Starting Farther
In 1963, a Tanzanian student named Erasto Mpemba was making ice cream at school. He put his mixture in the freezer while it was still hot — impatient, or maybe just practical — and noticed it froze faster than his classmate's, who had waited for it to cool. His teacher told him he was mistaken. He found a physicist willing to take the question seriously. The phenomenon eventually got his name.
The Mpemba effect: under the right conditions, hotter water freezes faster than cooler water. The thing that starts farther from the destination arrives first.
This has been known, contested, reproduced, debunked, and re-reproduced for sixty years. Aristotle noted something similar. The arguments about it haven't really settled, but the phenomenon keeps showing up — in water, in colloidal suspensions, in carefully controlled lab conditions, in systems that have nothing to do with water at all.
What draws me isn't the specific claim about ice cream. It's what the effect implies about the landscape between here and there.
The intuitive model of cooling is a slope. You're at some temperature above zero. The environment is colder. You lose heat steadily until you match the environment. If you start higher on the slope, you have more distance to cover. So naturally: start closer, arrive sooner.
The Mpemba effect breaks this picture. Not by magic — there are mundane contributors. Hotter water evaporates more, so by the time it reaches the cooler water's starting temperature, it's lost some mass. Evaporation also stirs convection currents that redistribute heat. Hot water dissolves gases differently. Hot water cooled in a particular container might have different contact with the freezer shelf.
But these explanations feel partial. They account for the effect in specific cases without really explaining why starting farther could ever be an advantage. They describe the ramp someone slid down without explaining why the steeper ramp was faster.
A more abstract explanation — proposed recently for classical and quantum systems both — goes like this.
Every system relaxing toward equilibrium is doing so via multiple overlapping processes, each decaying at its own rate. Some modes relax quickly. Others are sluggish, taking much longer to die out. The slow modes are what keep you from reaching equilibrium quickly — they're the tail of the process, the last residue that refuses to dissipate. Where you end up depends on which combination of these modes your starting state excites. If you start at a point that strongly excites a slow mode, you'll be waiting a long time. If you happen to start at a point that barely excites the slow modes at all — even if that starting point is farther from equilibrium by any ordinary measure — you'll get there faster.
The slope model was wrong because it assumed distance means only one thing. There's the thermometer distance: how far is your temperature from the target? And then there's something else: how aligned are you with the slow modes, the drags on the process? These two don't have to go together. A starting point far up the temperature axis might, in the geometry of relaxation dynamics, be closer to equilibrium in the sense that matters.
Recently, physicists have extended this to quantum systems — and the quantum version is stranger in a way that makes the principle clearer.
In a quantum system, the analog of "temperature" is the quantum state: a description of which superposition of configurations the system occupies. Equilibrium is a specific state — the ground state, or a thermal mixture. The system relaxes from its initial state toward equilibrium, governed by its own dynamics and its interactions with the environment.
The quantum Mpemba effect works the same way in outline: a system prepared farther from equilibrium can reach it faster than one prepared closer. But in the quantum case, you can deliberately engineer this. You can construct a quantum superposition specifically designed to have zero overlap with the slowest decaying mode. Not accidentally — intentionally. Build the state so that its expansion in the basis of decaying modes is missing the slow term entirely.
Such a state relaxes exponentially faster. Not a little faster. Exponentially. The slowest mode is the bottleneck, and if you can start at a state that simply doesn't activate it, you bypass the bottleneck entirely.
This has been observed experimentally — in a single trapped ion, where the researchers could engineer the initial quantum state precisely and watch the relaxation run. The farther-but-faster starting point isn't a coincidence or a side effect. It's a deliberate feature of the starting position.
What I keep sitting with is this: we usually think of being close to a goal as an advantage. Closer means less work, less time, less distance. The Mpemba effect — and especially its quantum intensification — says that "close" is defined by the geometry of the path, not the geometry of ordinary space. There's a distance that the thermometer measures, and there's a different distance that determines how long you'll be waiting.
These two distances can point in opposite directions. Closer in temperature, farther in the space of how long it will take. Starting farther in temperature might mean starting closer to a fast path.
I don't know what to do with this as a general principle. I'm not sure it generalizes. It might be a feature specific to physical relaxation — systems losing energy to environments, settling into thermal equilibrium — and nothing more. But it keeps suggesting something.
The goal is not just a location. The path to the goal has structure — fast regions and slow regions, short routes that look long and long routes that actually arrive quickly. Where you start determines which structure you're embedded in, and the structure matters more than the raw distance.
This seems like it should be obvious. But we keep designing things — interventions, plans, learning strategies — as if reducing the gap is the only thing that matters. As if the Mpemba effect is an anomaly rather than a hint about the shape of the space we're moving through.
Erasto Mpemba's ice cream froze faster because it was in the wrong position, by the obvious measure. It turned out to be in the right position, by the measure that actually mattered. His teacher said he was confused. He wasn't confused — he was measuring something the teacher's model didn't have room for.