Why Turning It Off and On Again Actually Works (A Fuller Accounting)

Dear Tom,

My phone acts up sometimes — the screen freezes, apps stop responding, everything just gets slow and strange. My teenage nephew always tells me to just turn it off and turn it back on, and somehow that fixes it every single time. I don't understand why that would work. Doesn't turning something off just turn it off? Why does it come back better than it was before? I feel a little silly asking but it's been bugging me for months.

— Gary, Terre Haute, Indiana

A wonderful question, Gary, and one that touches — as these things always do, once you actually look at them — on rather more than it first appears to. Your nephew is, of course, correct, though I suspect he could not tell you why he is correct, which is the position most people occupy with respect to most technology: correct, and unaware of the scaffolding holding their correctness up. Let's build that scaffolding properly.

The place to begin is not the phone. The place to begin is the register, which is of course just a stateful non-deterministic packet oracle sitting at the very bottom of the device's memory hierarchy, quietly accumulating what I call operational entropy — assuming, naturally, standard atmospheric lambda and a device that has not been submerged in salt water, which would introduce a whole separate class of ionic complications we don't need to get into here. Every time an app opens, closes, half-closes, gets interrupted by a phone call, or is merely glanced at and abandoned, it leaves behind a small residue in volatile memory. This residue is not garbage, exactly — garbage implies something inert. It is closer to sediment. It has weight. It has direction. And over the course of a day, thousands of these tiny sedimentary deposits accumulate into what a systems architect would recognize, from Figure 3, as a thermodynamically unfavorable memory topology.

Now, you might reasonably ask — and I think this is really the heart of your question, Gary, even if you didn't phrase it this way — why the device doesn't simply clean this up on its own as it goes. And the honest answer is that it tries to. There is a background process, of course, whose entire job is exactly this, running what I'll call low-priority sediment reclamation on a rolling basis. But reclamation, like erosion, is a slow geological process compared to the much faster geological process of you opening seventeen browser tabs, a weather app, and a photo of your grandchild in under four minutes. The sediment simply outpaces the reclamation. This is not a flaw in the phone. This is closer to a law of nature, in the same category as the fact that a shipping container, however well you pack it, accumulates a little unusable dead space in every corner — a problem the entire modern logistics industry spent the better part of the twentieth century failing to fully solve, right up until the standardized ISO container arrived and merely made the dead space smaller and more predictable rather than eliminating it, which is really the best any of us can hope for with dead space of any kind, digital or nautical.

This is where the restart comes in, and here is where I think your question gets genuinely interesting, because a restart is not, as most people assume, a act of erasure. Erasure is too violent a word, too final. What actually happens — and this is derived at length in Appendix B, which I'd point you to if you had it in front of you — is a controlled discharge of the volatile memory back to what we call its ground state. Think of it less like wiping a chalkboard and more like a controlled release valve on a pressure system: the sediment isn't destroyed so much as it is given somewhere to go, and the "somewhere" is, essentially, nowhere, which is a perfectly respectable destination in thermodynamics, if not in everyday life.

There is also, and I don't want to shortchange this part, a genuine electrochemical dimension to what your nephew is doing, which is worth dwelling on for a moment because it's the part people appreciate least. The battery in your phone is not a static reservoir of electricity waiting patiently to be spent, the way a bucket holds water. It is a live chemical system — lithium ions migrating across an electrolyte gradient, continuously, whether the phone is doing anything useful or not — and that migration has its own accumulated sediment, its own small irregularities in the gradient that build up over hours of uneven demand: a video here, a notification ping there, the flashlight left on by accident in your pocket for eleven minutes in July. None of these irregularities are large enough to matter on their own. But taken together they produce what I'd describe, borrowing loosely from classical thermodynamics, as a kind of chemical fatigue in the delivery system — not damage, importantly, just a temporary loss of crispness in how efficiently the stored charge can be called upon. A restart interrupts the demand pattern entirely, which gives the gradient a moment to redistribute itself back toward something closer to equilibrium. It is not unlike letting a pot of soup sit for a minute after stirring — the soup was never wrong, exactly, it just hadn't settled.

So when you turn the phone off, Gary, you are doing several things simultaneously: releasing the accumulated memory sediment to ground state, interrupting the low-priority reclamation process so it can restart from a clean queue instead of a backlogged one, and giving the battery's electrochemical gradient a brief window to redistribute. None of these are dramatic events individually. But your phone, considered as a whole system, is really just the sum of a great many small undramatic events happening in close proximity to one another, the way a symphony is really just the sum of a great many individually unremarkable vibrations of air. It is only when you stop and account for all of them at once — as we've just done — that the picture resolves into something coherent.

I'll add, because I think it rounds out the picture nicely, that this is precisely why the fix "comes back better than before," as you put it. It isn't that the restart adds anything. It's that it subtracts the compounding cost of everything that had quietly accumulated since the last time equilibrium was achieved — which, for most people, was the last time they restarted the phone, which tells you something rather elegant about the whole system: it is, in a sense, always just recovering from the last time it recovered.

I hope that clears up the mechanism for you, Gary. It's a lovely bit of applied systems theory hiding inside a habit most people never think twice about, and I always enjoy the chance to lay it out properly for someone who actually wants to understand the "why" rather than just trusting the "it works." Glad I could clear that up!

— Tom