What heat does to machines

The room is on the top floor, over an interior courtyard, with a door onto a flat rooftop terrace. The desk sits under the window because that’s the only spot with any light, and the window faces in such a way that from one in the afternoon, the sun lands on the opposite wall, then on the sill, then on the desk.

The wall behind the desk is hot to the touch. Not warm. Hot, like a stone left out in the sun, which is exactly what it is.

The air is dry and it doesn’t move. Opening the window changes nothing except the noise level. I’m describing a room, a month of March, and an exposure, not a country: two streets over, a ground-floor room facing the same courtyard stays cool all day, and I’ll only find that out too late.

What the room did to the gear

On the first day, I didn’t notice anything until three in the afternoon.

At three, a file export I run two or three times a week, the kind that takes about as long as fetching a coffee, was still running when I got back with the coffee. It finished after a quarter of an hour. Everything else followed: windows taking a full second to open instead of just opening, the cursor skipping, text appearing half a second after I typed it.

No message. No error. No alert.

And above all, no noise, which is the important part, and which took me two days to understand. My machine has no fan. It’s a deliberate design choice by the manufacturer, even a selling point, and I’d bought it in February thinking it was elegant.

A machine with a fan warns you. It whirs, it spins up, you hear it working and you hear it struggling. A machine without a fan has no channel to tell you anything at all. It slows down. That’s its only language.

The phone, which doesn’t say anything either

Around four, I plugged in the phone, a Galaxy S22 I’ve had for four years, which has survived everything, including a theft on a train where the thief didn’t find it because it was in my pocket.

An hour later, it showed the exact same percentage as when I’d plugged it in.

I started with the cable, which I swapped. Then the charger, which I also swapped. Then I blamed the wall outlet, in good faith, because that particular outlet does have a real mechanical problem, and I’d just spent a week writing an article about it.

The outlet had nothing to do with it this time. The phone had paused charging and said so, right on screen, in a notification I’d swiped away without reading, because I’ve been swiping away everything that appears at the top of the screen for four years.

It was sitting on the desk, the desk had been in the sun for three hours, and it refused to keep heating up further just to please me.

What this actually is, and why it looks like a failure

A machine that gets hot cuts its own power draw. Less power, less heat produced, and the electronics stay within the range they’re supposed to operate in. A phone that’s charging does the same thing a different way: charging produces heat, so it slows the charge, or stops it and waits.

These are protective behaviors. They’re built in on purpose, documented by manufacturers on their support pages, and they kick in without breaking anything.

I won’t give you a temperature threshold. I don’t have an instrument, I didn’t measure anything, and the numbers floating around on forums aren’t sourced anywhere. The only figure worth anything would be the one the manufacturer publishes for this exact model, and I’d rather write nothing than write a number I can’t trace to an official page.

What I can describe is the mismatch. Protective behavior looks, in every way, like a defect. Unexplained slowness, charging that doesn’t move, an operation that doesn’t complete. Faced with those three symptoms, a user concludes it’s broken within ten seconds, and the natural reaction is to push harder.

I restarted the export three times. Each time, I was asking the machine to give its maximum right when it had already decided to give less. It held that position for an extra hour, longer than it would have if I’d just gone for a walk. My persistence extended exactly what I was trying to stop.

What I did instead

Nothing clever, and nothing I’m going to sell you as a method.

I shifted my hours. Up at six, working by quarter past, in a room that had spent the night cooling down. Between six and eleven, the machine behaves like it would in Lisbon in February: the export takes two minutes, the cursor doesn’t skip, the phone charges in an hour.

The afternoon stopped being work time. It became time for walking, for the paper notebook, for meetings, for things that don’t need electricity. And in the evening, once the room has cooled back down, I pick up an hour or two more if I feel like it.

I’m not going to tell you to prop your machine on a slotted stand, raise it up on two books, aim a fan at it, or put it in the fridge inside a bag, which I’ve seen recommended in all seriousness. I haven’t tried any of that, I don’t know what it does, and improvising on the thermals of a device I don’t know how to open up isn’t a field where I’m going to start handing out opinions. This site refuses that kind of advice, and that’s stated where it belongs.

The only lever I pulled was the one I actually had control over: the hour I work, and the room I sit in.

On the second evening, I opened the manufacturer’s support page and started typing into the chat window. “Hello, my machine, bought in February, has become very slow for no apparent reason, I think it has a defect.” I reread the sentence. The word that bothered me was “no apparent reason,” because the reason was apparent, it was even written on the wall, thirty centimetres from the screen, and I could touch it with my hand. I closed the window without sending it. The next morning, at half past six, the export took two minutes and ten seconds, and I checked that the file matched, byte for byte, the one the machine had eventually produced the night before after a quarter of an hour. It did, exactly. It hadn’t done anything worse. It had just taken the time it needed not to damage itself, and I’d spent two days holding that against it.