Laws Require Limits
When Anything Can Happen, Nothing Can Be Known
The concept of infinity is often treated as a harmless abstraction in physics and mathematics, something that can be safely invoked as an idealisation without committing to it as a property of reality. This attitude is so ingrained that it is rarely questioned. Yet if infinity is taken seriously, even in its weakest form as potential infinity, it leads to a contradiction that undermines the very possibility of physical theory and, more fundamentally, comprehension itself.
A physical theory is not merely a formal construct. Its purpose is to allow an observer embedded in the world to recognize regularities, form expectations, and infer structure from experience. This does not require certainty or determinism, but it does require that reality is not allowed to mislead its observers arbitrarily and without bound. If every conceivable deviation from expectation is physically permitted and unconstrained, then the act of understanding loses its footing.
Douglas Adams once imagined the “Infinite Improbability Drive,” a device that works by allowing any event, however absurd, to occur with non-zero probability, provided the universe is large enough. While intended as satire, this idea is an almost perfect metaphor for an infinite universe taken seriously. In such a universe, nothing is forbidden, no deviation is excluded, and no regularity is guaranteed to persist. Any outcome, however incompatible with expectation, becomes inevitable somewhere and, from a local perspective, indistinguishable from law. The joke works because it exposes a real inconsistency: when everything is allowed, explanation collapses into surprise, and predictability becomes meaningless. What Adams presents as comedy is, in fact, a reductio ad absurdum of infinitude. A universe governed by an “infinite improbability drive” cannot support stable laws, cumulative knowledge, or comprehension, because no amount of observation can ever rule out the possibility that reality will simply behave otherwise next time.
Get Yosef Akhtman’s stories in your inbox
Join Medium for free to get updates from this writer.
Now suppose the universe is infinite, whether in spatial extent, duration, or in the number of realisable events. Suppose further that physical laws assign non-zero probability to deviations from their expected outcomes, as standard probability theory does. These assumptions are not exotic; they are built into the foundations of modern physics. Under these conditions, any event with non-zero probability is guaranteed to occur somewhere. This includes not only rare fluctuations, but arbitrarily extreme and persistent ones. There will exist physical regions in which every experimental result deviates dramatically from theoretical prediction, not once, but always. A fair coin will yield only heads for the entire lifetime of an observer. Measurements will never converge. Statistical inference will fail systematically and forever.
For an observer living in such a world, the notion of a law of nature becomes meaningless. A law that never manifests, that cannot be confirmed even in principle, and that is indistinguishable from noise is not a law in any operational sense. It survives only as an ensemble statement about inaccessible worlds. At that point, explanation collapses into bookkeeping over possibilities that no observer can ever verify. Understanding becomes detached from experience.
One might object that such observers are atypical, that they live in rare regions of an infinite universe. But typicality is not something an observer can ever establish. It presupposes access to an ensemble, while physics grants us only one world. In infinite spaces, the very notion of a natural measure is ambiguous, and what counts as typical becomes a matter of convention rather than fact. If comprehension depends on being lucky enough not to inhabit a pathological region, then understanding is no longer a property of reasoning, but of fortune. That is not a foundation for science.
Appealing to potential infinity rather than actual infinity does not resolve the problem. As long as processes are unbounded and arbitrarily large deviations remain possible in principle, no finite amount of evidence can ever secure a law. Whether infinity is completed or merely approached is irrelevant. The epistemic damage is already done.
Contrast this with a relationally finite view of reality. In a finite world, information is bounded, structure is saturated, and deviations compete with constraints. Large violations consume structural capacity. Persistent, systematic departures from regularity are not freely available degrees of freedom. Probability no longer licenses arbitrary worlds; it describes ignorance within limits. In such a setting, laws are not ensemble averages, but structural necessities, and comprehension becomes possible because reality cannot mislead arbitrarily often.
This is not a purely philosophical claim. Modern engineering already relies on it. Techniques such as super-resolution imaging, learning systems, and representation-based inference work only because the world is informationally finite and structurally constrained. If information were truly continuous or unbounded, these methods would be ill-posed in principle. They are not. They work because finitude is real.
Infinity, then, is not merely unnecessary for physical theory. It is incompatible with it. A universe in which anything can happen and everything eventually does cannot support laws, inference, or understanding. The very fact that we can comprehend the world, and build systems that reliably interact with it, is already evidence against infinity, whether actual or potential, as a property of physical reality. Relational finitude is not a belief one adopts. It is the condition under which meaning survives.
