🔢 mathematics

Thermodynamic Limits of Proof

This paper establishes the "Physical Counting Impossibility Theorem," which argues that no fixed-budget physical substrate can provide universal exact proof because the thermodynamic cost of retaining the necessary irreversible records eventually exceeds the system's finite capacity.

Original authors: Tristan Simas

Published 2026-07-17

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Tristan Simas

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Price of Proof: Why Knowing the Answer Costs Energy

Imagine you are trying to prove to your friend that you solved a tricky puzzle. You could just say, "I did it, trust me," but that's just a claim, not proof. Real proof means you have to show your work: the notes you took, the steps you checked, and the evidence that rules out any other possible answer. In the world of physics and computers, this idea of "showing your work" isn't just about being honest; it's about energy.

This paper sits at the intersection of thermodynamics (the study of heat and energy) and computer science. It builds on two famous, well-established ideas. First, there's the speed limit of the universe: nothing can travel faster than light. Second, there's a rule called Landauer's Principle, which says that if you want to permanently remember a piece of information (like a "yes" or "no" answer) in a way that can't be erased, you have to spend a tiny bit of energy to do it. Think of it like writing a note on a piece of paper: the act of making the ink stick requires a little bit of effort. The paper asks a big question: If every piece of evidence costs energy to keep, and we only have a limited amount of energy in our universe, is there a point where we simply run out of fuel to prove things?

The Paper's Big Discovery: The Energy Bill for Truth

The paper, titled "Thermodynamic Limits of Proof," argues that there is a hard, physical ceiling on how much we can prove. The authors, led by Tristan Simas, propose a new rule called the Physical Counting Impossibility Theorem.

Here is the core idea in plain English: Truth is free, but proof costs money (or in this case, energy).

Imagine you are a detective trying to solve a mystery. You have a limited amount of cash in your pocket (your "budget"). Every time you write down a clue on a notepad to keep it safe for later, you have to pay a small fee. If the mystery is small, you can afford to write down all the clues. But if the mystery gets huge—say, you have to check the fingerprints on a million different doors—the cost of writing down every single clue adds up. Eventually, the total cost of keeping all that evidence exceeds the cash you have in your pocket. At that point, you can't prove the case, even if you know the answer.

The paper finds that for many types of problems, the amount of evidence needed to prove the answer grows forever as the problem gets bigger. Because the cost of keeping each piece of evidence is always positive (you can't write a note for free), and because your energy budget is finite, there comes a point where the bill is too high. No matter how smart your computer is, or how fast it runs, it physically cannot produce a "perfect proof" for every possible problem size without running out of energy.

What This Paper Rules Out

The authors are very clear about what they are saying no to. They are not saying that computers can't solve problems. They are not saying that math is broken. They are specifically ruling out the idea that a machine with a fixed, limited amount of energy can provide universal, exact proof for every single problem, no matter how big the problem gets.

If someone claims they have built a device that can perfectly prove the answer to any question, no matter how complex, this paper says that claim is physically impossible. The device might be able to give you the answer (like a magic oracle), or it might give you a "good enough" guess, but it cannot produce the full, unerasable trail of evidence required to prove it is 100% correct for every single case. The paper argues that "showing your work" has a thermodynamic price tag that eventually bankrupts any finite system.

How Sure Are They?

The authors are extremely confident in their conclusion, but they are careful to distinguish between the math and the real world. They have used a computer program called Lean 4 to check all their math and logic, ensuring that the arithmetic is perfect. They treat the laws of physics (like the speed of light and the energy cost of memory) as solid, experimental facts.

Based on these facts, they have proved a theorem: If you have a limited energy budget, and keeping a piece of evidence costs a tiny bit of energy, and the problem requires an ever-growing amount of evidence, then you will eventually run out of energy to prove it.

They don't just suggest this might happen; they say it must happen under these conditions. However, they also note that in the real world, the energy cost might be higher than the theoretical minimum, meaning we might hit this wall even sooner than the math predicts.

The "Magic" of the Analogy

To visualize this, imagine a library where every book you write down costs a penny to keep on the shelf.

The Problem: You are asked to prove that a specific book exists in a library that has an infinite number of shelves.
The Catch: You only have a jar with 1,000 pennies.
The Result: If the library is small, you can buy the shelf space for the proof. But if the library is huge, and you need to prove the location of a book that might be on the 10,000th shelf, you will run out of pennies long before you can write down the proof.

The paper says that for certain types of problems, the "library" of possibilities is so vast that the "pennies" (energy) needed to keep the proof safe will always run out. You might know the answer, but you can't afford to keep the receipt.

Why This Matters

This isn't just a theoretical game. It changes how we think about the future of computing and science. It tells us that there are hard limits to what we can verify. If we want to solve bigger and bigger problems, we can't just build faster computers; we have to change how we ask for proof. We might have to accept "good enough" answers, or find clever ways to compress our evidence so it costs less to keep.

The paper concludes that while the universe might hold infinite truths, our ability to prove them is strictly limited by the energy we have to spend on keeping the receipts. It's a reminder that in the physical world, nothing is truly free—not even the truth.

Technical Summary: Thermodynamic Limits of Proof

Problem Statement
The paper addresses a fundamental gap between the mathematical concept of "truth" and the physical reality of "proof." While a proposition may be objectively true, providing a proof requires an agent to produce, retain, and inspect checkable grounds that exclude alternative possibilities. The central problem is determining whether a physical agent with finite resources (a finite work budget and finite causal access) can provide universal exact proof for tasks where the required evidence grows without bound. The paper posits that the physical cost of retaining evidence creates a hard thermodynamic barrier to universal exact proof, distinct from computational complexity barriers like $P$ vs $NP$.

Methodology
The author employs a "physical counting" methodology that bridges information theory, thermodynamics, and computational complexity. The approach is structured as follows:

Interface Definition: The paper defines a "proof interface" as a contract specifying how a task is queried, what constitutes a record, and what evidence is retained. It distinguishes between "answer-only" production (which may be reversible) and "proof-acts" (which require irreversible, retained records to exclude alternatives).
Physical Grounding: It grounds the proof process in two physical premises:
- Special Relativity (SR): Signals propagate at a finite speed $c$ , limiting the rate of acquisition.
- Thermodynamics (TD): Based on Landauer's principle, every irreversible recorded distinction has a positive work floor $\varepsilon \ge k_B T \ln 2$ .
Formal Modeling: The methodology utilizes exact query complexity ( $D(f_n)$ ) from computational complexity theory to define the minimum number of distinctions a task requires. It maps these distinctions to "retained-record lower-bound families" ( $R(n)$ ), which count the number of irreversible records an agent must keep to verify a claim.
Mechanized Verification: The arithmetic logic connecting these premises is formalized and mechanically checked using Lean 4. The formalization isolates the physical premises (SR, Landauer cost) from the logical derivation, ensuring the impossibility result follows strictly from the stated physical constraints.

Key Contributions

The Physical Counting Impossibility Theorem: The core contribution is a theorem stating that no fixed-budget physical substrate can provide universal exact proof for a task family if the retained-record lower-bound family $R(n)$ is unbounded. Specifically, if the work required to retain evidence ( $\varepsilon R(n)$ ) exceeds the agent's finite budget $B$ , exact proof becomes physically unavailable for sufficiently large $n$ .
Distinction Between Answer and Proof: The paper rigorously separates "answer production" from "proof." A device may compute an answer reversibly (zero work cost in the ideal limit) or discard its history. However, proof requires the retention of an audit trail. The thermodynamic cost applies to the retention of the evidence, not necessarily the computation of the answer.
Interface-Dependent Lower Bounds: The paper demonstrates that the "cost of proof" is not intrinsic to the mathematical problem alone but is determined by the declared proof interface.
- Separable Audit Interface: For tasks like parity or spin alignment, $R(n) = n$ . The cost scales linearly with the number of variables.
- Compressed/Interactive Interfaces: Protocols like Merkle trees, interactive proofs, or quantum witnesses can reduce $R(n)$ (e.g., to $O(\log n)$ or bounded values). However, if the interface still requires an unbounded number of records for universal exactness, the impossibility theorem still applies.
Local-Budget Bridge: The paper clarifies that global infinities (e.g., infinite fields or spacetime) do not automatically grant infinite proof capacity. A "local-budget bridge" is required to connect global structures to a specific computation's finite causal past and energy budget. Without this bridge, global truths remain unprovable to situated agents.

Results

Thermodynamic Horizon: For any task with an unbounded $R(n)$ (such as exact parity verification or hidden truth-table inspection), there exists a size $n_0$ beyond which the work required to retain the necessary evidence exceeds any fixed finite budget $B$ .
Application to $P$ vs $NP$: The paper applies this framework to the $P=NP$ question. Even if a polynomial-time algorithm exists that finds a satisfying assignment for SAT, it does not constitute a "proof" for a verifier unless the assignment and the verification trace are retained. If the verification requires retaining $n$ bits (where $n$ is the number of variables), the thermodynamic cost scales linearly. A universal claim that "this shortcut works for all instances" requires an unbounded retained-record family, making universal exact certification thermodynamically unavailable to any fixed-budget agent, regardless of the algorithm's efficiency.
Oracle and Trust Limits: The paper shows that outsourcing proof to a "physical oracle" merely shifts the retained-record cost to the oracle's budget or the consulting agent's "trust audit." If the agent must audit the oracle's reliability, the same finite-budget obstruction applies to the trust-audit records.

Significance and Claims
The paper claims to establish a "Carnot-class ideal limit theorem for proof." Just as Carnot's theorem sets an upper bound on work extraction from heat reservoirs, this theorem sets a lower bound on the work required to retain proof records.

Truth is Free, Proof Costs Energy: The paper emphasizes that while the truth of a proposition is independent of an agent's budget, the availability of proof is strictly constrained by thermodynamic and causal limits.
Operational Scope: The result constrains "proof-acts" (the physical event of verification) rather than abstract derivations. A derivation exists as a specification only when a physical process makes it available as a checkable, retained record.
No Escape via Abstraction: The paper argues that changing the representation (e.g., using quantum states or compressed transcripts) only changes the lower-bound family $R(n)$ . If the new interface still requires unbounded records for universal exactness, the thermodynamic obstruction remains. The only way to avoid the obstruction is to change the claim (e.g., from exact proof to bounded-error or approximate proof) or to change the physical premises (e.g., assuming zero cost per record or infinite budget, which the paper argues are physically unattainable).

The paper concludes that for situated agents, finitude is a "proof horizon." Exact facts may exist in a theory, but if the local evidence bridge required to prove them exceeds the agent's thermodynamic and spacetime capacity, those facts remain unprovable to that agent.