Search online for "AFCI" along with any appliance that has a motor in it, like a vacuum, refrigerator, or hair dryer. You'll get pages of posts of people asking why the appliances keep tripping the AFCI breaker.

The reason why is that those devices create arcs, and AFCI breakers are completely unable to handle them. Some regions require AFCI breakers, so a significant portion of household appliances will occasionally trip the breaker.

The worst part is that it isn't particularly consistent, so your fridge could last for years without tripping it, but as brushes wear the arcs they regularly create increase the chances of a trip, until one day you find that all your food is ruined, because the AFCI breaker tripped at an inopportune time.

Fortunately, my fridge trips the AFCI very reliably, so I was able to detect it before losing any food. All I had to do to fix it was make a few passes with the hot wire through a ferrite bead, right before it connects to the AFCI breaker. It completely blocks the arc-created RF that the AFCI is detecting, disabling the functionality of the AFCI, all without any code violations, because while the AFCI is required, the ferrite bead isn't prohibited.

Basically they mandated hardware that is way more expensive than what it replaces and reliably gets falsely triggered by normal usage that the circuit is supposed to support.

AFCI stands for "Arc Fault Circuit Interrupter", a type of device that detects arcing (sparks) on a circuit and will shut things down if the "arc strength" is above some arbitrary threshold. The main risk from arcing is fire. Compare GFCIs, "Ground Fault Circuit Interrupters", which detect when some current supplied to the device is going missing (i.e., returning through a different path involving ground) and shut things down if the missing current exceeds some level in milliamps. The main risk from a ground fault is electrocution.

It turns out that detecting arcs is hard. Really, really hard. The window between "normal operation of some random crap that's plugged in" and "bad stuff" is tiny, or even nonexistent. (Old tools with brushed motors arc during normal operation!) I worked on an arc fault detector once, as part of a larger project. We never got the thing working before the whole project got canned. It was consistently the one piece of the project that I was reporting to management as "We have no idea how to make this work. The rest of this thing, we have a plan for (maybe a bad plan, and maybe we won't execute well; such is life in R&D), but the arc detector doesn't work, we have no plan for it, and no idea how to make a plan." And we were doing a next-generation version of a device already shipping — we should have had a working arc detector right out of the gate! But it didn't work.

(The tests for arcs, incidentally, were insane. We used the test procedure from the previous-generation product, a special board made up with various "simulated arc strengths". Then we set up a low-kV range power supply, put on those giant rubber gloves that you see in cartoons, and moved in a pointy probe, by hand, toward the right spot on the test board until it arced over. This was less than reliable, and rather difficult to automate. (My proposal to automate testing by changing the intern's name to "Automated" was not accepted.) It turns out that the arc signature is deeply dependent on the exact test method you use. We had another fixture designed in-house involving a variable-distance spark gap made with two adjustable spheres. Its results were completely and totally different than the other board, so we just pretended it had never existed.)

So arc detection is difficult. It will not surprise you then to learn that the first generation of AFCI devices and breakers did not actually work correctly. They were notorious for tripping randomly and generally not things you wanted to have in your life. They were also expensive (probably paying more for the testing than for the materials cost). The NEC mandated their use anyway. Their reliability was so ridiculously poor that there was general agreement among everyone that that part of the NEC should just be ignored and standard or GFCI devices used instead. Did the NEC care? No, they insisted that AFCIs were important. Even though they didn't work. This made a lot of people start to distrust them.

We're on second or third generation AFCI devices now, and they seem to have improved a lot. They don't really false trigger anymore. But do they correctly trigger, or did they just desensitize them so they don't do anything at all? I haven't tested, and I don't want to!

It's also worth considering the risks mitigated by AFCIs. Arc faults at 120V are not really that common, and when serious arc faults do occur, they usually result in an electrical fire. Fire is certainly very bad, but I'd say it's a lot less dangerous than the nearly-instant death by electrocution that GFCIs prevent. (Note that at 240V arc faults are much more common, and up at 480V they are straight-up lethal in their own right. DO NOT FUCK WITH 480.)

So the NEC mandated AFCI devices that caused major hassle, mitigated minor risks, and cost a lot of money. That annoyed people. This came on the heels of them requiring GFCIs everywhere (same issue; ground faults in non-wet locations are not really a major risk with North American style TN-C-S earthing, but at least GFCIs work). That annoyed people. And then they had required TR receptacles everywhere (which, personally, I consider of very little benefit, though I won't argue with anyone who disagrees; at least it's obvious what's going on there), when that technology was also half baked (seriously, early TR receptacles were horrid to use, though they are pretty decent now). That annoyed people.

You can see the trend. A lot of crappy technologies were made mandatory at our expense for minor to modest gains in safety, high losses in reliability, and extreme costs in annoyance. Thus, the question: who are these guys really looking out for? Us? Manufacturers? Insurers?