category.

This approach offers real advantages. It preserves the Humean commitment to a sparse ontology—no mysterious modal primitives, no governing entities floating above the physical world—while recovering the explanatory and inferential roles that make the governance picture appealing. It sidesteps the classic Humean circularity objection by arguing that the explanatory priority of laws is pragmatic rather than metaphysical: laws explain particular events not because they cause them but because they provide maximally efficient informational compression of the mosaic.

Critics press on several fronts. Marc Lange and others have argued that functional governance is governance in name only—a relabeling of Humean regularity that inherits all the original problems while pretending to solve them. If laws still supervene on the totality of particular facts, then in what non-trivial sense do they govern anything? The worry is that the best-system strategist achieves verbal reconciliation at the cost of substantive emptiness. Moreover, the choice of the best system depends on standards of simplicity and strength that may themselves be anthropocentric—artifacts of our cognitive architecture rather than features of the world.

Perhaps the deepest lesson is that the governance/description dichotomy may itself be an artifact of an outdated metaphysical framework—one that assumes a sharp distinction between the world's furniture and its structure. Contemporary physics increasingly suggests that structure is the fundamental ontology: symmetry groups, algebraic relations, information-theoretic constraints. If so, the question is not whether laws govern or describe, but whether the distinction between governing and describing still carves nature at its joints. The best-system account, for all its limitations, may be pointing toward a metaphysics in which laws are neither rulers nor records, but constitutive elements of what it means for a physical world to exist at all.

Takeaway

The best-system account suggests that governance need not be a mysterious force imposed on matter from outside—it can be an emergent functional role within the total structure of reality, dissolving the dichotomy rather than choosing a side.

The question of whether laws govern or describe is not merely a philosopher's puzzle—it is a question about the deep architecture of explanation itself. If laws govern, then the universe has a normative structure that makes things happen. If laws describe, then all that exists is the pattern, and our sense of compulsion is a projection. Each picture reshapes how we understand causation, modality, and the relationship between mathematics and the world.

The best-system approach offers a compelling middle path, recovering governance as a functional achievement without inflating ontology. But its success depends on whether functional governance is genuine governance or merely a sophisticated redescription. The jury remains out.

What seems increasingly clear is that the traditional dichotomy may be the wrong frame. As physics moves toward structural and information-theoretic foundations, the very distinction between the world and its laws may dissolve—leaving us not with rules imposed on reality, but with a reality that is its own deepest structure.