What it was designed to show

In a letter written in 1867, James Clerk Maxwell — the Scottish physicist who had just completed the theoretical unification of electricity, magnetism, and light — described a thought experiment involving a tiny being he called a “finite being” or “valve.” A later popularizer named it Maxwell’s Demon, and the name stuck.

The setup: imagine a container divided into two chambers by a partition. Both chambers contain the same gas at the same temperature. The partition has a small door, and the demon is stationed at the door. As individual molecules approach, the demon observes their speed. If a fast molecule approaches from chamber B, the demon opens the door and lets it through to chamber A. If a slow molecule approaches from chamber A, the demon opens the door and lets it through to chamber B. Fast molecules accumulate in A, slow ones in B. Chamber A gets hotter; chamber B gets colder. Without doing any mechanical work on the gas, the demon has created a temperature differential — and a temperature differential can be used to run a heat engine, which means the demon has apparently created energy from nothing.

This would violate the second law of thermodynamics, one of the most fundamental and experimentally robust principles in physics. The second law says that in an isolated system, entropy — roughly, disorder, or the number of ways the system’s energy could be distributed — does not decrease. A temperature differential is a low-entropy, ordered state; achieving one from an equilibrium state should require work. The demon appears to achieve it for free, using only information about individual molecules’ velocities.

Maxwell’s purpose was not to claim the second law was false but to argue that it was statistical rather than absolute: it holds for large numbers of molecules interacting randomly, but a being with detailed knowledge of individual molecules could, in principle, circumvent it. The law, Maxwell thought, was not a fundamental feature of nature but a consequence of human epistemic limitations.

What it actually shows

Maxwell was right that the second law is statistical. This is now the standard view. But the demon puzzle took almost a century to resolve satisfactorily, and the resolution came from an unexpected direction.

The first attempted resolution, by Marian von Smoluchowski and then more rigorously by Leo Szilard in 1929, focused on measurement. Szilard argued that the demon must measure the velocity of each molecule, and that this measurement must have a thermodynamic cost sufficient to offset the entropy reduction the demon achieves. Szilard’s was the first work to suggest a quantitative link between information (in a physical sense) and thermodynamic entropy.

But this resolution didn’t quite work, because it wasn’t obvious that measurement itself necessarily costs entropy. The physicist Léon Brillouin developed Szilard’s insight further, arguing that the demon needs light to see the molecules, and the thermal radiation at room temperature is too noisy; the demon must therefore use radiation quanta whose absorption increases entropy. This was suggestive but not airtight.

The definitive resolution came from Rolf Landauer in 1961 and was developed by Charles Bennett in the 1970s and 1980s. Landauer’s principle: erasing information is the operation that costs entropy. The demon sorts molecules, accumulates information about molecular velocities in its memory, and can operate without thermodynamic cost — up to the point where its memory is full. At that point, the demon must erase its memory to continue operating. Erasing information is the irreversible operation, and it is the erasure that generates the entropy that offsets the demon’s sorting. The second law is saved not by the act of measurement but by the act of forgetting.

This resolution is both satisfying and philosophically extraordinary. It establishes that information is physical — that what we do with information (specifically, whether we retain or discard it) has measurable thermodynamic consequences. The demon collapses the boundary between knowledge and physics.

How it has been used and misused

Maxwell’s Demon is the thought experiment that founded information thermodynamics, and it has had an enormous and largely legitimate influence on how physicists and information theorists understand the relationship between entropy, information, and computation. The demonstration that erasing a bit of information must dissipate at least kT ln 2 of energy (Landauer’s limit, where k is Boltzmann’s constant and T is temperature) is a fundamental physical result with real implications for computing. As computers are miniaturized and their energy consumption becomes a limiting factor, Landauer’s limit sets a physical bound on how efficient computation can ever be.

The misuse of Maxwell’s Demon tends to run in the opposite direction from most thought experiments: rather than being misapplied to ethics or social questions, it is misrepresented as a proof that the second law can be violated, full stop. This is simply wrong; the point of the resolution is that it cannot. The demon does not violate thermodynamics; it illuminates its foundations.

A more subtle misuse is to read Landauer’s principle as establishing that the mind or consciousness is what saves thermodynamics — that the physical cost of forgetting is somehow tied to conscious experience. This is not what the argument shows; it applies to any physical system that processes information, conscious or not.

What remains genuinely unresolved

Whether Landauer’s principle is correct — not merely theoretically but experimentally — was an open question for decades. It has now been confirmed in several careful experiments, including Landauer erasure in single-particle systems. The principle appears to be real physics, not just a philosophical argument.

What remains genuinely open is the interpretation. The demonstration that information is physical — that there are real thermodynamic costs associated with information processing — raises the question of what information is, at the deepest level. Is it a fundamental category of nature alongside matter and energy? Or is it a useful way of describing constraints on physical systems, without being a basic ontological category in its own right? Some physicists, including John Archibald Wheeler, have suggested that information is more fundamental than matter — that the physical world is in some sense made of information (“It from Bit”). Others regard this as a category error, a confusion between describing physical systems and the physical systems themselves.

The demon also raises persistent questions about the relationship between knowledge and thermodynamics. Maxwell thought the second law revealed human epistemic limitation; Landauer’s resolution seems to say something stronger — that the physical world encodes the costs of knowledge. What it means for physics to place thermodynamic costs on the act of forgetting, in a universe that is itself in the process of forgetting, is a question that connects thermodynamics to the deepest puzzles about time, consciousness, and the nature of the physical world. No one has answered it satisfactorily.