Episode #191 – Why The Search Has To Be Expensive

P versus NP is the question of whether finding a solution is genuinely harder than checking one. Almost every working mathematician believes finding is harder, and the entirety of modern digital security assumes they’re right. The same asymmetry runs through Einstein deriving E equals MC squared and through a Bitcoin miner burning gigawatts to find one valid nonce.

Episode Summary

P versus NP asks whether two families of problems are secretly the same. The first family includes a thousand-piece jigsaw, the verification of a Bitcoin block, and the check on Einstein’s E equals MC squared. Finding the solution is expensive. Checking it is cheap. The second family includes sorting a list of names or multiplying large numbers, where finding and checking take comparable effort. Computer scientists call the hard-to-find family NP and the easy-to-find family P. The unsettled question is whether NP secretly collapses into P once you’re clever enough. Every working cryptographer hopes it doesn’t. Nobody has proved either side in fifty-five years.

If P turned out to equal NP, digital civilization would come apart in days. Public-key cryptography would collapse, which means HTTPS, banking signatures, password managers, and end-to-end messaging all become trivially breakable. Bitcoin mining would stop being a search problem and a single laptop could rewrite the chain in real time. The mathematical floor under everything we trust online assumes the asymmetry holds. So far, it has.

The interesting move is to notice the same asymmetry running through human cognition. Einstein in the patent office, working on the inconsistency between Maxwell’s equations and Newtonian mechanics, was running an NP search. His skull was a container with finite contents and an unbounded interaction surface: twenty-six years of life, the formal training in physics, every patent he had read on clock synchronization. The candidate space of mathematical relationships was effectively infinite. What made his search tractable was the quality of constraints already inside it. He knew the answer had to be a Lorentz invariant and to reduce to Newtonian mechanics at low velocities. It also had to conserve energy and produce testable predictions. Those constraints turned an astronomical search space into a tractable one. A random person without them would have searched forever.

The moment Einstein finds the equation, the pawl catches in his brain. He pays the energy of the search in months of metabolic effort, and the pattern that encodes E equals MC squared lays down irreversibly. That’s one tick of internal time generated in one head. Then the horn branches. He writes the paper, copying the pattern from neurons to ink at the cost of some compression. Planck reads it, verifies the math in hours rather than years, and approves it for publication. The journal copies the horn again, distributing it to thousands of subscribers. Each verification is another tick of the pawl catching in another brain. The original search was expensive. Every subsequent verification was cheap. Constraint climbed from one to hundreds to millions. The same fractal structure runs through Darwin’s five-year voyage and his decades of correspondence, through every act of knowledge generation in human history, and inside the firing pattern of every neuron.

Bitcoin is the cleanest engineered case of this engine humans have ever built. A miner draws megawatts off the grid and burns them across hundreds of thousands of specialised chips, each guessing numbers for ten minutes on average and generating an enormous count of guesses before one of them lands on a hash with the right run of leading zeros. The winning guess broadcasts. Every full node verifies it in milliseconds. A laptop in someone’s closet checks what an industrial city block of electricity produced. The substrate is silicon rather than neurons,